Stop playing at the KPI Roulette Wheel
Key Performance Indicators (KPI’s) like Customer Acquisition Cost and Conversion Rates are a side-effect of design, policy and product. Change the any one of these, change your KPI’s.
The problem comes when you realise your redesign has ended up becoming a game at the worlds largest casino and you’re sat at the wheel with all your companies executives around you.
You’ve staked your bet, you’ve spread along a set of the latest trends in web development and research that’s currently talked about at all the conferences.
When you sign off the spec-sheet, and all your contracts are signed, all bets are closed. You’ve already decided what your KPI’s are likely to be — but not a single person in the house knows what it is — until you launch, and the ball stops rolling.
Let’s take a look at what our possible outcomes are.
1. Worst Case Scenario
Most people tend to think that the worst case for a redesign is that you drop a load of percentage points on your KPI’s. I’d argue that staying exactly the same would be the absolute worst case. You actually didn’t learn anything.
What?! We’ve wasted a hundred grand to find out absolutely nothing?!
2. Losing Percentage Points
Awesome! This is where I like to live. Being down with the people who are in eCommerce hell. Some of the best questions for this time include:
- What have you done wrong?
- What can we learn from this?
- Are you going to stop listening to people who lick your ass for cash?
Fastest route to resolution: Revert. Flip flop. Run back to what you just put in the trash. Get it back out of the trash. Say you’re sorry, and kiss it like it’s your momma.
3. Best case
You’ve jumped up a few points, even better! But what are you going to do now? Wait 3 more years until you’re ready to do your next redesign? Play on the table again?
Or are we going to all going to cut this crap and move to a better strategy.
How can you tell if something is wrong? Explain why your site performance is improving. No lying, assumptions or stabs in the dark.
