A new client arrives with a budget and a reasonable instinct: try everything, see what works. Ninety days later there is data from six channels, none of it statistically meaningful, and no clear answer about what to do next.
Splitting a modest budget six ways does not give you six tests. It gives you six inconclusive results and a spent budget.
Days 1–14: tracking before spending
Nothing goes live until we can attribute an enquiry to its source with confidence. That means conversion events defined at the outcome that matters — a qualified enquiry or a funded account, not a page view — plus server-side tracking where the browser cannot be trusted.
This feels like a two-week delay. It is not. It is the difference between ninety days of data you can act on and ninety days of numbers you will argue about.
Days 15–45: one channel, done properly
Pick the single channel where intent is highest for your offer. For most businesses selling a considered service, that is paid search — people actively looking for what you do. For visual and impulse-led products, it is usually paid social.
Then commit enough budget to that channel to reach significance within the window. A tight campaign structure, ads matched to specific landing pages, and disciplined negative keyword work will tell you more in a month than six half-funded experiments will tell you in a year.
- One campaign per commercial theme, not one campaign for everything.
- Ad copy that matches the landing page headline, literally.
- Negative keywords reviewed weekly for the first month — this is where budget quietly leaks.
- No automated bidding until there is enough conversion data to feed it.
Days 46–75: fix the landing page, not the ads
By now you will have a cost per click you can live with and a conversion rate you cannot. This is normal, and it is almost always the page rather than the traffic.
Match the page to the promise in the ad. Remove every field from the form that is not needed to have a first conversation. Put proof — numbers, named clients, real reviews — above the fold. Make the phone number tappable. These are not growth-hacking tricks; they are the difference between a two and a six percent conversion rate.
Days 76–90: scale only what held
Increase spend only where cost per acquisition has held stable for two consecutive weeks. Raise it in increments of roughly twenty percent, not by doubling — algorithms re-enter a learning phase when you shock them, and you lose the stability you just established.
This is the point to add a second channel, funded by proven returns from the first rather than by hope. And it is the point where retargeting starts to make sense, because you finally have an audience worth retargeting.
Boring, sequential, and profitable beats broad, simultaneous, and inconclusive — every quarter.
