After five years of investment, data functions are well established. Now they have to prove that they are generating a positive return for the money they have received. David Reed looks at some of the ways in which organisations can prove the value created.
It’s payback time for data. After five good years during which budgets have flowed towards data management and analytics, the time has come to start showing what the business is getting for its investment. As James Morgan, head of information strategy and management in the business intelligence unit at Telefónica, says: “A lot of the work we do is putting targets and value behind the things we are trying to achieve and then being held up to those targets.”
He is not alone in facing that pressure this year. Data is attracting significant interest at board level, whether through the continuing distribution of analytics down to business user level or as a result of the excitement surrounding Big Data. Monetising data has been backing up that interest, but now comes the reality of that focus - being exposed to the demand for clear reporting and returns that other functions have long been held to.
Morgan describes the importance of getting a “line of sight” between what the data or analytics operation is doing and how it is driving the customer experience, product sales, retail performance and the rest. At the same time, data has to hold on to the vision which first made it appealling. “How do you balance the right long-term strategy and ensure it is not too distracted by doing the more immediate initiatives that are being demanded?” he asks.
Good question - and one the industry will do well to find an answer to in 2013. Data management may be essential to keeping core business processes running, like customer relationship management and marketing. They can even justify investing in analytics, although ad-hoc resources will often suffice. Building that into a fully-blown, standalone operation is a much greater commitment, however.
The returns from doing this can be considerable. Aberdeen Research surveyed 145 organisations last year into their deployment of business intelligence tools. In its report, “Coffee and BI are for Closers: Mixing Analytics with Customer-Facing Activity”, it found that best-in-class operations (the top 20% of those it researched) had achieved an average 29% increase in organic revenue over the past 12 months. For the half of firms in the middle, that growth was only 10%, while laggards saw a 4% decline.
Even better, the top performers raised their operating profit by 28% in the same period, against a 9% gain in the middle and a 16% fall at the bottom. Across sales, marketing, customer service and field service, it was also discovered that leading on analytics shortened the sales cycle by 10%, meaning that leads converted into cash more quickly.
Those are real-world gains that prove data can be monetised in a variety of ways. Some of those gains are the result of finding new opportunities, such as the SmartStep mobile marketing service which O2 is now offering retailers to send messages to potential customers as they pass by a store.
EE has developed a similar proposition which leverages its own investment into data and analytics as a new media sales opportunity. It is based on analysis of its call data records and cell locations which it can then profile against specific retail geographies. “It makes targeting mobile advertising easier if you can get this type of granular information,” as Ben Lethbridge told last year’s DataIQ Future conference.
In one programme carried out for the Westfield mall, store visits were compared against footfall and mobile browsing activity. This discovered that 31% of visitors to the shopping centre were visiting Amazon’s website via their mobile and a further 7% were visiting Groupon. Those are indicators of potential sales leakage and discount hunting which retail marketers can develop plans to deal with.
Without its investment into data and analytics, EE would not have been able to find those insights or to have monetised them through third party advertisers. Becoming an effective media owner is one dimension in how large organisations are leveraging data in new and profitable ways.
It also represents something of a trend happening within the data industry - finding new revenue-driving uses for data beyond their primary purpose (subject to appropriate permissions). A customer record can be targeted with first-party cross and up-sell offers, but it may also be available for third-party, non-competitive marketing.
That is what The Data Partnership is offering through its 2nd Use data. It runs regular telephone and online surveys asking questions specifically commissioned by brands. They have a 30-day exclusivity over that data, after which it can be sold as a lead to others. What makes that attractive is that consumers who responded to the survey are still likely to be active, but the delay makes the record significantly cheaper than if users captured it for themselves.
“It has been proven to work if you use it the right way,” says John Pooley, director of The Data Partnership. “These records have not been bombarded. As 2nd Use, they are likely to have received one call from the company which commissioned the survey - and they may not even have been contacted at all.”
It is partly inefficiency by first users of the data which has spurred the decision to put the files onto the market. “A lot of companies don’t use it properly - they get a new feed of data, hit it once, then wait for the next batch,” he says. Pooley blames call centre managers who are often inexperienced with data or who are only running operations 9 to 5 - that means records will get dialled when nobody is at home. If there is no response, the data itself often gets blamed. Paradoxically, this is also what is making 2nd Use valuable for resale, since it has not been exhausted.
When considering how to demonstrate a return on investment from the database, permission-to-market is one place where many organisations start. It is a nexus for processes that capture data, such as websites (through registration forms), loyalty programmes and CRM, and those which rely on it, such as marketing.
“You can assess your database for whether you just have a postal address, an email address or both. That is about your ability to contact and market to somebody which is an indicator of its value. You can multiply that across the whole base and work out a financial argument,” says Lee Witherell, insight director at Merkle. “It is not financially accurate, but it is a soft measure.” A next step is to consider customer lifetime value and how data dimensions - from permissions and quality to depth of variables - help to realise it.
Witherell says that more advanced marketers are starting to consider how the use of data to personalise and target offers delivers uplift against what would have happened anyway. “You find that there are some people who will buy from you anyway, so you don’t need to invest in them. That frees you to spend more on people who might have a higher yield, for example. It is a view that is coming back into fashion,” he says.
Differential marketing depends on data to enable this kind of value segmentation. That data has become more available and insightful than it was 17 years ago when the concept was first proposed by Garth Hallberg in 1996. However, Witherell notes that many marketers struggle with the idea of leaving this segment untouched. “The difficulty they have when they look at it that way is, why not invest budget where you will get a ROI because it is easy?” he says.
Proving what keeps those buyers coming requires the more complex modelling of econometrics. This unpicks the contribution of different parts of the mix to the whole and can help to modify budget choices. Running savings made as a result back to the contribution made by the underlying data is not often done, however, even if the models could not have been built without it.
For Charles Ping, CEO of Fuel Data, understanding if data has been worth investing in starts with simpler measures. “The first thing in understanding data’s value is making sure you have got all the information necessary to differentiate one customer from another,” he says. That may sound simple, yet it can be critical - as in regulated industries which have to ensure they do not blend different customer accounts - and also difficult to achieve - most companies would not state their data accuracy beyond 98%.
“The whole point of realising the value of data is about saying the right thing to the right person and also who you do not want to say anything to. Propensity models are most powerful in dropping the bottom 20%, so you have to be certain you have got all the information you need to drive that decision,” he says.
Failure to integrate customer data is a commonplace problem and can therefore become a basis on which to argue for investment. When customer data integration has taken place, there needs to be sufficient detail to allow for meaningful differences in any segmentation that is applied. That can be harder to achieve than might be supposed - how many databases have simple variables on them, such as gender, which could be critical indicators for marketing?
That might help to argue the case for investment into upstream processes where data is first captured or entered. “Are you recording it in a way that allows for analytics to inform what you are doing?” asks Ping. If monetising data through licensing it to third parties is part of the business case, then making these processes more accurate is even more critical.
Ping notes that the idea of using data to make better deals is starting to penetrate more widely across business. “For example, if you are a generic news website, your news pages are vanilla, but your personal finance pages are chocolate chip. So if you can identify a visitor in the finance section and target them in news, you can still charge a premium based on their interest in money. That is the whole game,” he says.
Key business decisions might be made based on these considerations. On a paywalled site, 80% of revenue probably comes from subscription sales. So using data to drive a 5% increase in ad revenues is hardly worth the effort, as it only increases overall yield by 1%. “You are better off putting your money into subscription growth,” argues Ping.
In some respects, data is like insurance - you do not see a positive benefit until you really need it. So much of the modelling and forecasting carried out by the data and analytics function results in negative outcomes - choosing not to invest in certain areas of the business or customer segments.
Can those be counted as cost-savings and used to prove the value of data? Anecdotally, this clearly happens - marketing presentations are full of the efficiencies which have been achieved but raraely mention monetising data.
But as data continues to attract investment over the next few years, it will need to work a bit harder to convince. Ping says: “It doesn’t matter how you make the case, businesses are looking for payback in year one and on the basis of a return on capital employed.”