We said no 163 times

Across 33 distribution plans the planner picked a channel 74 times and ruled one out 163 times: what it kept refusing, and the channel nobody calls a channel that quietly won

Between late June and late August we generated distribution plans for 33 products. Every plan scores the same ten channels and marks each one focus, test later, or avoid for now. That is 330 verdicts, sitting in a database, that I had never actually looked at in aggregate.

So I ran the query. Focus came back 74 times. Avoid for now came back 163 times. More than twice as often as it recommended anything, the plan recommended against something.

That is not how distribution advice usually reads. The genre is a list of nineteen tactics and an implied instruction to attempt all of them. What the planner mostly produces is a shorter list of things to ignore this quarter, which is less satisfying to receive and, I think, more useful to act on.

Here is the whole table.

Paid ads went nought for thirty-three

Thirty-two avoid, one test later, zero focus. Not a single product out of 33 got told to start with ads.

I did not tune it that way, and when I saw the number my first thought was that the prompt must be biased. Reading the reasoning changed my mind. Ads are a multiplier, and almost every product in the set was still missing the thing being multiplied. You need to know which sentence makes someone sign up before you pay to show that sentence to strangers. Spending money to find that out works, but it is the most expensive way to run the experiment, and the founders in this dataset mostly had a landing page, no data, and a budget measured in hundreds.

The one product that got test later rather than avoid was the one with an existing conversion rate worth scaling. That is the actual rule hiding under the 32.

The channel nobody calls a channel

Top of the table is trust packaging, at 6.72 average fit and 17 focus recommendations. It is not a channel in the way Reddit is a channel. It is the pricing page that answers the obvious objection, the screenshot that shows the product actually running, the about page with a human on it, the proof that somebody else already paid.

It keeps winning because it is the only item on the list that does not need traffic to pay off. Ten of those 17 expected a signal within days. Everything else on the board is a way to send more people to a page; this is the one that decides what happens when they arrive. If a hundred people visited last week and two signed up, the cheapest available growth is not another hundred visitors.

The one I did not enjoy reading

SEO was the first move five times out of 33. Sixteen test later, twelve avoid for now. Of the five that did get it, four expected months before any signal.

You are reading that on a blog we are deliberately building to rank, which is either a contradiction or the point. We can afford months. This is a compounding asset for a product that intends to exist in two years, and the cost of writing it is my evenings rather than a budget. A founder with 40 users and three months of runway is playing a different game, and telling that person to start a content engine is close to telling them to take up a hobby.

SEO is rarely wrong. It is frequently the wrong thing to do first, which is a distinction the nineteen-tactics genre never makes.

Two channels live permanently in limbo

Partnerships: one focus, sixteen test later, sixteen avoid. Creator and short-form video: two focus, ten test later, twenty-one avoid.

Partnerships is the interesting one. It is almost never rejected outright and almost never chosen, which is the profile of something that would probably work and that nobody has the standing to attempt yet. You cannot partner your way out of having no users, because the other side of a partnership is doing arithmetic about your audience. Creator is blunter: it scores badly for products whose buyer does not scroll, and there are a lot of those.

Outbound splits the room

Eleven focus, eighteen avoid, four test later. That is the widest gap between enthusiasm and dismissal in the whole table, and it is the one channel where the average score tells you nothing useful at all.

It is not a verdict on outbound. It is a verdict on product shape. When the buyer is a company, the contract is worth more than a few hundred a year, and you can name twenty accounts that would benefit this quarter, outbound came back focus and expected a signal in days. When the product was a five-dollar-a-month consumer utility, it came back avoid, because the arithmetic never closes: you cannot spend forty minutes researching someone who will pay you sixty dollars a year.

Both of those are the same channel with the same tooling. The thing that changed was who was on the other end, which is why founder-led outbound is worth real effort for some of the products in this set and a waste of a quarter for most of them.

What this is not

These are recommendations, not results. Nobody in this dataset has reported back with revenue. The table tells you what one model, running one prompt, concluded after reading 33 landing pages, and a model that has read the entire internet on marketing has absorbed the internet's conventional wisdom along with everything else.

Thirty-three is also a small number, and two months is a short window. If I publish this again in a year against measured outcomes rather than opinions it will be a better article, and some of it will be embarrassing. That version is worth waiting for. This one is worth writing down now, because the ratio is the part I did not expect and the part that survives whatever the outcomes turn out to be.

Two thirds of the available channels were wrong for two thirds of these products. Not permanently wrong, not wrong in principle: wrong this quarter, for this product, at this stage. If you want the procedure rather than the numbers, the six questions that rule most of them out are written up separately. The plan is free and takes about thirty seconds, so if you would rather have it done for you, give it your URL and argue with what it tells you.