Today, technology gives you tools to learn more than you could ever use about how people interact with your content. But the key to optimization is not about technology and certainly not about trying to reconcile statistically insignificant discrepancies in reporting. Let’s first assume you performed extensive QA on your analytics instrumentation, and that you are comfortable that you are collecting accurate data.
Analytics tools already work really well. And much optimization today focuses on automating incremental changes, or context-relevant content delivery. We are talking about something more broad than what an optimization tool delivers. In this broader sense, the difference between success and failure is about process — more importantly, a repeatable process.
Here are the steps:
- Define Drivers (figure out your content goals)
- Build Metrics (decide what to measure, and with what tool[s])
- Plan Actions (decide what to do after reviewing metrics)
- Create Changes (execute the actions that were planned)
- Measure Success (measure again; see if goals were achieved)
Let’s talk about how each of these steps can work in a real world situation.
1. Define Drivers
This part of the process is essentially a strategy and a discovery exercise. It will call upon people who have both very high-level requirements as well as those who have more targeted, specific requirements of the site. But the strategy needs to grow out of business needs, not the capability of the measurement tools.
For instance, you may determine that your sales reps need to get more leads out of the site (or app).
You’ve just defined one of your key drivers.
2. Build Metrics
The level of complexity found in digital analytics tools continues to increase dramatically. However, they are foundationally designed to deliver “reports” based on site activity.
Make sure you work with a team of professionals who really know how to make these powerful tools collect and display data. An out-of-the-box instrumentation is fine for smaller companies, but the enterprise must go further. And must also know the limits of usefulness. Instead of trying to report on everything you can collect, go back and look at your drivers. Focus on getting the data right as it relates to that.
3. Plan Actions
This is the part where software can’t take you. Only a certain level of knowledge about both marketing goals and technology will get you through.
It’s the part where, even when there’s a richly detailed report lying around, nothing else happens. Unfortunately, reports alone are not going to somehow turn into a form of “advice.” Someone inside or outside the organization has to do this. These days an entire professional class has grown up around it, and they call themselves “analysts.” Another popular phrase describing this is “data storytelling.”
Humans need to take numbers and graphics and make it mean something to stakeholders who don’t spend their days and nights with analysis. They need to target problem areas (for instance: is the contact form too long? Do you have a way of re-targeting folks who leave before completion?) and highlight them for decisive action.
4. Create Changes
Change is hard. It means people will have to readjust their behavior. They will have to take responsibility for content decisions that may not have gone off as planned.
The changes might be on the content side (for instance, a different offer), the design side (a page that’s easier to understand), a technology side (the page needs to load quicker), or an information architecture side (the important page needs to be easier to find). Most organizations will know where to find the capable folks to do this kind of work. But it’s harder to get them to admit they need to make the changes.
Sometimes this step gets lost in automation. There is nothing wrong with multivariate testing, but this kind of optimization typically targets only smaller content elements. And often, you can see lift in making changes on the fly. These days, it’s easy enough to manage a CDN so it responds intelligently to the visitor, optimizing content for them out of existing elements. But sometimes the moves need to be more strategic. And more painful to make.
If you can communicate this requirement across all of your channels, you will have tackled the most persistent problem in optimization. Next, you will have to find out if the changes have worked.
5. Measure Success
In some ways, this follows the “shampoo principle”: rinse and repeat.
If changes have worked out well, you’ll see a jump in the percentage of prospects that were converted to customers. More users who did what you were inducing them to do. Often, content optimization tools will let you know this in something like real time — but again, it’s not always incremental or situational change that’s needed. And if you’ve made wholesale changes, this phase is much more involved than looking at the results of an optimization tool.
In a fast-changing environment, it’s hard to determine what new content elements are equivalent to the old. That’s why it’s important to separate strategic goals (ex. “more leads”) from report nomenclature. You may have to classify new content in a way that allows you to compare it to what you had before.
Finally, accept that there is no “finish line”: repeated rounds of measurement and change will lead to continually enhanced results.
Emily Ma, product director of Tencent’s advertising platform products department, was a keynote speaker at ClickZ Live Shanghai where she discussed the ... read more
The terms that customers type into your site search function can help you to gain an understanding of user behaviour and can be used to optimise ... read more
Google Analytics comes with lots of standard reports and settings, but with a little customisation you can extract much more value. One way is ... read more