SEO

Which search channels should you focus on?

A free three-question tool that returns a ranked shortlist of search surfaces plus, more usefully, the ones to drop. No signup, weights published.

2 min read
seoseo-strategychannel-strategyfree-toolsplanning

Audience, then job, then surface, then format, then outcome. In that order, and never platform first.

Platform-first planning produces a channel with no job attached to it, and a channel with no job attached cannot be evaluated, which is why those accounts never get closed.

Five answers, and if you cannot fill in all five, you have a channel idea rather than a plan.

The tool

Here is a tool that does the surface step. Three questions, and it hands back a ranked shortlist plus, more usefully, a list of what to drop.

Three questions

Which surfaces are actually yours?

Answer three questions about the business and this ranks the six surfaces for it. It is my judgement with numbers attached, not a calculation, and the weights behind it are readable in the repo. Treat the output as a starting shortlist you then argue with.

01 What do you actually sell?

Pick the closest. If two fit, pick the one that produces most of the revenue.

02 Who decides, and how long do they take?

The longer the decision, the more the validate job matters and the less discovery alone will do.

03 How does the purchase actually happen?

The act job. Wherever this lands is the surface you cannot afford to be sloppy on.

Answer all three and the ranking appears. Nothing is sent anywhere; the arithmetic runs in your browser.

A starting shortlist, not an answer. Argue with it using what you know about your buyers
Weights are in the repo, in data/search-everywhere-lesson.ts, and they are my judgement

Take the bottom half seriously

Where you should not spend time is as important as where you should, and it is the half nobody writes down.

For a B2B SaaS company Pinterest is almost certainly irrelevant. For emergency plumbing, Maps beats every AI surface combined, and I would tell that client to ignore ChatGPT entirely this year without any hesitation at all.

The tool is my judgement with numbers attached, not a calculation, and the weights are readable in the repo. If you and it disagree, you win.

Now write the stop list

Where this fits

The six surfaces this tool sorts, and the five search jobs it weighs them against, are explained in search everywhere optimization, step one of the SEO roadmap. If your buyers turn out to be mostly on web search, step two is how search actually works. If you want to see how different the answer gets across business types first, there are four mapped journeys in customer journey SEO.

Frequently asked

Questions About This Article

How do I decide which search channels to invest in?

Audience, then job, then surface, then format, then outcome, in that order and never platform first. Name one buyer, name the job they are doing when you want to reach them, pick the surface that serves that job for that buyer, then the format that surface rewards, then the one number that would tell you it worked. If you cannot fill in all five, you have a channel idea rather than a plan.

Why does the tool tell me what to drop?

Because where you should not spend time is as important as where you should, and it is the half nobody writes down. Every surface has a minimum viable investment before it returns anything, and that minimum is set by the competition on that surface rather than by your ambition. Half a presence on six surfaces clears the bar on none of them.

Is the recommendation a calculation?

No. It is my judgement with numbers attached, and the weights are readable in the repository. It is a structured second opinion, not an answer. If you and it disagree about your own market, you win.

How many channels should a small team run?

Two or three, chosen because your buyers use them, done well enough to clear the bar on each. The common failure is adding five at once: a dead account on each surface eleven months later, and the same traffic you started with. The channels were not the mistake. Adding five at once was.