When a week goes badly, the temptation is to change everything: new audience, new message, new pitch. Then a good week arrives and you have no idea why.
Three variables
Every test in Switchback is a product (P), an audience (A) and a message (M). The code on each lead row, like P1 · A2 · M3, tells you exactly which combination it belongs to.
One change per switch back
When you switch back, change one letter. If M3 beats M2 with the same audience, the message did it. If A2 beats A1 with the same message, the audience did.
Keep the old results
A new version never overwrites the old one. That's what makes the comparison possible, weeks later, when you've forgotten what you changed.
