The volume of customer data is proliferating at an ever-increasing rate. There is data from transactions, from website visits, from social networks and of course and from many other sources. And there is hidden data about customers fundamental needs that drive successful innovation. All of this data needs bringing together intelligently, so that it can be used by customers and companies to co-create value together.
Companies have mostly focused on the mostly low-hanging fruit of transaction data, such as mobile phone call records, items purchased with a credit card, or supermarket items purchased, to-date. In the hands of a customer-intelligent company like Tesco, this can yield remarkable results. But for most other not-so-smart companies, the results are nowhere near as impressive. They have spent vast sums of money on marketing automation that struggles to achieve much more than a 10% increase in marketing success. And throwing yet more expensive technology at the data won’t help in the long-term either: the current craze of real-time analytics is just more technology for the same results but quicker.
Customers are getting restless too. Fed up with being spammed by direct marketers abusing the low hanging fruit of transaction data, they are starting to demand control over their own data and to be paid by companies wishing to use it for marketing. This is threat to the old way of doing things but an opportunity for innovative companies. UK mobile telco startup Blyk offers young customers free mobile telephony in return for listening to adverts on their phones. Blyk more than doubled its target of 100,000 customers in the UK in its first year. And customers complained that they got too few adverts, rather than too many. Try asking most bank or telco customers if they want more direct marketing! More and more customers are fed-up of direct marketing spam and opting completely out of marketing altogether. Direct marketers have created their own tragedy of the marketing commons .
As the opportunities to use customer data multiply but customers become less accepting of direct marketing, companies need to take a new look at the sources of customer data, at how useful the data is and at how best to use it to co-create mutual value with customers. Just like Kaplan & Norton did in the 1990s for performance management data, companies now need a Balanced Scorecard for Customer Data:
- Transaction Data – Like financial data in Kaplan & Norton’s original Balanced Scorecard, transaction data looks backwards at customers’ previous purchases. Companies assume that what customers did in the past is a good guide to what they will do in the future, which has lead to the widespread use of e.g. RFM models in catalogue retailing. But as companies have found out, the past is a very imperfect guide to the future. Even though most companies still rely on transaction data for the majority of their customer analytics, the results are typically <10% improvements in marketing response rates at best. The average response rate over all direct marketing is still only a miserable 1-2%.
US credit card company Capital One is one of the legends of customer analytics. It’s business analysts continuously analyse customer transaction data to identify opportunities to make highly selective offers to micro-segments of as few as a few hundreds of customers over a very short window of opportunity.
- Contextual Purchasing Data – Like process or customer data in the original scorecard, contextual purchase data looks at things that guide current purchasing decisions. This might include clickstream data about a customers use of a website, critical events in the customer’s lifecycle, or other point of purchase data that helps make sense of the customers individual buying behaviour. Of all the data available to companies, contextual purchase data is some of the most useful, yet few companies have access to this data today. As a result, they don’t really know what drives a customer to purchase a product or to leave it on the shelf. Where companies use contextual purchase data, the results can be up to 30% improvements in marketing response rates.
Toyota Financial Services identified new vehicle customers coming to the end of their finance or lease contracts and analysed their contractual conditions. By making a highly customised vehicle plus leasing offer for each customer coming to the end of their contract, Toyota increased the number of customers repurchasing another new vehicle and taking out a new finance or leasing contract by over 30%.
- Purchase Influencer Data – Like the contextual purchasing data already discussed, purchase influencer data looks at others who influence customers’ current purchasing decisions. This might include on-line social network data, telephone calling community data and even data about who accompanies you when shopping. We are social animals and as been shown many times, are much more influenced by others, their opinions and their behaviour, than we like to think. This applies both to customers’ online and off-line behaviour. In fact, despite the current emphasis on on-line social networks like Facebook, in reality, the vast majority of social influence is still wielded off-line through friends and family.
Finnish social network analysis startup Xtract analysed the customer calling communities of a UK triple-play telco to identify the most influential customers. By targeting these customers with retention offers, the telco reduced customer retention by 26%, compared to 19% from traditional churn analytics, a 37% incremental reduction in customer churn.
- Customer Needs Data – Like innovation & growth data in the original scorecard, customer needs data looks forward at what guides core customer buying behaviour. This is data about customers’ aspirations, about the jobs that customer are trying to do and about the outcomes customer desire from doing them. Customer needs are stable over time but the products they hire to do them change, e.g. the job of listening to music at home has not changed, but customers have hired 78s, LPs, 45s, tapes and most recently iPods to do the job. In the same way, understanding what customers really need can be harnessed to drive innovation in products, services and experiences to better meet their needs. Ultimately, this is the only data that will help you define how you can better meet customers’ needs better in the future. It is the only data with which to guide the future growth and success of your company.
Outcome-driven Innovation consultancy Strategyn works with companies like Microsoft, Motorola and Bosch to harness customer needs to drive innovation. These companies have seen success rates of up to 80% for new products, services and experiences introduced into the market. That is a huge difference to the 80% failure rate for most new product introductions.
Companies need to start to think about all aspects of customer data, not just the low-hanging fruit of transactional data. That includes contextual purchase data, purchase influencer data and customer needs data. The companies that win through customer analytics will be those that can do this successfully and those that involve customers more fully in the process.
What do you think? Is transaction data enough for your needs. Or do you need to build a Balanced Scorecard of customer data?
Post a comment and get the conversation going.
Tip of the hat to Mike Compton of Optima for the phone discussion that triggered this train of thought.
Graham Hill
Customer-driven Innovator
Follow me on Twitter
Interested in Customer Driven Innovation? Join the Customer Driven Innovation groups on LinkedIn or Facebook to learn more
Further Reading:
Kaplan & Norton, The Balanced Scorecard – Measures That Drive Performance
– well structured and supported with relevant examples. It works well with the knowledgeable community of Customer Thinkers, but for the wider audience of marketers and customer management practitioners it is perhaps worth highlighting that the proposed tool is not a K&N Balanced Scorecard: the similarity is that it is also multi-dimensional and you look at many types and aspects of data, but it needs to be designed differently – for the purpose of leveraging data assets (as opposed to operational performance measurement).
The proposed approach: taking inventory of data, introducing a classification structure by type, then defining relevant uses for each type (and combinations thereof)- is very logical and is (nearly) a blueprint for someone with the ambition and time to roll their sleeves and develop the tool.
A possible alternative approach that I would normally take is a typical ‘reverse engineering’ process: What challenges do we need to resolve, wnat do we want to do with customers? -> establish a classification of tasks and processes by purpose and outcome -> What data would we need for this? -> Is it available and accurate, where it resides, where can it be obtained from? The end result may well be the same, but in this way I feel it would be better aligned with desired business results (and customer experiences, too).
In either case, this has the potential to become a rather useful neat little tool. We saw it here first and owe the idea to Graham. It may also be a collaborative opportunity: I can offer some volunteering time; anyone else interested to play with this and come up with a prototype tool?
Vladimir Dimitroff
Director, PRISM Consulting
Hi Vlad
Thanks for your comment. It is always great to hear your erudite opinions.
The purpose of the blog post was to show companies that there is more to customer data than just transaction data. Transaction data provides many actionable insights, but the addition of current data about the purchase process and future-looking data about customers’ needs can provide far more.
The Balanced Scorecard model was used simply because the principle of balancing different types of data to overcome the overreliance on rearward-looking transaction data is similar to the problems Kaplan & Norton tried to solve when developing the Balanced Scorecard (the overreliance on rearward-looking financial measures).
I can see the advantages of the inside-out approach you suggest for further developing the Customer Data Scorecard. By reviewing what is available that supports what marketers want to do, they can improve the execution of their marketing plans. That is one approach. But I don’t think it is necessarily the best one. It focusses on the marketer and not on the customer.
My experience as an operational marketer is that marketers’ plans are often developed without anything more than a cursory undersanding of what customers really want, of how they buy and of what alternatives they are looking at. Marketers in effect plan to spend their budget achieving internal marketing goals, not to get customers to spend their budget by helping them meet their own goals. As we have seen, the effectiveness of marketing has been falling for years. And why, to be blunt about it, so much of what passes for marketing today is simply unfocussed crap!
If I can help marketers (and CRMers more generally) to understand that they need to look at all four aspects of customer data, then they will automatically start to understand what customers really need, how they buy and thus, how to bring together just the right information to drive innovations in marketing, to improve how they go to market and to grow the top-line.
As you suggest, more work needs to be spent on developing the framework further. I have already started thinking about how to create a simple toolset around the Customer Data Scorecard that marketers can use to do their job better.
Ultimately, marketing should start with understanding customers’ needs. Now where have I heard that before. Oh yes, in Ted Levitt’s seminal article on ‘Marketing Myopia’ in the 1970s!
Graham Hill
Customer-driven Innovator
Follow me on Twitter
Interested in Customer Driven Innovation? Join the Customer Driven Innovation groups on LinkedIn or Facebook to learn more.
> “…Ultimately, marketing should start with understanding customers’ needs. Now where have I heard that before. Oh yes, in Ted Levitt’s seminal article on ‘Marketing Myopia’ in the 1970s!”
They have been trying – even before Levitt, but unfortunately Needs are something fuzzy and fluffy, difficult to quantify and analyse. Used to percentages and pie charts, for more than half a century they have been using ‘proxies’ – intermediate variables acting as indicators of needs. Demographics, lifestyles, attitudes, behaviours – they’ve got so deep into such ‘insight’ that they forgot what it was about, they don’t (even try to) see the real need behind the ‘proxy’.
This involves a lot of assumptions (‘couples with babies are likely to need Pampers’ and ‘young mobile subscribers probably prefer pop music to classical’). (Remember business rule #1? ‘Don’t assume!’)Decades ago this was OK when segmenting markets (masses of anonymous people) as opposed to customers (identifiable individuals about whom we hold rich data). This is unforgivable today but, sadly, most marketers still live in the 20-th century…
Little excuse for the conitnuing obsessions with demographics (e.g. everyone is after that ‘Youth’ segment..), when it is well proven that behaviours (transactional, purchasing, outside our business) are a far more reliable predictor of Needs – and are far better documented in our databases. One just needs ‘a look’ (a fairly simple analysis) at in-house data, yet big money is spent on market research extrapolating ‘insight’ (should we call it ‘outsight’?) from ‘representative’ samples of strangers…
These ‘random’ thoughts are, in fact, things I would like to see as dimensions/criteria in a Customer Data BSC. And more 🙂
Vladimir Dimitroff
Director, PRISM Consulting
Hi Vlad
Thanks for the follow-on comment.
The purpose of the Customer Data Scorecard was to highlight that companies need more varied data to improve their customer analytics.
I agree with you about demographics, psycographics and all the various commercial aggregation databases. They are better than absolutely nothing, e.g. if you are completely new to a market and have no idea who or where to target, but they are much less useful than historical transaction data. Unfortunately, the effectiveness of historical transactions is declining as markets become more fragmented and it has never been all that high anyway. An average response rate for direct marketing of 1-2% is not my definition of success, even though it may still be profitable for marketers. And customers are sick and tired of being marketed at by companies that are only interested in their wallets.
Companies need to gather new kinds of data, such as the contextual purchasing data that describes how customers buy and the purchase influencer data that describes who influences customers to buy. Work in online retail clickstream analysis and telco customer calling community analysis is a great start in this area. But even this is not enough without understanding customers’ needs over their entire consumption cycle. Particularly their post-sale needs where most of the value is delivered to the customer (but which most companies treat as generating costs-to-be-avoided).
Customer needs are hard to quantify if you use the traditional Voice of the Customer (VoC) approach. VoC suffers from three core problems:
These difficulties have plagued the traditional VoC approach. Despite the attraction of giving customers their voice, VoC-driven innovation doesn’t have a strong record of success either. VoC-driven new products has a failure rate in the market little better than the 80% failure rate of invention-driven new products.
To overcome these difficulties, leading-edge customer researchers now asks customers about their needs in terms of the jobs customers are trying to do and the outcomes they desire from doing them. This provides a standadr for defining needs that is missing from VoC. It also provides a foundation for thinking about the end-to-end customer experience that relates strongly to the other three categories of customer data that companies should collect.
As I said in my earlier response to your comment, “Ultimately, marketing should start with understanding customers’ needs”. Through jobs & desired outcomes we finally have a simple tool to really understand what customers’ need. And to relate them through the other categories of customer data to insight-driven action.
Graham Hill
Customer-driven Innovator
Follow me on Twitter
Interested in Customer Driven Innovation? Join the Customer Driven Innovation groups on LinkedIn or Facebook to learn more.
Great discussion, right on the money! I have had the chance to visit some of my clients in Europe this week where I have been proposing some (though not all) of these same ideas.
To paraphrase you, and to drill-down further on the clickstream aspects:
* web analytics can provide transaction data for online purchases
* but what most companies are still blind to is the fact that online purchases are only a small subset compared to online influenced OFFline purchases.
* So much more importantly, clickstream analysis can provide that contextual data on individuals’ preferences for direct marketers to use
* At the aggregate level, web analytics can also document some customer preference data, if not needs data. E.g. car manufacturers measure the interactions with their online car configurators to forecast next month’s purchases by product end even including details such as the preferred engine option. According to one case study the online data showed almost a perfect correlation!
Finally, how good of you to include the purchase influencer point! This is the part I didn’t include myself. But this data is very different than web analytics data:
1. This data is not available at the individual level, only in aggregate
2. This data doesn’t come from web analytics but from blog monitoring, etc.
Because of 1+2 I thought that your reference to a “BSC” made a ton of sense since the latter also shows disparate types of metrics.
Thanks as always!
Akin
Hi Akin
Thanks for your comment. It is always great to hear your thoughts and ideas.
Your description of clickstream purchase context data is very useful. It is one kind of contextual data. Others include information about customer events, e.g. moving house, about product triggers, e.g. laser printer toner empty, about retail experiences, e.g. shopping path in a supermatket and other information. To misquote Donald Rumsfeld, we still don’t know what we don’t know about the context of customer purchasing.
It is important to look beyond the microcosm of on-line data. As sophsiticated writers about CRM, we often forget that the vast majority of people are still occasional or non-internet users, even when looking for big ticket items. The macroscosm is mostly off-line and will stay so for some time. The real challenge isn’t integrating on-line data into the customer data scorecard, but identifying what off-line data is useful but missing and how to gather it.
I don’t believe the automotive car configurator case study. Having spent years working in automotive, the configurator tends to be first used some months ahead of the point of purchase rather than at the last minute. And once again, the majority of customers never go near the manufacturer’s website, let alone use an on-line configurator. They look at their friends’ cars, go visit a dealership, or look at car magazines. They live in the real world.
The customer’s world is still 90% off-line. And that’s exactly where we need to look for their data.
Graham Hill
Customer-driven Innovator
Follow me on Twitter
Interested in Customer Driven Innovation? Join the Customer Driven Innovation groups on LinkedIn or Facebook to learn more.
Graham: A book I recently read, Planet Google, describes how Google went the other way on customer data. It was fascinating to learn how valuable the company’s advertising model became without the information that marketers often laboriously collect. Although it seemed odd at first that the company could profitably sell advertising directed toward an individual with no other known attributes other than the string of characters he or she entered into a search window, the success of their early gambit is now well known.
On the confounding side, an article that appeared in today’s Washington Post (Multiracial Pupils to be Counted in a New Way) brought other issues that fundamentally impact customer data. Traditionally, students were categorized in one of five racial or ethnic groups. Now respondents to demographic questionnaires will be able to check “all categories that apply.” As the article states, the change “reflects the evolution of a country now led by a president born to a white Kansan mother and a black Kenyan father.”
Further, in the past, I’ve taken it for granted that some data such as gender was relatively easy to collect. Not any more . . . recently I read about one survey that had a statistically significant number of people check both “male” and “female,” to account for the fact that those respondents were transgender. I’m sure there’s more than one brand manager at a major CPG company that is well aware of this issue!
As the school survey points out, societal changes will continue to present challenges for how people are tracked and how the data will be used for improving a wide range of public programs along with commercial products and services. The categories that the counties used dated to the 1960’s and were updated for use beginning in 2010–a span of fifty years! What changes are in the offing fifty years from now?
Hi Andy
Great comment.
Google is a company that has three of the four types of data. They have transaction data because searches are the transactions. They have online purchase context data becuase they know all about how we search and what keywords we use. And they have purchase influencer data because they see which sites we look at as part of the search process and hopefully, before clicking through to an adwords powered seller of whatever we are seeking. The only thing missing is customer needs data, although they could probably construct a family of meta jobs by looking at search keywords in new ways. A scary thought about a company that has been eager on occasions to let others do evil with our, oops, I mean their data.
I purposefully missed out demographic data in the post. I find it to be of limited use for customer management, particularly the more data-driven kind. It may be useful if I know absolutely nothing else about you, e.g. if you are not yet a customer and have no transaction history, but not very useful otherwise.
Even as society changes, and people have many ethnicities (the mind boggles) and many genders (as a former genetecist I recognise either XY, XX and many multiploid version of the two like XYY) and many other descriptors, it is their behaviour in terms of the balanced scorecard of customer data that I will be looking at for guidance. As someone should have said, and probably did at some point, “I am not a man, I am a number in a Google database”. Just call me Number Six from now on.
If you think Google is frightening, just look at the Sixth Sense project at MIT, courtesy of TED:
http://www.ted.com/talks/pattie_maes_demos_the_sixth_sense.html
This is the most amazing stuff. Really. Scary.
Graham Hill
Customer-driven Innovator
Follow me on Twitter
Interested in Customer Driven Innovation? Join the Customer Driven Innovation groups on LinkedIn or Facebook to learn more.