How long do you test a method before dropping it?

That's impossible to answer without knowing the context. You'll need to provide more details - are we talking about a headline or copy revamp? A new marketing platform? Some changes show immediate results, others can take weeks, months, even years to manifest.

The testing period and metrics you measure will depend on what you're trying to test and what you want to achieve.

I took 3 years of statistics in college and did pretty well with it. Fact is, people say they're doing "split testing" but they have nowhere near enough instances (called the n number) of the thing to find any statistical relevancy..

There is a myth that an n of 30 is enough. I don't agree, unless it's to determine something like "will a hammer smash an egg?.

First, I would gather at least a n of 100 before making a decision about whether a thing works or not.

Second, you have to be very careful to ensure you are treating your observations like a high school science experiment. This means changing ONE variable at a time and then waiting for that good n number of results to come in.

The problem is, people muddle. Their sales page doesn't work, so simultaneously they change the color, the headline, the CTA button, and add a video. Now, even if results do improve, you have no idea what actually did it and can't replicate that outcome.

Third, if you are learning a skill, like how to run paid ads to bring in at least somewhat prequalified leads, or how to sell using a specific process, then you are going to have to do many repetitions and probably get some coaching to get good at it. This is not a simple thing. You may be doing something wrong but not see it. I would expect at least 90 days with lots of repetition (aka 'reps') and feedback to learn quickly.

So if your n=100 and you are getting one observation in a day, it's going take you 4 months to collect enough data to know anything for sure.

I can tell you having helped people build businesses since 2012, and watched a lot of people who basically approached their new business idea like it was a New Years Resolution, most people give up after just three days of sustained effort. Consequently, they learn nothing and struggle year after year. Sad, but true.

You illah or wellah bcs this?

(Do naht ask Moi how I split tested the fkr.)

Yes testing period is long enough

That depends on a number of factors. It depends on what you're testing. What your outcomes that you want, it depends on budget etc. What exactly are you testing?

I've learned that testing is not just about waiting a certain number of days-it's about being honest with how seriously you applied the method. There were times I thought something didn't work, but later realized I was inconsistent or half-trying it.
These days, I focus more on steady execution and watching for patterns instead of quick wins.

For me the biggest mistake is changing too many things at once. You tweak the headline, swap the offer, change the traffic source all in the same week and then wonder why nothing works. I try to isolate one variable at a time and give it at least 2-4 weeks depending on how fast the data comes in. If I'm running paid ads I can usually tell within a week or two. Something like SEO or content marketing though, you need way more patience. The real question isn't how long you test but whether you're actually tracking the right numbers while you do it.