From tracking to transacting

Web analytics tell you a lot, but not everything you need to know. For that you have to interpret the reports or integrate their data into broader BI packages. David Reed finds out why this is so hard to do.

When does a report become business intelligence? To some data practitioners, there seems little difference. If you have got the numbers on how some aspect of your organisation has been performing, that is all you need to decide whether to make changes.

Others disagree. They see reports as backwards looking, static and limited. True business intelligence needs to have a context, interpretation and recommendations. Without these, they argue that numbers are either too reassuring (if they are positive) or do not tell you what needs changing (if they are negative).

Nowhere is this more evident than in the realm of web analytics. Anybody with an online presence is likely to be deluged with numbers about their site, from gross volumes of clicks to favourite areas. So is that all that is needed to track the impact of a site on your business, or do you need to go further to make sure visitors are transacting?

Christian Howse, digital solutions strategist at WebTrends, is clear about the difference. “Web analytics is a misnomer – really they are web reporting tools. They point out what is happening on your site, how many visitors, percentage who transact, but not what should happen based on that behaviour,” he says.

One of the problems he identifies is that web analytics is very siloed in most organisations. Not only do web analysts tend just to be producing fixed reports, but those reports are only distributed within the division of the business, rather than across the whole enterprise. “It is a 270 degree view, not a 360 one,” says Howse.

For him, the problem is with the process, rather than the technology or data. “An organisation may have web analytics in place, but when they launch a new microsite, for example, they start from scratch and put Google Analytics on it. That is measuring what you can, not what you want to,” he says.

While the data generated in this way is useful to the specific marketing team which is managing the site, its greater value is limited. Often it is fear of waiting a long time for IT to develop the right solution that drives people towards an immediate one, whatever its faults.

“You tend not to hear the term ‘process’ very often in the new media world,” says Howse. “For an organisation to be competitive, every decision has got to be based on data. You can only do that if you know where your data is and how to link it to the big picture.”

Problems with integrating online and offline data are endemic, with 69 per cent of marketing directors surveyed by WebTrends saying they struggled with this. Howse has spent the year working with a major retail bank on a data integration project of this sort, applying Teradata and KXEN tools to all channels of data. “That opened my eyes. Previously, we had just been doing standard reports without seeing how they fit into the whole organisation,” he says.

Agencies often run into this problem when they are called in by the marketing department to help sort out online activity. There may be a view that a website is not delivering the business results it should. Demonstrating why this is the case and identifying what needs to be done can be difficult.

At Archibald Ingall Stretton, digital planner Kristin Berg says: “I see reports out of web analytics tools produced by clients and nobody knows what to do with them.” These reports frequently have a large email circulation list but are infrequently read – an unread message is just spam, regardless of its content.

“I take the numbers out of those reports and into a spreadsheet so they can be integrated with other numbers and cross-tabbed and analysed. I can also control how they are presented,” she says.

Interpretation is a critical dimension of the elevation of web analytics into business intelligence. This may mean highlighting key numbers, showing trends or things for the client to watch out for. Or it could mean the way the information is displayed is critical. Berg often produces summary dashboards with green and red indicators for “at a glance” viewing.

“Marketing and strategy departments often don’t have the data on which to make decisions about what to do next. They do get reports, but are not sure how to use them. Once you don’t know what something is, it becomes junk,” she says.

Since many web analytics tools have pre-determined formats, it may be necessary to introduce this human layer of analysis. At a simple level, this may only involve wrapping an email message around the report which debriefs it in summary. “I write bullet points – five to ten for a website report,” says Berg.

More sophisticated processes will extract the data for integration with other data sources. An important aspect of working with a skilled partner is the extent to which the organisation may come to rely on them. “Often we become their long-term resource,” she notes.

Skills shortages in the analytics space are all too common, with web analytics a particularly hard-pressed area. One consequence of this is to increase the use of standard reporting tools in-house. This reduces the requirement for humanr esources, but also tends to lock the organisation into a cycle of static reporting.

When a website is performing well, this might not matter since the business will only see the positive data on conversion and revenue, not the potential which is being missed. If performance starts to dip, however, the search tends to begin for better insight.

“We are currently seeing a lot of interest among clients. The bigger the organisation, the bigger the problem they have in this area,” says Grant Keller, managing director of digital marketing consultancy Acceleration Europe. “They tend to have a gaping chasm between what’s technically capable and the business reports they need.”

His company works with clients to establish what their business KPIs are and what therefore needs to be done in order to provide those metrics. “Our clients feel they don’t have the right intelligence to make business decisions and meet their targets.”

The fault often lies with the way that IT departments get asked to implement web analytics tools. “They don’t have an understanding before that implementation what the business wants to learn from it,” says Keller.

Changing this involves a number of things – sometimes the tool itself needs to be replaced, although often it has just been set up wrong and can be improved. “To find that out, you need to start with your end in mind. You really need to get under the skn of the business requirement, what they are trying to achieve and where they need to measure performance,” he says.

This knowledge is usually in the minds of the business users who are getting static reports. But they may not have been able to convert it into a clear enough brief for IT. It can also be a challenge to convert web-based KPIs into business-driving indicators. “The tools are there, but they are not implemented correctly,” argues Keller.

Another challenge to full integration is the different levels of knowledge, experience and expectations around each type of tool. In this respect, there tend to be two camps with very different ideas about what a solution should look like.

“People that have worked in traditional channels want to drill down, get extra analysis and drive out insight relevant to the business process,” says Gareth Mitchel Jones, director of analysis at Experian Integrated Marketing. “The people working on the website are interested in the number of clicks and pathways. That is useful for its purpose, but it doesn’t give you the understand of what is driving that customer.”

His company has been able to gain significant traction by helping clients to use the outputs from the second type of tool to achieve the outcomes desired from the first type. A major challenge, however, is that the availability of free web analytics tools that generate high volumes of data often clouds the perception of just how valuable those reports really are.

“When you do traditional analysis of online data, it is costly. That is because of the human involvement. It is not just about producing better reports, but about having somebody interpreting them,” he points out.

Developments on the vendor side are slowly making it easier to move reports from their tactical status into that of strategic resource. Either through the consolidation of web analytics into marketing automation solutions or their integration with BI tools, the technology is becoming less siloed, even if the process for its use remains so.

“The problem is a sore point in the industry,” says Akin Arikan, director of product strategy at Unica. “Getting from traditional reporting to optimisation has been difficult. Things that play a role in that are reporting that doesn’t do anything. Before you can improve that, you need to get from reporting to analysis. I’m not sure if most people recognise the difference.”

As an example of this, he points to a typical web analytics report showing that 10 per cent of traffic on a site converted to sale. “Is that good or bad? What can you do based on that information? Nothing unless you can get into greater granularity of data,” he says.

Drilling further down may reveal that conversions from visits driven through paid-search are running at 2 per cent while natural search is converting at 15 per cent. That immediately provides some options about what can be done to improve performance, especially if further data mining shows which paid search terms are the best performers of all.

“From the tools side, the tool needs to make it easy to ask new questions. That is not always the case because implementation has been to difficult for business users,” says Arikan. Communicating best practice is one way to build pressure for change, either of the tool or the process. “But when people are not listening, how can you help them?” he wonders.


    Leave a comment