Trends in Business Intelligence for Marketers

Trends in Business Intelligence for Marketers

David M. Raab

DM Review

August 2007

 

            The general trends in the business intelligence marketplace are well known: deployment to front-line operations, replacement of descriptive with predictive analysis, extension to unstructured data sources, greater use of hosted services, and more advanced visualization.  All of these certainly apply to marketing departments.  But they’ve been exhaustively discussed elsewhere, so there’s little point in reviewing them here.

            Are there any business intelligence trends that are more unique to marketing?  Let me propose two that are definitely happening, and another that’s a bit more speculative.

            One clear development is the consolidation of Web and non-Web analytics.  An observer from Mars might ask why these were ever separated in the first place, but it wasn’t done on purpose: the newness of the Web and the technical difficulties of capturing Web behavior ensured that initial Web analytics systems would be stand-alone products.  Only now that a generally accepted approach has evolved for acquiring Web interaction data—essentially, build tracking codes into Web pages—it is possible to shift the focus from data capture to true analysis.  (Not that the data capture problem is fully solved: page tags are inadequate for many types of Web data, including user-generated content, rich media content, and some dynamic Web pages.  So we can still expect more innovation in that area.)

            In merging Web and non-Web analytics, the biggest challenge is customer data integration: linking Web identities such as cookies, IP addresses and email addresses with offline identities such as names, addresses and account numbers.  Precious little can safely be inferred from conventional matching techniques, since there is often no connection between the text of, say, an email address and a postal address.  More likely, marketers will rely on registration and similar direct data sources to link online identities with their real world counterparts.

            Once customer identities have been linked, marketing intelligence systems will need methods to handle much richer sets of interactions across time and multiple channels.  The simple one stimulus:one response model that seeks to attribute a particular customer behavior to a particular marketing contact has become hopelessly outmoded in a world where many contacts are recorded before, during and after each business transaction.  What’s new is not that there are many contacts, but that they’re actually recorded and therefore available for measurement and optimization.

            Business intelligence vendors are just beginning to offer analytical tools that can show the relationships among multiple, cross-channel contacts and business results.  This is where techniques such as visualization, time-series analysis and simulation come to the fore.  It also requires deeper integration with financial information to make sure the analysis can focus on meaningful business results, not just whichever behaviors are easily observed.

            The second clear trend is greater interest in marketing performance measurement.   Of course, there’s nothing new about marketing measurement, or use of business intelligence for performance measurement in general.  But the topic is attracting much more attention than it used to: there are more conferences, white papers and surveys than ever before.  One explanation is the rapid expansion of the digital channels (Web, email, mobile) has marketers looking for new ways to allocate resources: since the old rules of thumb didn’t include the new channels, marketers need different approaches to determine the appropriate mix.  And while many marketers would probably be content to make the allocations based on their personal instincts, chief executives and finance departments are pressing for less subjective approaches.

            One practical consequence of reporting across many efforts in many channels will be a greater need to aggregate details from hundreds or thousands of marketing programs into meaningful summaries.  This implies technologies such as balanced scorecards and hierarchical reporting structures.  These will have to show major trends, performance against plan, and significant variances, while still allowing users to drill down into details on demand.  The techniques that business intelligence vendors have developed to serve areas outside of marketing should generally prove sufficient for these applications.

            But the complex interrelationships among multi-channel contacts will also push business intelligence vendors to deploy marketing measurement systems that go beyond reporting of past results to predictions of future behavior.   These predictions are not simple statistical forecasts or probability models, but multi-factor simulations that estimate how each customer contact affects later activities, and also shows the impact on those activities of external factors such as competitor behavior and business conditions.  Only when extraneous activities are factored out—using a reliable, consistent method that is credible to managers outside of marketing—can performance measurement systems really report on the impact of specific marketing decisions.

            In other words, both the increasing variety of marketing channels and the increasing pressure for marketing performance measurement are pushing business intelligence vendors towards more powerful analytics in general and customer behavior simulations in particular.  Simulation, in turn, is closely connected with business process management, so it may push business intelligence vendors to extend their products in that direction.

            So far there’s been little concrete evidence of this movement so I still consider it speculation rather than a confirmed trend.  But it’s hard to imagine marketers finding any tool other than simulation that can organize their information to be comprehensive, comprehensible, and actionable.  Of course, they always have the option to do nothing, but this seems unlikely: the pitfalls of using disconnected, short term measures are too obvious to ignore.  So we can expect sophisticated business simulations to play an increasingly central role in business intelligence for marketers.

 

*                     *                      *

 

David M. Raab is a Principal at Raab Associates Inc., a consultancy specializing in marketing technology and analytics.  He can be reached at draab@raabassociates.com.

 

 

 

 

Technical Measures for Data Quality Investments

Technical Measures for Data Quality Investments

David M. Raab

DM Review

June 2008

 

Last month’s column presented several return on investment calculations for data quality projects.  These were the measures that business people look at: profit per customer, promotion effectiveness, value per response, return on promotion expense.  Let’s look at technical measures of data quality for those same cases. 

 

- profit per customer.  An automobile dealer made service history available to salespeople while a new car purchase was being negotiated.  The business value came from targeted offers to increase use of the highly profitable service department. 

 

Technical data quality measures included:

 

            - speed of access: this is the time it takes the salesperson to retrieve data for a customer.  Multiple queries may be needed before the system returns a satisfactory result, and salespeople will not bother if it takes too much effort.  Elapsed time would be gathered from system logs. 

   

            - match rate: this is the proportion of successful matches returned by the system.  There are separate statistics for correct matches, false matches, and missed matches.  Match accuracy is often difficult to measure because the correct answer is not known.  But in this case, the customer will know whether she has previously used the service department.  The salesperson should therefore know when to look and keep trying until the system returns a match.  This means the most important measure is “correct results returned on the first try,” as shown by the number of successful single-search sessions.  Successful searches are followed by a request to view the underlying data.  Abandoned searches are not.

 

            - service data quality: this includes all quality components—accuracy, completeness, consistency, currency, and suitability to task.  Since the service history is derived from the service department’s billing system, it should be reasonably accurate, current and complete.  This would be confirmed by the company’s normal auditing functions.  Consistency is measured by profiling the data over time to identify unexpected values or value distributions.  Profiling can also detect improper or fraudulent billing—something the service manager may or may not be particularly eager to explore. 

 

            Suitability to task is a particular challenge, since the data is being used for something other than its original purpose.  The system must summarize the raw service data to show aggregate purchases, changes in usage patterns, types of work (e.g. all routine maintenance or only major repairs), and inferences about customer needs (high mileage, off-road travel, heavy loads, etc.).  Summarization depends on the core data quality measures of accuracy, completeness and consistency. 

 

            Even summarized data can be difficult for a salesperson to interpret, so the system should also recommend a best offer. Recommendation quality is measured by tracking how many recommendations are presented by the salespeople, how many of these are accepted by the customers, and their long-term impact on customer profitability.  Presentations and acceptances can be measured directly so long as salespeople record their results.  Long-term impact requires tracking customers over time.

 

Similar technical data quality measures apply to the other three cases.  Space is limited, so, briefly:

 

- promotion effectiveness.  This was a project to improve accuracy of a packaged goods manufacturer’s lists of distributor contacts.  The business value was better execution of retail promotions.  Technical data quality measures include:

 

            - list accuracy: determined by random telephone calls to the distributors to verify the names on the existing lists.   Returned mail and rejected email addresses may also provide information.

 

            - update speed: determined by tracking how often the sales force provides list updates.  This will identify salespeople who are not participating. 

 

- value per response.  This described an online marketer’s project to reduce bad debt and improve product recommendations through better real-time access to customer history.  Technical data quality measures include:

 

            - match rates with internal systems: measures include the percentage of successful matches, the percentage of confident matches (using a system-generated confidence score), and the percentage of multiple matches (more than one customer record matches a single input).  Here, independent validation of match accuracy may not be available.

 

            - match rates from external sources: confidence scores may not available, so the only measure is the match rate itself.  Some verification is needed to measure false matches—a particular issue with external vendors who are paid on the number of hits.

 

            - quality of results from internal systems.  Completeness is measured by the scope of data provided: purchases, payments, returns, refunds, and service interactions.  These may originate in several different systems.  Currency is measured by how long it takes a new transaction to become available.  It can range from milliseconds to a month.

 

            Important non-data quality measures include response time and prediction accuracy. 

 

- return on promotion.  This described direct response marketer who used lifetime value to optimize promotion spending.   Technical data quality measures included: 

 

            - cost data: accuracy, completeness, and consistency.  The most important measure is percentage of missing values, since many marketers fail to record the necessary information in the marketing system.  Another key measure is variation between the marketing system and the accounting system, since entries in the market system may not be revised to reflect actuals.

  

   - customer integration: accuracy and completeness.  Records for the same customer are set up independently in several systems and then merged.  Measures of incomplete merges include refunds without a corresponding purchase, and repurchases without an initial order. 

 

*                     *                      *

 

David M. Raab is a Principal at Raab Associates Inc., a consultancy specializing in marketing technology and analytics.  He can be reached at draab@raabassociates.com.

 

 

Calculating the Return on Data Quality Investments

Calculating the Return on Data Quality Investments

David M. Raab

DM Review

May 2008

 

Everyone agrees that data quality is important, but that doesn’t make them willing to pay for it.  Any manager asked to approve a significant data quality project will rightly want to understand its return on investment.

 

Sometimes the justification is a simple cost reduction: fewer duplicate mailings, fewer misdirected shipments, fewer customer service transactions to correct mistakes.  But often the benefit comes from higher revenue: better data will allow more effective promotions or more precise (that is, higher) pricing.  Data management professionals are frequently unfamiliar with the details of such analyses.  Even the business counterparts who are supposed to provide the necessary input may not know how to structure them to satisfy financial gatekeepers.

 

The challenge usually lies with the value calculation: the “R” in ROI.  After all, investment is not much different from any other systems project.  But how do you estimate the value of incremental revenue—or, indeed, what that revenue might be?

 

Here are several real-world projects that involved benefits from improved data quality.  Although they cover just a handful of the possible situations, they may inspire insights that apply to your own business.

 

- higher profit per customer.  An automobile dealer wanted to increase revenues from its service operation, the primary source of business profits.  A proposed integration project would make the service history of existing customers available to salespeople while a new car purchase was being negotiated.  The value came from helping salespeople to make the most appropriate service-related offer for each purchaser.  Existing service department users would be offered a long-term contract to lock in their business, while sporadic users would be given discount coupons to encourage them to come back.  In this situation, returns were measured in terms of increased profit per customer.  This incorporated a company-wide view that included profit on the sale itself, profit on future service revenues, and profit from financing activities.

 

- improved promotion effectiveness.  A consumer goods manufacturer relied heavily on retail promotions executed through its distributors.  The manufacturer was aware that it often paid for promotions that never reached the store aisles.  A little detective work found that materials for these promotions were sometimes not delivered to the distributor or, more often, they were not placed in the store by the distributor’s field staff.  Digging further, it turned out that the company’s contact lists for distributor field staff were often outdated.  As a result, distributors often did not receive notice of planned promotions or know who to contact when materials were received unexpectedly or expected materials were missing.  The value of the project to fix these lists was based on a reduction in wasted promotion materials—which accounted for about half the total marketing budget—and the revenue gain from increasing the number of promotions that were actually executed.

 

- increased value per response.  An online marketing organization used email and Web advertising to generate orders.   For each response, the company had to decide whether to require payment in advance of shipment.  This was a delicate balancing act, since pre-payment reduced the number of orders, but credit often resulted in bad debt.  In theory, the company could identify likely no-pay customers based on previous behavior, but poor matching and disconnected fulfillment systems meant only a portion of this history was available while the order was being processed.  The project to improve matching and data access was justified by the value of better credit decisions: orders would increase because more good customers were given credit, while bad debt would drop because more bad customers had to pay in advance.  Access to previous purchase history also would allow the company to better recommend additional products to buy once the initial order had been accepted.  The profits from higher add-on sales per responder might actually exceed benefits of the improved credit decisions.

 

- optimal return on promotion expenses.  A direct response marketer acquired customers at a loss in order to make profits on future sales.  It had a wide of choice of products to offer for the initial promotion and a wide range of channels to reach them.  Each product group traditionally evaluated acquisition promotions based on cost per order and expected revenues for future sales within its own group.  Obstacles to a company-wide measurement of each customer included multiple account numbers for the same customer, and lack of accurate cost data for lifetime value calculations.  The value from removing these obstacles would be measured by the increase in company-wide profit from reallocating promotion expenses to the most profitable acquisition product and channel combinations.   This required calculating the lifetime value of customers acquired by increasing or decreasing promotion spending for each combination—a demanding but doable task.  Moving investment from the least to the most profitable options would result in a substantial long-term profit increase with no change in promotion expense.

 

In one sense, each of these projects is justified in the same way: by higher company profits.  But identifying the specific mechanisms that will generate these profits yields credible, understandable return on investment calculations.  These are much more likely to result in project funding than a generic appeal to the value of data quality.  Although details for your projects will be different, a similar approach should also serve them well.

 

*                     *                      *

 

David M. Raab is a Principal at Raab Associates Inc., a consultancy specializing in marketing technology and analytics.  He can be reached at draab@raabassociates.com.

 

 

 

 

 

 

New Technologies for Inbound Marketing

New Technologies for Inbound Marketing

by David M. Raab

Curtis Marketwise FIRST

April 2008

 

Bank marketers are increasingly recognizing the opportunities of customer-initiated contacts such as telephone calls and Web site visits.   Outbound messages risk being ignored, ill-targeted or intrusive, while inbound messages start with the customer’s attention, can be tailored directly to the situation, and are clearly triggered by the customer’s own actions.  As a result, they are much more likely than an outbound message to yield a productive result.

 

Given these advantages, why isn’t inbound marketing more common?  The problem is simple: like tongue-tied adolescents as a school dance, banks don’t know what to say when an opportunity presents itself.  So they stare down at their metaphorical feet, listen silently to the hold music, and eventually wander off without having tried to make a connection.

 

But help is on the way.  New technologies can teach banks to understand what customers want and how to offer it to them.  They can even deliver the right message at the exact moment it is needed.  Think of them as can’t-miss pickup lines for financial institutions.

 

The first challenge in successful inbound marketing is listening.  Having a human  involved helps—but call center and branch agents focus on solving the customer’s immediate problem, not assessing the situation for marketing opportunities.  Nor do most agents have the skills, training or personality to do a good job of marketing.  So whether an agent is involved or it’s a fully automated interaction on a Web site or ATM machine, technology should handle most of the marketing-related listening.

 

This listening has at least three components.  One is literally understanding the customer’s words.  The process might begin with spoken words that are converted to text through speech analysis software, or it may originate as text in an email message, Web query or agent call notes.  Either way, text analysis software will then parse the message to identify key attributes such as products mentioned, terms demanding attention (e.g., “attorney”), and emotional content.  Current technology can extract these sorts of items, but it hasn’t reached the stage where it can reliably understand the exact meaning of how they are being used.  That is, the software might recognize that a conversation involves free checking accounts, but not assess whether the customer already has an account or is considering opening a new one, let alone precisely which features would be most important.

 

This brings up the second component of listening: tracking specific activities in company systems.  Technology can monitor the events during an interaction—accounts opened or closed; deposit, transfer, and withdrawal transactions; balance inquiries; data from forms; search terms entered; Web pages viewed; and so on.  These are much less ambiguous than streams of text.

 

Current transactions can be further enriched by the third component of listening, which is placing the current interaction in context.  This brings in customer data such as existing accounts and balances, past transactions, service history, and background information in company systems or from external sources like a credit bureau.  It can also include non-customer data such as the current workload in the call center, current promotions, and profitability of specific products.

 

Taken together, these three forms of “listening” provide a rich view of a current interaction: a much richer view, in fact, than a human agent could assemble on her own.  It can be hard work to assemble all this data, and the initial implementation of an inbound marketing system is unlikely to be complete.  But even partial data can be adequate input to the next task: making sense of what’s happening and deciding what to do about it.

 

This step usually involves a combination of business rules and statistical models.  The models predict specific behaviors, such as probability of accepting a particular product offer or of closing an account.  The business rules make decisions, or recommendations if a human is involved: offer this product, waive that fee, present these selling points.  The rules themselves often incorporate model scores, which helps keep the rules simple: the rule might indicate it’s time to make a product offer, but let the model select the specific product based on likelihood of acceptance, profitability, expected impact on retention, and other factors.  Rules and models can also provide non-marketing guidance, such as flagging a transaction for fraud review or identifying a credit risk. 

 

Once the system has decided what to recommend, this must be fed back to the system conducting the interaction itself—the call center, branch workstation, Web site, ATM, or another.  Modern “customer-facing” systems are designed to make this possible without major modifications.  Older systems can be harder to work with, but technologies exist to allow superficial integration even if the inner workings of the customer-facing system remain hidden.

 

The final step in the inbound marketing process is learning from the results.  The system records the decisions it has made, what was actually presented to the customer (which may not be the same thing if a human discretion is involved), and how the customer responded.  Some systems automatically analyze this information and adjust future recommendations to be more effective.  In other systems, the information is analyzed separately and then reviewed by business people who decide whether changes are needed.  Automated and non-automated approaches each have their advantages, but in practice, even an automated system must be watched closely by human beings to ensure it doesn’t spin out of control.

 

What benefits can marketers expect from these sorts of systems?  Published results can be hard to find, but here are a few.  Key Bank increased revenue per call by 23% after installing a call center recommendation system from eglue www.e-glue.com.   Barclay’s Bank doubled the number of inquiries on portions of its Web site using Omniture’s Touch Clarity www.omniture.com to tailor content based on observed customer activity.  Holland’s Spaarbeleg retail bank added $30 million in sales on one million calls to its service center with SPSS www.spss.com PredictiveCallCenter. 

 

In other words, this isn’t science fiction.  Inbound marketing is a proven approach with very substantial benefits.  It’s one you should give a good close look at your own institution.

 

*                *                  *

David M. Raab is a Principal at Raab Associates Inc., a consultancy specializing in marketing technology and analytics.  He can be reached at draab@raabassociates.com.

Revamping the Software Selection Process: A Modest Proposal

Revamping the Software Selection Process: A Modest Proposal

David M. Raab

DM Review

March 2008 – April 2008

 

Last month I wrote about the importance of scenarios in helping to select the right software.  But even scenarios only help buyers understand how a product works.  To make a truly sound decision, they must know how it will impact their company.

 

This means the software must be assessed in the context of other company systems.  A list of requirements can be misleading because some of the requirements may already be met by other resources.  But you can’t just ignore those requirements: if a new product is better at them than the existing tools, that value of that improvement should matter.

 

To place the new product in context, the evaluation team must first describe how the company functions today, and then see what would change if the new product were added.  If several products are being considered, they must assess the combination of each product with the existing systems.

 

The mechanics of this approach are not so different from conventional assessments.  It still starts with a requirements list, preferably built by analyzing company processes through scenarios.  What’s different is the next step: preparing a set of scores for how well each requirement is met today.  This is the “base case”.  Then the team creates additional sets of scores for the base case plus each new system (less existing systems that would be removed, if any). 

 

The practical impact is that a new system gets credit if meets a requirement better than the existing systems, but is not penalized if it doesn’t.  Instead of comparing the new systems in isolation, you’re measuring the value they add to your business.  This is what you really want to know.

 

Converting “requirements met” to “business value added” is not easy.  With analytical systems, the business value often comes from making better decisions, not from processing transactions more quickly or at lower cost.  How do you measure the contribution that different analytical systems will make to better decisions?

 

One approach is to measure who does the actual work.  The logic is this: in most analytical systems, the primary goal is to answer managers’ questions.  So the best system is the one that gets managers their answers the most quickly.  This depends on how far from the manager each question must travel before reaching someone who can answer it.  

 

In other words, analytical systems involve tiers of users, starting with the manager herself.  If she can answer a question without help, she will.  If not, she will ask a someone on her staff who has greater technical skills and probably more powerful tools.  Let’s call that person a business analyst.  If the analyst can’t answer the question, he will hand it off to specialist such as a statistician working in a corporate decision management group.  If that group can’t get an answer—usually because it requires data that hasn’t already been exposed in a data warehouse or service environment—they turn to the IT department.

 

Each step in this chain adds time.  Therefore, the value added by an analytical system can be measured by how many answers it moves closer to the manager.  For example, automated predictive modeling systems shift the ability to build models from statisticians to business analysts, speeding up the process and lowering the cost dramatically.

 

The table below provides a concrete illustration.  It defines five levels of effort to answer a question:  read an existing report, drill down in a business intelligence system, analyze data extracted from an existing BI or reporting system, analyze data in a data warehouse (but not exposed to end-users in a BI system), and add data not already in the warehouse.   Numbers represent the percentage of questions of each type that can be answered by each user role (they add to 100% reading across).  Column totals show the capabilities and workload for each group.

 

In the example, “Base” shows the current situation and “New” shows capabilities after adding a new business intelligence tool.  As the “Change” row illustrates, the new tool slightly empowers Managers and greatly increases the capabilities of Business Analysts.  The workloads of Statisticians and IT are reduced accordingly.

 

 

% Answers Provided by Each User Type
Case	Question Type	Manager	Business Analyst	Statistician	IT
Base	Read Report	100	-	-	-
	Drilldown in BI	40	40	20	-
	Analyze from BI	30	40	30	-
	Analyze from Warehouse	-	30	70	-
	Add New Data	-	-	30	70
	Score Total	170	110	150	70
New	Read Report	100	-	-	-
	Drilldown in BI	50	50	-	-
	Analyze from BI	30	60	10	-
	Analyze from Warehouse	10	80	10	-
	Add New Data	-	40	30	30
	Score Total	190	230	50	30
Change	Score Total	+20	+120	-100	-40

 

So far so good.  But how do you find the best choice among several systems?  I’ll answer that question next month.

 

*                     *                      *

 

 (Second of a two-part series)

 

The first part of this series showed why the traditional approach of comparing software products against each other less useful than evaluating how they complement existing company systems.  It also proposed a new measure for the value of analytical systems: the number of user groups a question must pass through before reaching someone who is able to provide an answer.

 

The table below repeats the example from last month.  It shows five types of analytical questions, requiring different levels of technical skills and resources to answer.  The “Base” case shows the percentage of questions that can be answered by four types of users with existing systems, and the “New” case shows the percentages with a new system added.  The change in score totals shows that the new system allows Managers and Business Analysts to answer more questions, while shifting work away from Statisticians and IT.  The bar chart illustrates the same changes even more clearly.

 

% Answers Provided by Each User Type
Case	Question Type	Manager	Business Analyst	Statistician	IT
Base	Read Report	100	-	-	-
	Drilldown in BI	40	40	20	-
	Analyze from BI	30	40	30	-
	Analyze from Warehouse	-	30	70	-
	Add New Data	-	-	30	70
	Score Total	170	110	150	70
New	Read Report	100	-	-	-
	Drilldown in BI	50	50	-	-
	Analyze from BI	30	60	10	-
	Analyze from Warehouse	10	80	10	-
	Add New Data	-	40	30	30
	Score Total	190	230	50	30
Change	Score Total	+20	+120	-100	-40

 

This approach gives a good picture of how a new system will affect the organization.  If several new systems are being considered, comparing the charts for each would be enlightening.  But it wouldn’t tell you which one to buy.

 

Picking a single system requires combining the workload figures into a single number that can be used to rank the alternatives.  To do this, the detailed figures must be weighted on two dimensions: question type and user type.

 

Questions types are ultimately the same as user requirements.  Since weighting on user requirements is part of any evaluation methodology, this poses no new challenge.  (The example simply added the five requirement scores, which implicitly weights them equally.)

 

But weights for user types are something completely different.  These reflect the relative value of having different users answer the same question.  Our original premise is that answers from users “closer” to the manager are worth more because they are received sooner.  They are probably cheaper too.  So we know the value weight is highest for answers from managers and lowest for answers for IT. 

 

It’s possible to calculate precise ratios between these weights, based on factors like turnaround time, labor cost, accuracy, and opportunity cost.  But in most cases an intuitive estimate will suffice.  In the example below, weights of 4, 3, 2 and 1 have been applied to the original New and an alternative business intelligence product, New 2. 

   

Summary Value Calculation – New vs New 2
Case		Manager	Business Analyst	Statistician	IT	Combined Value
New	Score Total	190	230	50	30
	x Value Weight	4	3	2	1
	= Value	760	690	100	30	1,580
New  2	Score Total	200	140	130	30
	x Value Weight	4	3	2	1
	= Value	800	420	260	30	1,510

 

Examining the user-level scores, we see that New 2 gives Managers slightly more capability than New (800 vs. 760), but Business Analysts gain much less (420 vs. 690).  The combined value for New is higher than New 2 (1,580 vs. 1,510), making New the better choice.

 

Where does this leave us?

 

In a better place, I think, than the traditional selection process.  Instead of a horse race between product features, this approach puts focus where it should be: on value to your business.  It recognizes that the value of a new tool depends on the other tools already available, and it forces evaluation teams to explicitly study the impact of different tools on different users.  By creating a clearer picture of how each new tool will impact the way work actually gets done at the company, it leads to more realistic product assessments and ultimately to more productive selection choices.

 

*                     *                      *

 

David M. Raab is a Principal at Raab Associates Inc., a consultancy specializing in marketing technology and analytics.  He can be reached at draab@raabassociates.com.

 

 

 

 

 

 

 

Software Selection Mistakes

Software Selection Mistakes

David M. Raab

DM Review

February 2008

 

Selecting the right software cannot guarantee the success of a project, but picking the wrong system can ensure failure.  Here are three common errors and how to avoid them.

 

1. Overly detailed requirements.  This error is so common that it’s sometimes mistaken for a best practice.  Selection teams list hundreds of desired system functions and embed them in a Request for Proposal.  Vendors are expected to answer, often in a yes/no format, without any opportunity to understand the reasons for the requirements or to explain how their solution would meet them.  The selection team then scores the results, and the software with the most tics wins.

 

This approach creates much work and little information.  Vendors cannot describe how their products would best meet the client’s needs, and clients gain little insight into how the products function.  It’s better to help vendors build a thorough understanding of your situation and goals, and then let them propose a solution based on their product.  In addition to getting the vendor’s best ideas, this process gives your team a good idea of what the vendor will be like to work with.

 

This approach does not do away with the need to understand your requirements.  It’s perfectly possible for a vendor to propose a solution that won’t actually meet your needs.  You must build your requirements list in advance so you can compare it against the vendor’s recommendation.  And, yes, you should share this list with the vendors—this is a business project, not a child’s game of “gotcha”.

 

2.  Canned demonstrations.  Project teams often follow their Request for Proposal with invitations for the most promising vendors to demonstrate their software.  Demonstrations play the same role as an automobile test drive: they let you discover what it’s like to actually use the product.  But just as you wouldn’t be satisfied with sitting in the passenger seat while the dealer drives for you, you can’t simply watch someone else run a piece of software.  You need to take control, which includes both running the system and choosing what to test.  For an automobile, this might mean driving on different roads in different weather conditions.  Maybe you’d even hook up a trailer if that’s your intended use.  The software equivalent is running through the relevant business processes – setting up a campaign, processing an order, handling a phone call, and so on. 

 

The first step in this type of testing is personally running the tasks on the demonstration system.  This can be enlightening, because vendors often structure their planned demonstrations to avoid known weaknesses in their product.  On an even more basic level, something that looks simple in the hands of an experienced demonstrator can turn out to be considerably more painful when you are pushing the buttons yourself.

 

But you’ll usually want to go beyond the demonstration system to see how the product would function in your own environment with your own data.  If actually connecting to your own systems is not practical, the vendor can still show you the steps required to do it.  This will give you a much clearer idea of the work required to deploy the software and will help identify challenges it adopting it to your data models.  It’s important that your team have the right technical experts present for this discussion, so they can provide information, ask the right questions, and understand the implications of the vendor’s answers. 

 

3.  Uninformed evaluation.  Many software vendors offer evaluation copies of their products.  This is the exact opposite of a controlled demonstration since users can do whatever they want.  But users testing a product on their own may underestimate its capabilities because they don’t understand them properly.  This is why the auto salesman shows you the controls before your test drive.  Software vendors often provide an evaluation guide or tutorial that illustrates key product features.  Some share the complete user documentation.  For complex products, assistance may extend to offering the time of sales engineers or technical support staff. 

 

The mistake here is to not take advantage of those resources.  Evaluators often try to learn the products just by loading and running them.  They sometimes rationalize this as “testing for ease of use”, but unless you actually plan to deploy the software without training your staff, that’s a poor excuse.  You will eventually try to run your own processes on the evaluation system, but must first start by learning how it works.  Remember, your ultimate goal is to gather correct information about each product.  Undervaluing a product due to poor evaluation is as much an error as overvaluing it because of vendor hype. 

 

Different as they are, these errors all have on thing in common: avoiding them requires creation of scenarios that illustrate how the system will be used.  Scenarios provide a reference point for vendor proposals, determine which features to explore during a demonstration, and structure the time spent with an evaluation copy.   They ensure the evaluation is grounded in actual business needs and that it covers the key processes from start to finish.  Although creation of scenarios is hard work, it is the best way to avoid the ultimate selection nightmare of purchasing a product, installing it, and immediately discovering it doesn’t do what you really need.

 

*                     *                      *

 

David M. Raab is a Principal at Raab Associates Inc., a consultancy specializing in marketing technology and analytics.  He can be reached at draab@raabassociates.com.

 

 

 

 

 

Building Customer Relationships

Building Customer Relationships

by David M. Raab

Curtis Marketwise FIRST

February 2008

 

We’ve all heard the saying, “Military justice is to justice what military music is to music.”*.  Something similar applies to customer relationships: they also serve a purpose different from relationships in general.  Not to put too fine a point on it, the purpose of customer relationships is to make money.

 

Every businessperson knows this.  But it’s still easy to get carried away with the romance of building a relationship management program.  So let’s be clear: companies nurture customer relationships so they can sell more, reduce costs, or keep customers longer.  If a program doesn’t serve at least one of those goals, it isn’t worth having.

 

But how, exactly, do relationship management programs do this?   Other than creating warm feelings towards your bank—not a very reliable motivator, alas—relationship programs produce specific benefits.  These include:

 

- reduced sales effort because customers trust the bank will offer products that suit their needs

- more information for targeting because customers are more willing to share their data with you

- first chance at providing new products because customers turn to the bank before checking elsewhere

- less price competition because customers will do business with the bank unless an alternative is substantially cheaper

- more referrals because customers are satisfied with their own treatment

- higher switching costs for customers because they get multiple services from the bank

- lower service costs because customers understand bank processes and self-service systems

 

Relationship building investments can be judged by their contribution to these benefits.  Let’s see how a few common investment opportunities stack up.

 

- Customer Relationship Management (CRM) systems.  These systems deliver customer data to sales and service personnel.  But the data contributes little to customer relationships unless bankers are also given tools to use it wisely.  This is why basic CRM systems are often supplemented with analytics that identify the best treatments for each customer.  This helps bankers improve the relationship by making more relevant suggestions.  CRM systems can also make the bank easier to deal with—and thus preferred over competitors—by letting bankers easily find customer data, thereby speeding and simplifying many interactions.  In addition, the CRM system provides a convenient mechanism for capturing customer information in the first place.

 

- Self-service systems.  These create opportunities for customers to bank when, where and how they want to.  They include ATMs, Web sites, automated telephone systems, mobile devices, and whatever the technologists will dream up next.  Self-service has many advantages: lower costs, greater customer convenience, barriers to switching because of the effort to learn someone else’s systems, and incentive to add new services that share data or functions with existing ones.  The challenge to marketers is that customers often resist self-service systems at first, both because of the effort to learn them and because initial implementations often harder to use then traditional methods.  Yet the benefits that banks gain from these systems can justify substantial investments in encouraging adoption—something not all bankers have fully realized. 

 

- Customized services.  These are services such as alerts for low balances, overdrafts, or market events.  They are nearly always self-managed by customers, so they could be considered a type of self-service system.  But the technologies involved are different enough that they should be treated separately.  The relationship aspects are different too: these services are less about “high tech” efficiency than “high touch” personal treatment.  This suggests a different approach to promoting these services, even though they face the same adoption hurdles (customer awareness and training) as self-service.  Their primary relationship benefit is improved retention, since customers are reluctant to spend the time to convert to another bank’s system, and even more reluctant to convert to a bank that doesn’t offer the services at all.  On the other hand, they don’t really save money, since the services would not otherwise be provided at all.  And while they may generally improve a customer’s attitude towards the institution, they are largely tied to specific products, so they provide little direct incentive for customers to add new ones.

 

- Targeting.  Banks can choose from a variety of tools to select offers for individual customers.  Labels and technologies overlap, but it’s worth distinguishing three types of targeting systems based on how they are used:

 

            - Event detection systems analyze customer transactions for patterns that indicate opportunities such as a funds to invest or risk factors such as loss of a job.

 

            - Recommendation engines analyze interactions like Web site visits or telephone calls as they happen and suggest appropriate offers based on customer behavior.  These engines may also factor in other information about the customer if it can be linked in.

 

            - Predictive models use historical data to select offers for outbound promotions such as direct mail and email.  They are also sometimes applied in real time within recommendation engines.

 

In each case, the goal of the system is to make better recommendations.  This contributes directly to increased sales by making more effective use of business opportunities.  It also improves the relationship indirectly by building trust that the company “understands” customer needs and acts to fill them.  Of course, trust will not be built if the recommendations appear inappropriate or, even worse, contrary to the customer’s best interest.

 

- Branding.  Brand marketing and relationship marketing are sometimes treated as opposites, but this is a false dichotomy.  It is based on the idea that brand marketing is aimed at masses while relationship building targets individuals. This is often (but not always) true, but it doesn’t matter.  A strong brand will encourage customers to do business with the company, to trust it, and to accept premium pricing.  So even though the messages may be targeted differently, the business benefits are the same.  This positions brand marketing as a valid competitor for relationship building funds.

 

Which tool is the best choice for relationship investments?   That depends on the value provided in return.  Measuring that value can be difficult, but we know it will be based on the relationship benefits presented above.  Even an informal comparison of the benefits of the different relationship building tools will help you make a sound decision.

 

*usually attributed to Groucho Marx, sometimes to Georges Clemenceau.  Take your pick.

 

*                *                  *

 

David M. Raab is a Principal at Raab Associates Inc., a consultancy specializing in marketing technology and analytics.  He can be reached at draab@raabassociates.com.

 

 

 

 

 

 

 

Single View of the Truth

Single View of the Truth

David M. Raab

DM Review

January 2008

           

            One of the bedrock goals of data warehouse projects is a “single version of the truth”.  Yet truth is rarely so simple.  In the classic example, a “customer” is one thing to a sales person, another to the shipping department, and something else to accounts payable.  Good data warehouse designers recognize this and build different definitions into their systems, so users can access whichever version they need whenever they need it.

            Yet the notion of a single version of the truth persists—and warehouse teams invest huge resources negotiating shared business models to define it.  Why?

            The problem is often described as “dueling spreadsheets,” where managers argue over whose data is correct.  This is apparently something to avoid at all costs.

            Personally, I love a good debate over data.  But if you do want to prevent those arguments, you have to understand what causes them.  Just providing a “single version of the truth” won’t do the trick, precisely because any data warehouse rich enough to be useful will contain enough variations of the truth that it, too, can produce conflicting results.

            Let’s start at the beginning.  Managers rely on whatever data sources they have available.  In the absence of a warehouse, these are usually their local operational systems.  Managers use these systems not just because they are handy, but also because they understand their contents.  Since learning about a data set is often the hardest part of an analytical project, it’s perfectly reasonable for managers to rely the data they know. 

            “Dueling spreadsheets” happen because each manager’s local data set is an incomplete view of an entire problem.  Call center managers can see call center information, and might do an analysis that shows how to minimize call center costs.  But the service manager will see service costs that result from poor call center treatments, such as dispatching repair people for problems that could have been resolved over the phone.  Each manager can analyze her own data correctly and reach opposite conclusions about the best course of action. 

            Putting all that data into one warehouse wouldn’t solve the problem.  The single version of the truth (that is, a shared data model) will include both call center data and service data.  If each manager simply extracts her own department’s information, she will still end up with conflicting results.

            The only thing that will change this is if both managers pull both departments’ data.  Indeed, each manager really needs all the relevant data, which probably comes from many departments.  Here is where the central data warehouse truly adds value: it makes all that data accessible in an integrated format.

            But this brings us back to the original problem.  Managers will use the data they find most familiar.  Even if they have access to a comprehensive warehouse, pulling data for all different departments requires understanding where to find that data and how to combine it.  Managers are unlikely to take the time to learn how to do this.  Instead, they’ll either go back to their familiar local sources or pull the equivalent data from the central warehouse.  Either way, they get the same, incomplete result.

            This problem can be mitigated but not really solved.  It can’t be solved because managers don’t have the time or inclination to learn about proper warehouse procedures.  Mitigation means making it easy for managers to see the data they really need, even if they didn’t think to look for it.  It also means making it at least as easy to get that data from the warehouse as from local systems.   

            Waiting for the IT department to build a new data cube definitely does NOT count as easy, particularly if the manager could already pull it from the local system for herself. 

            It’s tempting to “solve” the problem by mandating use of warehouse information, but that probably won’t work.  Many managers will just do without the information rather than invest major time in learning a new system.  Or they’ll look at data from the local system and not show it to anyone else.  Or they’ll ask an analyst to do the work for them, but only when it’s worth the extra time, cost and trouble.  Remember we’re talking here about managers, who have considerable discretion in how they do their jobs.

            Analysts are another story.  Learning new systems is part of their job and, if they’ve made the right career choice, it’s something they enjoy.  Moreover, they are part of a community of other analysts that should enforce its own standards for the quality of work.  Therefore, it’s perfectly plausible to require that analysts draw their data from a warehouse.  Not that this should be a problem: all possible data assembled in one place is an analyst’s fondest dream come true.  Requiring them to use a warehouse should be about as hard as requiring a kid to visit a candy shop.

            In short, the warehouse is an analyst’s tool.  Getting managers to use it adds requirements which companies may or may not decide to meet.  Achieving a “single version of the truth” takes more than a unified data model: it means a thorough change in how companies analyze their data.  Warehouse teams that fail to recognize this can never meet expectations.

 

 

*                     *                      *

 

David M. Raab is a Principal at Raab Associates Inc., a consultancy specializing in marketing technology and analytics.  He can be reached at draab@raabassociates.com.

 

 

 

 

 

 

 

How to Judge a Columnar Database

How to Judge a Columnar Database

David M. Raab

DM Review

December 2007

 

            Columnar databases are an idea whose time has come – again.  First introduced in the 1970’s in (still available) products like Model 204 and ABABAS, this approach has resurfaced recently at Vertica (www.vertica.com) and to some extent at QD Technology (www.qdtechnology.com).  These join resurgent columnar database vendors Alterian (www.alterian.com) and SmartFocus (www.smartfocus.com), whose original products date to the 1990’s.

            As the name implies, columnar databases are organized by column rather than row: that is, all instances of a single data element (say, Customer Name) are stored together so they can be accessed as a unit.  This makes them particularly efficient at analytical queries, such as list selections, which often read a few data elements but need to see all instances of these elements.  In contrast, a conventional relational database stores data by rows, so all information for a particular record (row) is immediately accessible.  This makes sense for transactional queries which typically concern one record at a time.

            Columnar databases were largely eclipsed by relational systems in the 1980’s and 1990’s, when huge improvements in hardware price / performance allowed relational databases to compete effectively despite their analytical inefficiency.  Columnar technology may be re-emerging today because analytic databases are now so large that the hardware required to use relational systems is too expensive even at current prices. 

            Today’s columnar systems combine the columnar structure with techniques including indexing, compression, and parallelization.  But the fundamental questions asked in evaluating these systems are the still the same.

            - load time: how long does it take to convert source data into the columnar format?  This is the most basic question of all.  Load times are often measured in gigabytes per hour, which can be unbearably slow where tens or hundreds of gigabytes of data are involved.  The question often lacks a simple answer, since load speed can vary with the nature of the data and choices made by the user.  For example, some systems can store multiple versions of the same data, sorted in different sequences or at different levels of aggregation.  Users can build fewer versions in return for a quicker load, but may later pay a price in slower query times.  Realistic tests based on your own data are the best path to a clear answer.

            - incremental loads: once a set of data has been loaded, must everything be reloaded every time there is an update?  Many columnar systems allow an incremental load, taking in only new or changed records and merging them with the previous data.  But close attention to detail is critical, since incremental load functions vary widely.  Some incremental loads take as long as a full rebuild; some result in slower performance; some can add records but not change or delete them.  Often incremental loads must be supplemented periodically with a full rebuild.

            - data compression: some columnar systems greatly compress the source data so the resulting files take a fraction of the original disk space.  There may be trade-offs: uncompressing the data to read it can slow performance.  Other systems use less compression or store several versions of the compressed data, taking up more disk space but gaining other benefits in return.  The most suitable approach will depend on your circumstances.  Bear in mind that the difference in hardware requirements can be substantial.

            - structural limitations: columnar databases use different techniques to mimic a relational structure.  Some require the same primary key on all tables, meaning the database hierarchy is limited to two levels.  The limits imposed by a particular system may not seem to matter, but remember that your needs may change tomorrow.  Constraints that seem acceptable now could prevent you from expanding the system in the future.

            - access techniques: some  columnar databases can only be accessed using the vendor’s own query language and tools.  These can be quite powerful, including capabilities that are difficult or impossible using standard SQL.  But sometimes particular functions are missing, such as queries that compare values within or across records.  If you do need to access the system with SQL-based tools, determine exactly which SQL functions and dialects are supported.  It’s almost always a subset of full SQL; in particular, updates are rarely available.  Also be sure to find whether performance of SQL queries is comparable to performance with the system’s own query tools.  Sometimes the SQL queries run a great deal slower.

            - performance: columnar systems will usually outperform relational systems in nearly all circumstances, but the margin can vary widely.  Queries involving calculations or access to individual records may be as slow or slower than a properly indexed relational system.  Create a set of sample queries and test them against a prototype system. 

            - scalability: the whole point of columnar databases is to get good performance on large databases.  But you can’t assume every system will scale to tens or hundreds of terabytes.  For example, performance may depend on loading selected indexes into memory, so your hardware must have enough memory to do this.  As always, first ask whether the vendor has existing systems running at a scale similar to yours and speak to those references to get the details.  If you would be larger than any existing installations, then be sure to test before you buy.

 

*                     *                      *

 

David M. Raab is a Principal at Raab Associates Inc., a consultancy specializing in marketing technology and analytics.  He can be reached at draab@raabassociates.com.

 

 

 

 

 

Bank Marketing Technology Trends for 2008

Bank Marketing Technology Trends for 2008

by David M. Raab

Curtis Marketwise FIRST

December 2007

 

            There were plenty of shiny technical gizmos for bank marketers in 2007, and 2008 will bring even more.  But like that hot video game which is always out of stock at your local store, much of the latest technology will remain out of reach for many bank marketers.   Whether it’s too costly, incompatible with existing systems, or out of synch with current priorities, there are plenty of reasons to miss the Next Big Thing.

            So which technologies are likely to have widespread impact?  Here are some strong candidates.

 

·          Online marketing management systems.  Web pages, email and mobile devices aren’t new anymore, but the technology to use them is still evolving rapidly. The most important change next year will be the increased availability of integrated systems that make it cheaper and easier for marketers to assemble sophisticated cross-channel campaigns.  The latest tools unify Web and mobile ad placements, search keyword purchases, outbound emails, personalized landing pages, and automated email response.  They support these with shared customer databases, content management, project workflow, and results analysis.  Some can execute multi-step, rule-driven contact streams that essentially put customer management on auto-pilot—although this more than many institutions are ready to do.

 

      These systems let bank marketers set up, execute and evaluate integrated online campaigns for themselves, often at less cost than outside resources now charge to work in each channel separately.  Since the new systems are often hosted (that is, run on the vendor’s computers and purchased via monthly subscription) , the start-up cost and technical support burden are low enough that even institutions with limited resources can afford them.  The net result will be that increasing numbers of bank marketers can make greater, more creative use of the online channels.

 

·          New online techniques. Although older online techniques are well established, new concepts are still appearing.  Social networks like FaceBook and MySpace, virtual worlds like Second Life and Whyville, and communication tools like wikis and  blogs all present new opportunities and challenges for innovative bank marketers.  In most cases, the cost of entry is very low, so the usual barrier to experimentation is not present.  On a more prosaic level, tools like online surveys, e-newsletters, and “buzz” monitoring services provide new ways to understand and change consumer attitudes—again, at much lower cost than conventional methods of gathering and distributing similar information.  The real opportunity is gathering the same information for less money, but using the same budget to do more—for example, capturing actionable data from the entire customer base instead of surveying only a small sample.

 

·          Self-service analytics.  This is delicate topic because so much has been promised in the past, and so little delivered.  Clever software will not turn every branch associate into a Ph.D. statistician.  In fact, few front-line personnel have the time or inclination to engage in serious analytics regardless of the tools they are given.  But marketing managers and analysts do have that time and inclination; what they often lack to their immense frustration is access to the data.  New technologies such as in-memory databases and visualization software now allow managers and analysts to extract and analyze data directly from marketing databases, data warehouses, or sometimes core systems themselves, with minimal help from IT staff. 

 

      Specifically, the role of IT is to set up and manage the connections that make the extracts possible.  Few marketers have the technical skills to create these connections, but, even if they did, IT would still be in charge to ensure security and performance.  The change is that older business intelligence and reporting systems generally required IT to prepare data summaries in response to particular questions, while the newer technologies pull in the raw data and let marketers aggregate it for themselves.  Entry prices for products in this group–QlikView, Tableau, ADVIZOR, Spotfire, Miner3D—are in the thousands or, at most, tens of thousands of dollars.  This is well within the budget of nearly any marketing department, and an even greater value once you factor in the labor savings for IT.

 

            The low cost and high value of these technologies should make them very successful in the next year.  Other technologies will be more limited to institutions with an aggressive appetite for innovation.  Opportunities for these organizations include: 

 

·          Behavior detection.  Systems like Harte-Hanks Allink Agent and Conclusive Marketing SynapseEBM scan customer transactions for patterns that indicate a business opportunity and pass the resulting lead to automated systems or sales people for action.  The approach is well proven but can stumble over the connections between the behavior detection system and the systems that use its results.

 

·          Touchpoint analytics and guided selling.  This modifies front-line systems such as call centers, branch automation, ATMs and Web sites to add analytical software that monitors customer behavior during an interaction and recommends appropriate responses.  Again, the benefits are well documented but implementation is often hampered by difficulties integrating with the front-line systems.

 

·          Enterprise integration.  Local branches can access corporate marketing systems to generate custom direct mail, download advertising materials, and order promotional items.  Central lead generation and referral management systems can pass information to sales people and account officers.  Call centers can be decentralized so agents can work from home.  All are excellent opportunities to reduce cost and improve performance – if the corporate infrastructure is open to these sorts of connections.

 

·          Value-based customer management. There is a growing recognition of the importance of managing by customer value. This requires a host of technical and analytic resources: accurate profitability measurement; improved risk analysis; extensive predictive modeling; simulation and optimization.  Implementation is a long process that moves in stages: first the company must measure the value of individual customers; then, marketing and operational systems must customize treatments based on these values; finally, the company must measure the impact of each treatment on future value so it can select the best ones.  Companies like RBC Royal Bank have profited from this approach, but only after years of disciplined investment in people and process as well as technology.  If your bank has the vision and resources to move in this direction, 2008 would be a great time to start.

 

 

*                *                  *

 

David M. Raab is a Principal at Raab Associates Inc., a consultancy specializing in marketing technology and analytics.  He can be reached at draab@raabassociates.com.