How long do you usually test an offer before killing it?

[SuperLuckyAds]

I've been testing a few offers lately and keep changing my approach.​

Sometimes an offer looks terrible after the first couple of days, then starts converting once enough data comes in. Other times the early numbers look great and completely fall apart after scaling.

I'm curious how others handle this.

Do you set a fixed test budget for every offer, or do you keep running it until you see enough data to make a decision?

 

[thaaz]

I’d use spend and funnel data, not days. Set the target CPA before launching. Once tracking is verified and the normal conversion delay has passed, spending around 2–3x that CPA with zero conversions is a reasonable pause point, but not a hard rule.

Then look at where the funnel breaks. Weak CTR usually points to the creative or targeting. Clicks with no page action suggest the message or landing page. Checkout drop-off makes me look at the price or payment flow. Keep the main variables stable during the first test or the data becomes hard to read.

 

[sam.curun]

I've been testing a few offers lately and keep changing my approach.​

Sometimes an offer looks terrible after the first couple of days, then starts converting once enough data comes in. Other times the early numbers look great and completely fall apart after scaling.

I'm curious how others handle this.

Do you set a fixed test budget for every offer, or do you keep running it until you see enough data to make a decision?

While I normally have a test budget, I would not solely base my judgment on the first 1-2 days. I would lay out some kill/scaling rules from the beginning and evaluate the quality of conversion along with some data.

 

[darftwood]

While I normally have a test budget, I would not solely base my judgment on the first 1-2 days. I would lay out some kill/scaling rules from the beginning and evaluate the quality of conversion along with some data.

Exactly. It helps that kill and scale parameters are known before starting the test because it removes any emotion from the process of testing. I would also monitor CPA and conversion quality through a significant number of days rather than just one or two good/bad days.

 

[amyleon]

The first 1–2 days usually aren’t enough to judge an offer, especially when conversion data needs time to settle. I’d set clear kill/scale rules upfront and watch CPA along with conversion quality, so a few unusually good or bad days don’t drive the decision.

 

[missrose]

I typically allocate a testing period of two to four weeks, continuously optimizing along the way; if the ROI remains unsatisfactory after four weeks, I will decisively stop the campaign.