Everyone wants to know the right price for a PPV. There isn't one. There's a right price for a certain guy at a certain moment, which is why good price tests target a moment instead of hunting for one magic number.
What to test
These are the moments we test on our own account. Each one is its own test.
- The first paid offer to a guy who's never bought. The hardest sale, and every later sale depends on it.
- The start of his next script. After he's finished one, how high does the next one open?
- The restart price. A buyer comes back after a break. What does his first offer cost, relative to what he usually pays?
- The size of a cut. When a PPV doesn't sell and goes out again, how much off?
- The price band for singles and bundles, for guys who've already bought.
Test inside a tier. A follower and a big spender aren't the same buyer, so we give each tier its own price levels, and every guy stays on one level for good. Mix them and you'll get an average that fits nobody.
What to measure
Money per guy in the group, counting the guys who bought nothing, over a fixed window.
Conversion rate alone rewards cheap prices. Average sale alone rewards expensive ones. Say two groups have 100 guys each. Group A sees $10 and 20 guys buy: $200, or $2 a guy. Group B sees $20 and 12 buy: $240, or $2.40 a guy. A looks like the winner on conversion. B made more. Those are made-up numbers, but that's how a test gets read the wrong way round.
Keep counting after the first sale. A cheap first PPV might bring a second and a third. A high one might win the week and lose the month. That's why the window should be long enough to catch what he does next.
How long to run it
There's no honest fixed number of days, because what you need is buyers, not time. Our first 8 days had 24 buyers and 64 sales from 645 offers. Split that in two and you've got about a dozen buyers a side, which can't tell you anything about price.
Two rules we use. Run through at least one full pay cycle, because guys buy around their paydays and a test that only covers the week after payday tells you about that week. And keep going until no single guy can swing the result, which brings us to the main trap.
If you run several models, pool them. The same test in the same tier across your accounts reaches a usable number far sooner than one account alone.
What fools you
One whale
Our first 8 days: $1,865 gross, of which one fan paid $1,345. Without him, about $520. Now picture a price test running that week. Say the other 23 buyers split evenly, about $260 a side. He lands in group A. Group A shows about $1,600, group B about $260. You'd roll out A's price to everyone and congratulate yourself, and price had nothing to do with it. Put him in B and B wins.
So read every price test twice, with and without the top spender in each group. Look at how many guys bought as well as how many dollars came in. If the winner changes when you take one guy out, you don't have a winner.
Stopping when you're ahead
Early leads come and go. If you check daily and stop the first time you like what you see, you'll pick a lot of false winners. Decide the end date first.
Testing on guys who know the old price
A buyer already has an idea of what things cost, and a fan should never see a price lower than he already paid. New guys give you the cleanest read on a price. Test buyers on the moments that are really about them, like a restart or a next script.
Whales in the same test as everyone else
There are too few of them to test this way. Set whale prices from what each guy actually pays, and read them as a separate line.
What to do with the result
If it wins with and without your top guy, roll it out to that tier and that moment only. If it flips, keep it running or keep the price you've got. If it's close, take the one that's simpler to run, because a tie means price wasn't your problem.