Building analytics into your business processes

 This article, written by Neil Mason, was originally published on Clickz.com and is republished here with permission.

I’m increasingly convinced that the issues that most businesses face around the successful deployment of analytics in their business are not to do with their technologies but to do with their businesses processes. That view was reinforced this week when I was running a workshop with a group of students studying on a Masters Programme in Internet Retailing.

As part of the session I asked the class what they thought some of the key ingredients were in executing successful internet optimisation. They started off by throwing out items like “good data”, and “technology”, and then moved on with things like “flexibility”, “willingness to fail”, “good hypotheses”, “risk culture” and “strong business model”. I then asked the class how many of them worked in organisations that had those characteristics. No hands went up.

Often organisations are happy to spend money on new campaigns or large scale product development on their site without thinking or being explicitly clear about how they are going to measure the effectiveness of the campaign or the new piece of functionality. So how do they know that they have done a good job? The measurement piece also needs to be built into the campaign or product development process as well. Here is a simple framework to use as part of that process.

• State the objectives
• Define the success indicators or KPIs
• Perform a gap analysis
• Create the measurement roadmap

State the objectives

Before you start, be very clear about what you are trying to achieve. State your objectives and make sure that they are not what I call “marshmallow objectives”. Marshmallow objectives are ones that are soft and squidgy. They are not firm and they don’t hold up to scrutiny. These objectives are difficult to measure. People often talk about making objectives “SMART Objectives”. SMART objectives are Specific, Measurable, Achievable, Realistic and Time bound. You should try and make your objectives as SMART as possible. If you come up with objectives like “to improve the user experience”, force yourself to be Smarter and ask yourself questions like: “why?”, “which users?”, “in what way?”, “by when?”

Define the success indicators or KPIs

Next define the measures of success or the Key Performance Indicators. These should relate directly back to the objectives. If the objectives are SMART then the definition of the success metrics should be relatively straight forward. Constantly ask yourself “What does good look like?”. If your are achieving your objectives, what will be happening in the business or on your site? Sometimes it can be difficult to measure objectives directly with significant cost (see below) then you may need to come up with other success metrics that are indirect measures of success.

Perform a gap analysis

Once you know what it is you want to measure, perform and analysis of what you can measure now versus what it is that you ideally want to measure. Where are the gaps? Do you have the right measurements systems in place already and are they configured the right way? For example, f you want to improve likelihood of someone booking a holiday once they have done the research on the site, you might set up a success metrics that is something about visitors propensity to book at a later date. You may need a survey in place to be able to measure one. So do you need to set up a survey from scratch or do you just need to ask a specific question on an existing Voice of the Customer programme?

Create the measurement roadmap

The final stage is to create the measurement roadmap. This is effectively the plan of how you are going to measure the things that you need to measure, or as I like to say, “Count the things that count”. A key part of this is prioritising the work that needs to be done, being clear which gaps you are going to plug in your measurement systems and then how that work is going to get done. This could range from just ensuring that there is a specific custom report developed through to the implementation of a new piece of software or the adoption of a new service.

The process I have outlined can take days to complete or it might just take a matter of minutes. It might be a very strategic piece of work looking at the business overall or it might be quite tactical, for example running an A/B test on a landing page. Whichever it is, it’s a way of putting measurement and analytics at the heart of your business processes. So anytime you’re planning to do something, ask yourself (or your colleagues): What does good look like? How will we know we’ve done a good job?