How to get your web analytics implementation right

On the surface it would seem that indifference rules the roost and analytics is just another tick box on the requirements list.

However, on closer inspection, the criticism of apathy can often be harsh. More often than not, data obscurity lies in a lack of education amongst stakeholders uncertain as to what analytics really means and what it should do for them.

In my humble opinion (though I do love to voice it) the responsibility for getting the analytics implementation right lies with the person who runs business intelligence in-house and the agency selling the service.

Yes, the Client has to take ownership of the outcome but if you are selling a service, you should have your Client’s best interests at heart, not just an eye on the revenue stream.

Below are my recommended steps for effectively planning your web analytics implementation... 

Understanding what analytics is

A web analytics tool is not a sales reporting tool. Why? It collects data differently. If you want financial sales reports, take these from the web database. Analytics is about trend analysis. The revenue figures from something like Google Analytics or Omniture Site Catalyst will not match a sales report 100%. It is essential that expectations are clearly set from the start so you can understand what the role of analytics is within your business.

An analytics tools represents a sub-set of your data and is effective in monitoring trends over time. This means you can evaluate customer behaviour patterns and the impact of site changes on key metrics.

This is a simplified explanation. If you want to learn more about the difference between analytics and sales reporting tools in relation to data capture methods, I recommend contacting Michael Feiner of AEP Convert, a seasoned analytics specialist who can be found on Twitter.

Don’t sign-up with anyone who doesn’t audit your commercial requirements

This is not negotiable. It is impossible to implement an analytics solution that will cover your unique business needs without auditing requirements across stakeholder groups.

Many agencies will have an out-the-box solution but how do you know this satisfies your needs? On the surface the report suite may look comprehensive but believe me, the devil is in the detail and no two companies have identical requirements.

For example, do you attribute revenue by the last click model when evaluating marketing efficiency? Or does your Finance team demand a more granular analysis whereby you can stack campaigns to understand the contribution that one campaign may make to long tail conversion?

Consider a PPC campaign that using the last click model is returning a low ROI of 35% and the decision is to divert budget into the more profitable email program.

On closer inspection, your analytics tool may tell you that this PPC campaign has actually influenced purchasing in other channels and when you attribute revenue share, the ROI suddenly leaps to 350%. Now does your commercial decision change?

If this level of detail is not mapped out in a scoping exercise, you will be left with a reporting suite that gives only partial information and clouds judgement.

Identifying commercial goals

Your analytics solution must be tailored to your commercial goals. How does your e-commerce channel support commercial goals and how to you measure this contribution? If you have multi-channel aspirations, you need to factor this into planning. For example, what analysis will you require of the impact of your website on driving store visits and sales?

A recent Google Analytics audit I did for a client revealed gaps in implementation for tracking on-page events such as the send-to-friend feature on product details pages. Elements like this were missed in the initial implementation because the agency did not map requirements into a specification document.

Don’t forget external websites that are linked to your e-commerce platform, such as white label supplier sites. Reporting from a wide supplier base is usually inconsistent, so mapping out a tagging structure for your partners and then giving them an implementation guide is important. Some may not want to support this, but that is a consideration for your commercial team.

Understanding relevance

There is no point generating a complex web of data and reports that you will not have the time to analyse. Data is only useful if it is relevant and you can take action. Only collect information that you will use and that can help your business make intelligent commercial decisions.

At the same time make sure the analytics tools you invest in are scalable for the future. If you know that in 12 months time you need to integrate data from a telephony system, make sure this can be plugged in without onerous cost.

Understanding relevance involves knowing what is most important to you and how much resource is at your disposal. This has to relate back to the commercial goals identified. Start with top level priorities such as optimising conversion and improving page engagement to reduce bounce rates. Don’t try to do everything at once otherwise you will drown in data and struggle to make constructive choices.

Mapping out inter-dependencies with other systems

Who else in the business needs to exchange data with the analytics software? This could be a central CRM team managing the email program that wants an API to feed segmentation data into the CRM database to drive targeted campaigns. Or it could be your Finance team wanting to slice and dice data to support revenue attribution modelling.

By mapping out these requirements you will get a more accurate idea of the cost of analytics and this will help the business decide how far to go with its implementation. Everything has a cost, either in upfront development or ongoing license fees, so this mapping exercise is important commercially to avoid blowing your budget and invoking the wrath of Finance.

Getting a comprehensive implementation specification

Before anything is done on your website you need a comprehensive specification that details what the analytics solution will deliver. This should cover the reports that will be supported and the implementation requirements for developers to follow.

The devil is in the detail. Relate this specification back to your commercial goals and make sure that the scope covers your requirements. Don’t sign-off anything until you are confident you have covered all bases. If you don’t give this stage sufficient attention you will pay the price, often quite literally, post implementation.

Rigorous testing

A detailed functional spec does not guarantee a flawless implementation. e-commerce platforms are complex beasts, so you need to test the analytics implementation thoroughly to ensure that data is populating the reports in the right way and the page tagging is accurate.

To achieve this you need a technical appreciation of how analytics tagging works – for Google Analytics this is relatively easy as much of the code is consistent but for solutions like Omniture and Coremetrics you might want to consider validation from a solution specialist (who does not work for the agency delivering the implementation if you use third party suppliers).

Experience tells me that it will take a few iterations before the reporting suite is populating correctly and this additional time and cost is well worth the effort to give you the reassurance of data integrity.

Defining the support framework

What happens if something goes wrong? What level of support are you commercially entitled to? It is important that you know who will help with technical problems and who can support you with the commercial application of analytics from a business-user perspective.

Key requirements are to have a tight SLA with your analytics provider with clear support channels including live technical support and an Account Manager to provide a real voice at the end of a phone line.

If you are using an agency to implement the analytics solution, then you need to clarify their level of responsibility post-implementation. Do they support the analytics platform? Or do they simply cover issues with the page tagging from a technical perspective?

Summary

Don’t assume that just because somebody has implemented an analytics solution before it means it is the right implementation for your business. Use the
following checklist to make sure you plan your analytics requirements in detail:

   1. Make sure you understand what the role of web analytics is.
   2. Review commercial requirements across stakeholder groups.
   3. Use an analytics specialist if you don’t have the in-house knowledge to understand what you want to achieve.
   4. Produce a comprehensive technical implementation specification.
   5. Agree a clear SLA with any third party suppliers.
   6. Make sure you have a structured support framework to help with the day-to-day usage of your analytics tools.
   7. If it doesn’t make sense, ask people for advice.