Why the angle decides the launch
There is a comforting idea that a good enough distribution partner can carry anything. It is not true, and the arithmetic shows why. Creators engaged multiplied by their median reach gives you a floor.
Everything above that floor comes from people choosing to pass the thing along, and nobody passes along a claim that gave them nothing to say.
This is the part of a launch that cannot be bought. You can buy the first wave. Whether a second wave arrives is decided by what the first wave was carrying, and that is a writing problem several weeks before it is a distribution problem.
The failure is usually invisible in the report. The launch delivers its impressions, the numbers look reasonable, and nothing happens afterwards. That is what a weak angle looks like on a dashboard: an expensive floor and no ceiling.
The one-sentence test
We work positioning down to a single sentence that states what the product does, for whom, better than what. No adjectives. Adjectives are where vagueness hides, and every one of them is a decision somebody avoided making.
Three things that sentence has to survive:
- Deletion of the superlatives. Remove every instance of fastest, most powerful, best-in-class and seamless. If what remains says nothing, the superlatives were carrying the sentence and there was no claim underneath.
- The "compared to what" question. Faster than what. Better than which alternative. A comparison the reader can hold is a claim; an improvement with no reference point is a mood.
- The specificity check. Could a competitor put their own name on this sentence and publish it unchanged? If so it describes a category, not a product.
Most launch copy fails all three, which is why so much of it reads interchangeably. The reader has seen forty versions of it and recognises the shape before reading the words.
Claims people can argue with
The test we actually use is whether a technical reader could disagree with the sentence.
A claim nobody can dispute generates no replies, and replies are a large part of what drives reach. Ranking systems weight engagement per impression, and a post that produces reading but no response is telling the algorithm that people saw it and felt nothing. The algorithm page covers the mechanism in detail.
Being arguable is not the same as being contrarian. Contrarian for its own sake collapses the moment somebody pushes back and you cannot hold the position. The real test is whether you will still be defending it in the replies that evening, because you will be.
Budget time on launch day to argue. A thesis abandoned under the first bit of pressure is worse than no thesis, since the abandonment is now the story.
The strongest version of this is a claim that costs you something. Naming a limitation, saying what the product is bad at, or taking a position that some of your market disagrees with all read as credible precisely because a marketing team would have removed them.
An argument is one way to give people a reason to respond and it is only the first.
A launch can also carry an in-joke a particular community will catch, a number sceptics will want to check, or a human detail that makes a person visible inside a corporate format, and those reach groups an argument never touches.
Give them something to reply to covers the other five, and how to aim each one at a different audience.
Where the narrative has to come from
We shape narratives. We do not invent them, and any distribution partner claiming otherwise for a deep technical product is telling you something useful about themselves.
When a client is building real-time world models or an inference stack, the people who wrote it understand it in ways no agency will in a two-week engagement.
Pretending otherwise produces copy that reads as fluent and lands as hollow, and the audience for technical products detects that immediately. They have seen it many times and they are unforgiving about it.
So the useful division is this. You supply the substance: the specifics, the numbers, the thing that surprised your own team. We work it into something that travels, then brief creators from it so they can write it in their own voice.
A content brief of one or two pages is usually enough, and the parts we most want are the ones an engineer would find interesting rather than the ones a marketing page would lead with.
The most valuable single input is the detail that could only come from having built the thing. It cannot be researched, it cannot be guessed, and it is the reason a launch sounds like a company rather than a category.
One more reason to settle the sentence before the launch rather than during it. A launch is the single moment when the largest number of people write down what your company is, on their own sites, in their own words.
Those pages are what answer engines read later when a buyer asks the category question, and a week where eleven people described you eleven ways produces a description you then have to live with.
Five ways narratives fail
| Pattern | Why it fails |
|---|---|
| The category claim | "The platform for X." Somebody larger already owns that association, so it reads as ambition rather than fact. |
| The adjective stack | Fast, powerful, intuitive, secure. Four words that survive deletion without changing the meaning of anything. |
| The roadmap as news | Announcing intent rather than a shipped thing. Readers have learned that the follow-through rate on these is poor. |
| The unattributed metric | "10× faster" with no baseline, no workload and no period. Treated as noise because every launch contains one. |
| The excitement opener | "We are excited to announce." Spends the first line of the post, which is the only line most people read, on your emotional state. |
Each of these is a decision that did not get made. The category claim avoids picking a segment. The adjectives avoid picking a comparison.
The roadmap avoids waiting until something exists. Narrative work is mostly making those decisions before the launch rather than during the replies.
Questions we get asked
Can you write our positioning for us?
We shape it rather than invent it. The specifics that make a launch travel are things only the team that built the product knows, and a distribution partner claiming to understand a real-time world model or an inference stack as well as the people who wrote it would be lying. You supply the substance, we work it down to something that carries.
Should a launch narrative be contrarian?
It should be arguable, which is not the same thing. A claim nobody can dispute generates no replies, and replies are a large part of what drives reach. Being contrarian for its own sake fails the moment someone pushes back and you cannot defend it, so the test is whether you would still hold the position in the replies at 9pm.
How long does narrative work take before a launch?
The writing is fast once the decision is made. The decision is the slow part, because it usually surfaces a disagreement inside the company about what the product is for.
Starting a week before the drop is comfortable. Starting the day before means shipping whatever consensus already existed, which is rarely the sharpest option available.