Saying what you are in one sentence
The hour-long positioning exercise we run before any channel work: five questions in order, the sentence they produce, and the two tests it has to pass.
Before any channel work, we run this. It takes an hour and it decides everything downstream, because a campaign for a company that cannot say what it is will be a well-executed campaign for nobody. The mistake it exists to catch is not bad writing. It is a company describing the thing it wishes it were selling instead of the thing a buyer already has a word for.
Why this comes before the channel
In a developer community, positioning is not a brand exercise. It is the first sentence of a post, and it is the one the reader decides on.
A post that opens with a category people recognise gets read. A post that opens with an invented category — a "workflow intelligence layer", a "developer experience platform" — gets one of two responses: a comment asking what it actually does, or nothing at all. Both are expensive. The first makes you explain yourself in public, under a post you wrote; the second burns a placement you only get once in that room.
So the sentence has to exist before the first post does, and it has to be one the client's own engineers would say out loud.
The five questions, in order
- 1What category are you in? Not the one you would like to invent — the one a buyer already has a budget line for. If nobody has a budget for your category, you are selling against "do nothing", which is a different campaign.
- 2Who is the alternative? Usually not a competitor. Usually a spreadsheet, an internal script somebody maintains grudgingly, an agency, or doing nothing at all.
- 3What do you do that the alternative cannot? One thing. If you have three, you have not chosen, and a reader will not choose for you.
- 4Who is it most true for? The narrower the answer, the better every sentence you write afterwards. "Teams running their own Postgres" is a position. "Developers" is a demographic.
- 5What is the proof? A number, a named customer, a demo somebody can run. An unprovable claim is a claim you cannot lead with, and in a technical room an unprovable claim is the one that gets picked apart first.
Answer them in order. Question three is unanswerable until two is settled, and most arguments about three are really disagreements about two.
The sentence
Then it is one sentence, in this shape:
For [who], [product] is the [category] that [the one thing], unlike [the alternative].
It will be ugly. It is not copy — it is the thing copy is written from. Nobody outside the room ever reads this sentence; they read the twenty posts that were written against it.
The three answers that are usually wrong the first time
- The category is a wish. A team calls itself an "AI agent platform" because that is what raised money last year, while every one of its users would describe it as a job queue. Ship the post that says job queue; the other one gets ignored by both audiences.
- The alternative is a competitor nobody uses. Ask the last five customers what they were doing the week before they bought. It is almost never the competitor on the comparison page.
- The proof is a feature. "Sub-millisecond" is a claim. "Sub-millisecond, and here is the benchmark and the machine it ran on" is proof. In a technical room the second one survives the replies and the first one starts them.
The two tests
Somebody outside the company can repeat it back after hearing it once. And a salesperson can say it on a call without wincing.
The second test catches more than the first. A sentence a founder likes and a salesperson will not say is a sentence that describes an ambition rather than a product, and it fails on contact with a room full of engineers.
What we do with the answers
The five answers become the constraints on every post we write for you:
- The category decides which subreddits and threads are on-topic at all. A product in the wrong room is spam even when nobody meant it that way.
- The alternative decides what the post argues against, which is the only honest reason for a technical audience to read it.
- The one thing is what a post is allowed to claim. Anything else goes in the replies, if asked.
- The who decides the vocabulary, down to whether we write "Postgres" or "PostgreSQL".
- The proof decides whether a post can carry a number at all. If you did not supply one, we do not write one.
When the answer is that there is no position yet
It happens, and it is worth saying out loud rather than starting anyway. A product three weeks from a pivot, or one whose first ten users each bought it for a different reason, does not have a position to launch against — it has a research question.
In that case the channel work we would recommend is different: fewer posts, more questions asked in public, and a month of reading what people reply. That is cheaper than a campaign, and it produces the answer to question two, which is the one that was missing.
If the answer is that these communities will not care either way, we say so and pass, before you have spent anything. What we take on and what we do not is the shorter version of that position.