I usually start with a quick test using ads. I would direct the ads to a landing page that I created or a competitors landing page. But If I can’t even get a decent CPC or CTR on my ads, and I plan on heavily relying on ads to drive traffic, I abandon the idea.
If CPC and CTRs look good, then I would further build out the funnel to continue testing. This time for conversion rates and cost per acquisition.
Next step is the landing page. For a quick drag and drop page, you can use unbounce. Or if it’s an ecommerce shop, I use shopify.
I also have the entire backend payments set up to accept real orders. This is the best indicator of interest as opposed to an add to cart or an email sign up. If real orders do come in, I write the customer an email to explain that I’m out of stock or currently not selling, then refund the order.
Run some ads to the landing page in order to gather data quickly. If CPAs are near my targets, I move forward with developing the product. If CPAs are not near, but there is room for improvement (in the CTR of the ads, or the CVR of the landing page), I may make adjustments to the funnel and test again.
Usually a minimum of 1 week so I can account for variance due to the days of the week. But this would depend on what metric I am using as my main KPI for the test, and my daily budget.
If CPA was my KPI, I am going to have less data than if I were analyzing based off of CPC. So in that case, I would be running my test for a longer period, or spending more budget to test.