So... What's your conversion problem? It's not rare in this business to find practitioners optimizing websites... without knowing what problems they are trying to fix in first place.
They know the symptoms (high bounce rate, low conversion, poor revenue) but they just don't have a clue what illness they are suffering.
Is due to a qualifying issue? The Value proposal is not clear? Are the users understanding the benefits of the product? Is it a trust generation problem? Is the page clear about what to do next?
They just don't think about it. So they start MayBe testing.
Sometimes, they happen to "fix the problem" and claim "Victory" after a 5% lift in conversion rate.
Is that all? Yep, as long as you don't dig deep in to the causes of your bad performance.
As Kim Goodwin states in her excellent book the aim is: "Understanding the problem before solving it".
In other words, not skipping the stage 2 of the Scientific Method: Research.
Choosing which Landing you want to improve using KPI's is OK, but it's just the stage 1 on the Scientific Method: Stating the problem.
Stage 2 implies research to understand it.
In the case of the landing, we could use for this purpose:
- Web analytics data segmenting your KPI's by traffic sources, and keywords.
- Segmented data for visitor's dimensions such as location, type of visitor, operating system, isMobile or language.
- Heat maps tool to understand how users are interacting with our site (don't use Google Analytics for that, please!). Crazy Egg costs 9.9$ per month.
- Site Search data. The number one source of data to understand customers problems.
- Recording user's sessions and analyze their mouse trails. You can use Inspectlet for that.
- Online survey tool to have metrics like: visit purpose, satisfaction, task completion rate and open fields.
- Using user feedback tools (Kampyle, User Voice, Feedback, etc) and also online chat logs, customer service reports, etc.
- Analyzing some competitors. Pricing, offers, rankings, reviews, comments, industry data and papers, etc.
- Talking to stakeholders (if any) to find out about the product's background.
- Whatever helps to understand the problem ;)
Notice that so far we are changing nothing. That's why Usability and User Testing tools are let a side (we would only performance usability test of the current version in order to have insights, but not to test a new solution.... yet).
Like my good friend (and best UX manager) Victor Solàwould say, this is just good 'old UX mixed with some Web Analytics.
And he's right. UX has been doing this for years. CRO just puts some Web Analytics and A/B testing on top of it.
Once you've done that research, hypothesis will show up for them selves. And you'll be able to understand better what your problem is. Therefore, you'll be able to fix it better.
Your conversions and your confidence will thank it.
BTW, I'm looking for new Conversion Optimization Adventures.