13 Aug 2012: The overwhelming majority of marketers regularly accept that building a single customer view strategy is utterly vital in today’s complex, multichannel world – so why is it that so few of them are actually doing anything about it, asks James Lawson?

A 2012 Experian survey identified a host of reasons to build a Single Customer View (SCV) database, with cost cutting (90%) and increased revenue per customer (80%) coming top. Yet more than half the companies interviewed (57%) had taken no action beyond the planning stage and just a quarter (24%) had a strategy in place. So what’s the problem?

Plan your build

Achieving and maintaining a single, integrated view of customers and products is rarely a “big bang” project. Instead, it is more of a continuing and ever-evolving journey, one where the ambitions and applications for the final database will be different for every business.

The first and most important question in an SCV project is, why are you doing this? Good project planning demands a step-by-step rundown of the strategy behind the project and the difference its success will make to the business. GB Group has a specific methodology it works through for all SCV projects.

“They need to define what they want to achieve and use that to drive the design phase of the project,” says Darren Cudlip, the company’s Sales Director. “What is the contact strategy they want to put in place? What are the key questions they need answered? If you start out by defining the technical solution first, it can easily lead to something that’s too complex and unachievable.”

Costing any business case to demonstrate ROI and gain funding is unavoidable, so it’s essential to set metrics that accurately measure current and target goals. Regular monitoring will keep the project on track and show progress or otherwise. Benchmarking metrics against industry standards will also aid understanding of how much there still is to do and how well you are doing compared to similar successful projects elsewhere

“You need to lay out what defines success up front,” says David Barker, Head of CDI Europe at Acxiom. “You must understand the success criteria for the project and how you are going to measure them.”

Going through all the data sources and deciding which ones are the most important is a vital early step. To work out those priorities, go back to the goals of the project. Once a project starts rolling, it’s easy to start to try and fix everything. As well as an excessively complex implementation, this can mean that the SCV fails to deal with the problem it was initially set up to tackle.

“We come across hugely ambitious projects fairly often,” says Joel Curry, Managing Director of Experian QAS. “The project teams need to understand and focus on the primary use case. If your ROI justification is to increase the number of returning website visitors, then you need to put those visitors first and then build the rest of the SCV.”

With a hierarchy of feeds that reflects their importance to the project, it’s time to find out what sort of state the source data is in – and be prepared to adjust expectations in the light of the findings. Companies need to audit source systems with as much rigour as possible. Ideally, this would be done by a trusted, independent expert both for reasons of accuracy and also in case there are any internal politics involved in ownership.

A clear estimate of the complexity, time and cost involved in cleansing each feed is an essential part of any audit. The old Pareto 80:20 rule has application here: far better to put one or two very dirty and non-core source systems to the side for now and concentrate on fixing and integrating the key feeds.

“There are massively varying quality standards in different systems,” says Curry. “There’s no point in attaching the website feed until you fix the quality of data coming from it. It’s far better to build in a modular fashion and get the SCV going with the other channels first.”

Gaining full approval for access to feeds is also critical, as is complete commitment from project participants, including a realistic estimate of the amount of time they are able to devote to the project from their own schedule. They must have the authority to make quick decisions rather than having to go back to the stakeholders every time.

Deciding on the required data items and where they will come from is another vital part of initial planning. Are you going to incorporate any data from partners or other third party owners? Is some may spark the realisation that some vital data is missing. In the case of the key variables used to link files together, any gaps here must be dealt with. That means either data collection or buying in those fields from external providers.

“Email and postal address are the key pieces of data that enable you to link individuals together across channels,” says Katharine Hulls, VP Marketing at Celebrus. “By encouraging and even incentivising registration, then sign in or log on across touchpoints, you can then log and link the multiple devices that individuals use.”

Definitions are another classic area for confusion. Even something as apparently simple as, “what is a customer?” will elicit very different responses from each department. Some form of corporate Master Data Management (if in place) will help but thrashing out what constitutes an enquirer, prospect, lapser or contract start date can make for revealing conversations.

“As you get into it, you find there is more than one answer,” says Barker. “Mapping product hierarchies can also become very complex very quickly and there are often historical product lines that also have to go into the SCV. You need to discover as much of this as possible at the outset.”

Flexibility is another critical element of a successful SCV project. Even the best-planned project will be overtaken by events as new feeds appear or as departments make extensive change requests. That means having a database schema that can adapt quickly to new feeds and a project team that can do the same.

“A decade ago, a database might have five sources and add a new one annually,” says Barker. “Now there might be 100 to 150 feeds with between 10 and 20 new ones each year.”

Barker advises, “having the conversation up front”, with any contractor. “You need to ask what it will cost and if the system can handle that – is it flexible enough?” he says.

New era, new challenges

Much of the advice thus far could apply to any SCV project in the last couple of decades. But the combination of real-time updates, ballooning data volumes and tricky online data represent something of a more recent challenge. However, the same basics apply: work out what you want to do and why. If the benefits outweigh cost and risk, only then think about the details of the technical solution.

“One of our clients was able to sell higher value products in the call centre based on their understanding of online behaviour,” says Hulls. “They could make these follow-up calls because they were able to link that data together.”

In the case of offline data, the decision on which items should be brought into the SCV can be complex. Web analytics generates vast pools of data, most of it anonymous, and aggregating it in some way is essential to make it actionable.

“Your call centre and shops may be only open for eight hours a day, but your web and mobile site are bashing away 24/7 and this means your SCV needs to be more responsive,” says Mark Wilding, Director at Cognesia. “Real-time may be over-stating it, but certainly near-time may be more appropriate.”

Serving ads or making offers in real time based on analysis of historical behaviour across channels is sexy stuff, but this sort of ambition can end up sinking a project. Again, a true cost/benefit analysis should reveal the scope of any work and if it’s justifiable. Talking in depth to marketing may reveal that offline analysis of web behaviour combined with other channel information would satisfy some or all of their needs.

“We’ve seen customers achieve a nice balance by treating some of their web data as ‘ad-hoc’ or ‘campaign based’ and therefore not having the big step of permanently applying all of it to customer’s records,” says Wilding, who advises picking a campaign that will “generate real value” and using that as a proof of concept.

“Knowing the campaign objective helps identify the web data that will be required,” he says, “and whether it should be stored permanently or in some other way. It also makes it easier to identify what data will need to be shared with the communication tool, be it email or otherwise.”

An operational database might hold any data required to drive real-time processes such as web offers, email triggering or ad serving, while a less-frequently-updated marketing database might be used for analytics and direct mail selection. Campaign management software can sit on top to manage the ruleset and decisioning required across channels as well as governing the various refreshes and sychronisation between systems at the desired intervals. Alternatively, the refreshes might be controlled by another data management tool such as Informatica.

“It’s unrealistic to expect a traditional SCV in real time,” says Cudlip, referring to the set of cleanse, dedupe, PAFing and other processing jobs in a typical marketing database refresh. “You need a different architecture to do real time, typically an offline batch-updated element that will talk to an operational real-time database.”

Triggered activity

Event-driven communications such as triggering an email offer from an abandoned basket is a good example. Does that need millisecond response – “real real time” as Acxiom’s Barker puts it – or is an hourly refresh perfectly fine? As highly available, operational systems with backup and failover don’t come cheap, these decisions have a big impact on overall cost.

To decide means mapping out the various phases of the possible customer journeys in each channel, their different timings and the possible triggered messaging or other actions at each point. Once again, a solid figure for projected extra income based on estimated conversion levels or order values is needed in order to rank each activity and the data it requires by their value to the business.”

“You might know you can recover 4% of abandoned spend if you email within 4 hours,” explains Cudlip. “Marketers want to bring in individual-level data to serve ads in real time, and this can lead to an explosion of data. You need to be very specific about the elements you need.”

Whether on- or offline, retail or batch, experienced consultants can make a big difference at all stages of a project, and are particularly effective at the start. For example, engaging an experienced MSP to build and test a static extract helps more accurately predict how cross-sell or retention work might raise income. But their knowledge of the classic pitfalls that kill SCV projects early and how to avoid them is even more attractive.

For example, QAS’s Curry comes across companies that take on SCV projects with purely internal resources, fully prepared to build everything from scratch. “There’s lots of craft that goes into building an SCV and you really do need experience,” he says. “They may have a very confident IT manager, but there is a whole raft of common issues they won’t know about. A lot of our current projects started in-house two years ago, but didn’t work out.”

Build your plan

Even in a company that is familiar with its data and has good technical resource, SCV projects can prove tricky. Political problems can derail the schedule while a takeover or merger will knock any SCV project out of kilter. Staying flexible and remembering that an SCV is a journey rather than a destination is the only way to handle the inevitable changes.

“Where we see client errors, it’s usually in the design and planning rather than the technical area,” says Barker.

“You need to understand the detail of what marketing really wants and have people who understand the business committed to the project as a full-time job.”