Best practice in implementing web analytics in companies

Miles Bennett, CEO of web analytics consultancy Numberminds is a web analytics specialist who helps companies improve their reporting and analysis capabilities so they can get better results from their web marketing. They has helped many of the UK and European retail financial services sites set up and get value from their web analytics solutions. These brands include Unilever, Barclaycard, Barclays Bank, Mint Credit Card, Royal Bank of Scotland, NatWest, Abbey.com, Lombard Direct, Ulster Bank and Coutts.

I've known Miles for a few years now and he is a regular speaker and acquaintance from the London Marketing Optimization Summits (www.emetrics.org and www.searchenginestrategies.com).  Since he sees many of the barriers and problems companies face in implementing web analytics he's in a great position to advise on how to avoid these problems by getting web analytics implementations right first time.

Q1. Looking at the big picture of digital marketing reporting and performance improvement, what would you say are the most common problems that are most detrimental to a business that companies should guard against?

It surprises me that the more companies I get involved with, the problems remain the same; politics (aka empire building), IT delays, Design Agency involvement and a general lack of departments working together.

Above all there is often a surprising lack of a cohesive reporting strategy and website objectives even for corporate communications sites.  Most reports aren't considered to be living documents that need to be constantly refined.  I'm sure I'm not alone in the knowledge that most emailed reports are consigned to the deleted bin without even being opened, typically because the recipients don't understand the report.

People don't realise that they need to take their websites seriously it isn't just a case of throwing a few pages out into the ether and then washing their hands and saying "Ah well that will do!".  It WON'T!  Measure it!

Companies invest millions (in some cases) in redesigns, content management systems, back end processing, usability and sadly web analytics and reporting is often the last thing to be considered.

Yet, there some simple questions to ask to help web analytics contribute:

1.      What is the purpose of your site?

2.      What are your measures of success?  Break these down into critical and soft factors.

3.      If you have an existing site - what is wrong? What can be salvaged? Perform competitor analysis.

4.      If it's a new site, again competitor analysis.  Determine competitor web marketing strategies.

Only when you have answered these questions should the selection and configuration of analytics tools be addressed.

Q2. For me, developing custom metrics, custom dashboards and emailed reports that are actionable to improve business results are crucial. Could you give some tips on the general approaches you use to identify appropriate KPIs and how they should be structured?

I totally agree but, still it comes down to how advanced our clients are - giving them the entire sweetie-store isn't always the best option.  Some simply don't understand that visits, page views and visitors are merely numbers unless there is more actionable insight.  First, segmenation of visitors is essential.  Next, determine which metrics can be used to add value to your report.

Remember that employees,  especially senior stakeholders aren't going to log into WebTrends or Google.  They need information that they can parse in seconds.  Give them graphs, gauges, conditional formatting.  This is especially true of marketing people who often manage multiple campaigns and need to know what to do QUICKLY to make best use of their budgets.

To review existing reports I find there are some basic reports which are critical for many different types of business which  then work with the management team to define the best KPIs and metrics from that. In all reports the main focus is typically:

1.      Where are the visitors coming from?

2.      Where are they entering the site?

3.      How long are they staying?

4.      Where are the exit points?

5.      How do conversion rates vary?

6.      How do you define marketing ROI?

7.      Monetisation - what is effective in generating revenue and profit.

Q3. The latest Econsultancy research shows that 77% of surveyed companies use Google Analytics, but with many (57%) using both Google Analytics and a paid service. How should companies decide whether a paid solution is worthwhile?

Google Analytics offers not just an Analytics solution but also has an ad-serving tool and an A/B and multi-variant tool.  It is an excellent grounding for anyone wishing to get into the industry.   However, when you are looking at brands like banks and online commerce providers (e.g. www.play.com) then the question of data security is raised.  The paid for solutions like WebTrends and Omniture have gone through this "pen testing" time and time again and clients trust their with their customer data (albeit matched offline with their CRM systems, but still personally identifiable information). Google's terms of service prevent information being held about individual customers.

Online and offline data integration is essential for a holistic marketing reporting tool. This can be more challenging in GA.  You'll also want to integrate your CRM information and profile your customers better.  For example are all your customers that come through social media - 18-25 year old, single living at home?

The other thing that needs to be mentioned is that web analytics (whatever tool you are using) is not a part-time job.  Someone cannot be given this as an additional soft responsibility.  Paid for solutions need people investment, they need someone who has a passion for data.

Q4. It's one thing to have an analytics system configured correctly, it's quite another to get value from it through staff using it sufficiently and appropriately. Which approaches have you seen which work help encourage productive usage of web analytics system.

My clients have primarily taken the web analytics data and integrated it within existing reports (sometimes these reports are 10 megabytes in size).  I've read about a woman who accosted colleagues in the corridor and by the water-cooler and challenged them if with questions like "who is our biggest referrer" or "what is the search term which is driving most organic traffic".  If they got the answer right then they won a chocolate bar.  The last I heard was that she was now getting USB lava lamps for winners from a gadget website.

I wouldn't advise that this would work in all situations! It could be classed as a gimmick or a myth and I can see people running in the opposite direction and avoiding the water cooler altogether!

Nothing can beat good dashboard design, good commentary and useable recommendations.  Be careful with the terminology you use and make sure you have made your point within one or two sentences.  No one really likes reading reports because historically they have always been the bearer of bad news and web analytics is even harder to 'get'.  Bear that in mind when you're writing your next dashboard.

Q5. When more than one web analytics system is used for reporting, then there will often be substantial differences in visitor volumes and page views. How should web specialist explain the significance of these differences to their less savvy colleagues?

It's simple - give them the truth. Each web analytics tool has different collection methodologies.  The terminology is the same and I and my colleagues have worked hard at the Web Analytics Association (www.webanalyticsassociation.org) to ensure that the terminology is shared across platforms (I also did similar work with the Direct Marketing Association back in the late 90s in relation to telemarketing).

If you're running two tools in tandem then there has to be a reason for that.  Are you migrating from one tool to another?  Are you trying to prove to the budget holders that migrating from a free tool to a paid solution provides more insight?  I hope it isn't just because the graphs in Google Analytics provide a better image grab that those of WebTrends.

If you are insistent on running two tools constantly then understand the limitations and the differences and post these percentage differences on the report (along with the source).  So XITI is 25% less than Google Analytics (and explain that this is because of "x") otherwise the larger number will be believed, in most circumstances.

Q5. Looking to the future, what excites you most about where we are now with web analytics against where we could be

I am just amazed at the growth of interest in web analytics but, in the same breath disappointed that more companies aren't embracing the technology and the insight.

I've also seen the market move from focusing on the site content to what other people say about you, the rise of social media is one of our most interesting challenges.

Primarily, I love being involved with some of the greatest web analysts on the planet and making web analytics a must have discipline and not an afterthought!