Your opinion in this PPC strategy

Hello Gerard.

>> 1. I set a campaign to the same audience with 1 ad with 3 different variables (images, text).

That's not one ad, that's three. You should never test just one ad at at time, always two and never more, although I've been known to break that rule sometimes.

You test only two ads so that they both get shown to about the same kind of audience. You don't want to dilute the stats which would make your test meaningless.

>> 2. I run the campaign for 7 days to find which ad had the best results.

It's not the time that is important. It's the clicks. For each ad.

If you run for 7 days (that's a week like the song says), and get just five clicks per ad, that's not enough. My standard is a minimum of 20 clicks per ad and more is better. Regardless of how long it takes to get them although some other stats may make me break that rule.

>> 3. I set up a second campaign with the best result from the first test, but now I test for different audiences.

I don't know what you are selling but normally, a product has just one audience. If not, it will have one main audience which is the one you should be concentrating on.

>> 4. I keep testing for variables until I feel satisfied with a combination of elements.

Yes, keep testing different things. But never stop. I have a client for four years and I've got some high volume groups where I've tested dozens of ads and I still try to come up with something better. You don't see Coke and McDonalds ads from years ago do you?

>> 5. Then I start my main PPC campaign with the results from the exploration campaign.

What you describe is your main campaign, the only campaign which you keep testing different variables.


>> I believe this path can yield better results

Yes, but you still have to do it right. I've seen many who did exactly what you suggest. The problem is they did not test properly. They used virtually the same elements for all ads such as keep on using the same text for 20 different ads and expecting different results.

As I said, you should never stop, even if you have the best possible ad. There's always something better.

There's new feature too that you have to take advantage of. Adwords for instance late last year introduced callout ad extensions. Skeptical at first, I saw that they improved click rates. But my ads very often used that idea so I had to change to take callouts into account and change the ads. The callouts were based on what I knew worked but I could not have the callout repeat what the ad says. This sort of thing happens whenever there's a new feature: you have to think about the impact and how to take advantage.

You should base decisions based on number of clicks, not time. That's at least 20 for each ad being tested and more is better. Time is irrelevant here.

You may be thinking, why 20 or more? It's more statistically significant. Think of it this way. If you had Ad1 going 5/40 (12.5% CTR) and Ad2 6/50 (12%) that is not statistically significant. If Ad2 had just one more click, its CTR would be 14%. That would not tell you anything. But if the stats were ten times greater, 50/400 and 60/500, adding that one more click would result in a 12.2% rate so you'd be more confident that Ad1 is indeed the better one, if only marginally, over the long run. I'm of course looking only at CTR and not considering conversions here.

Another way to look at it is that one extra click in 20 is only 5% more so the margin of error and significance is lesser.