The Laziest Price Test I Ever Ran


I once did some work for a company that charged an annual membership for access to a library of video content; for the purpose of this story, we‘ll call them ExpertSchool.

Opportunity - Multiple Plans?

At the time, ExpertSchool was very discriminating about the quality of its Experts, but not so much about how much to charge, offering a single price for all potential customers.

The obvious comparable was Netflix, which offers multiple pricing plans at almost 2x the price without restricting the actual watching stuff.

What if we too could increase revenue per customer without curtailing the product?

The Obstacle: Price Changes are Expensive

As far as experiments go, “let’s change the price” is on the opposite end from “what if the button said Get Started instead of Start Now

Consider handling legacy plans, internationalization, cancellations, price changes for existing customers, porting changes to multiple app stores, etc.  Even if you end up not rolling out the changes, you will now forever have a bunch of customers subscribed to a plan that is no longer available. You’ll need to deal with that complexity for the rest of time.

Still, the opportunity is just as massive.  If only there was a way a lighter-weight way to figure out what the impact of a price change would be without paying the price change cost upfront.

The Solution: Everybody gets a car!

The key insight here was that if you exceed everybody’s expectations.
  1. Premium Plan
This is all of the features of your existing plan, but now priced at your target “premium” price.  When the customer chooses this plan, let them know that they got a promotion that drops the price to your only existing plan’s price.  

  1.  Basic Plan
Figure out which features you were planning to cut out of your plan (offline downloads, multiple simultaneous views, etc) and make that the new basic plan, your existing plans’ price.  Once customer select that plan, let them know that, as a promotion, they’ll get all of the Premium plan features at no extra cost.  

3. Medium Plan(s)
You can probably imagine where this is going.  If the customer picks a middle plan, you’re giving them both a discount and more features than they expected.

What sort of back-end changes does this kind of scheme require? If you’re paying careful attention, you’ll notice that in all three cases, customers ended up buying a plan that is in fact identical to what your original pricing plan was.  

Which is to say, you don’t need to change the back-end at all.  That means you’re effectively getting a price sensitivity experiment for the cost of a pricing page design and a pop-up.

OK but you can’t trust these results, they didn’t actually purchase

We ran the test. It was a significant winner.  Then it came time to go get buy-in, and the finance skeptics (rightfully) showed up.

  • “It’s nice that you can show this sort of improvement on the pricing page, but somebody saying they’ll pay more and actually entering their credit card is a completely different thing.  

This is a good and fair concern.  So, we re-ran the test but moved the “actually, you got a better deal than we first thought” pop-up to after the credit card entry.  The test results held up and all was well. “Sorry we gave you more and charged you less” is not an apology I’ve ever had to send.

This gave us confidence to kick off the “multiple pricing plans” workstream.  It took a while - knowing with confidence what the upside would be gave us the momentum to see things through.

OK cool story, what should I learn from this?

For any sufficiently complicated experiment, see if you can first come up with an experiment design that doesn’t actually require changing your product, letting you punt all of the hard work to once you know the change is worth making.