Getting started
What can a marketing tool know about you on day one?
Onboarding produces a confident market picture from one paragraph about your business. What that is worth, what it is labelled, and which parts you should overwrite.
How MarketWar handles it
Every list onboarding produces is labelled a STARTING HYPOTHESIS rather than a finding — the audience map, the personas, the problem map, the competitor angles, the keyword and question universes, the pillars and the 90-day play — and nothing invents a testimonial, a review or a performance metric. All variation comes from a seeded hash of the business and industry rather than from a wall clock, so the same description always produces the same picture and you can tell what changed when you edit it.
Every marketing platform asks you to describe your business and then produces a market picture: brand voice, audience segments, personas, competitor angles, content pillars, a ninety-day plan.
It looks like research. It is produced in four seconds from one paragraph, and what it is worth depends entirely on whether the product admits that.
What it is, stated
Every list here is labelled a starting hypothesis, not a finding. The audience map, the personas, the problem map, the competitor angle map, the keyword and question universes, the pillars, the plan: all of it is a structured starting point meant to be refined with real data.
That label is the whole difference. A hypothesis invites you to disagree; a finding invites you to plan against it. The same text under the wrong label produces a business running a year of marketing against four personas a model invented on a Tuesday.
What it genuinely gives you
A blank page is slow, and most small businesses never finish one. Structure is worth a lot:
- a first draft of the objections you will have to answer, which is the hardest list to write cold,
- a question universe — the things people actually type — which is where the content plan comes from,
- a problem map, which is where the copy comes from,
- a ninety-day sequence, so the work has an order.
Four seconds to the first version, and then the real work is editing it.
Which parts to overwrite immediately
The problem map and the objections. You already know these better than any model, because you have heard them on the phone. Replacing the generated versions with the sentences your customers actually use is twenty minutes of work and it improves everything downstream — the copy, the pages, the answers.
The personas, or delete them. A persona assembled from an industry assumption is the part most likely to be quietly wrong in a way that costs money. If you have real customers, replace it with three real ones, by name.
Nothing is invented
No testimonials, no reviews, no performance metrics.
That exclusion list is specific because those three are where generated onboarding does real damage. A fabricated testimonial is a fake review, which is a consumer-protection matter in most markets and not a growth tactic — and a product willing to generate one has told you what else it is willing to generate.
Reproducible, which sounds dull and is not
The same description always produces the same picture, because there is no randomness and no wall clock in it.
The payoff is diagnostic: when the picture changes, your description changed. With randomness in it you could never tell whether an improvement came from your edit or the dice, which makes iterating impossible.
How to know when you have outgrown it
When you have real feedback to cluster, real queries you appear for, and a number for how many people you have asked. At that point your own data replaces the hypotheses, and the question of what to write next has real signals behind it.
A business still running on day-one hypotheses in month three has mistaken the scaffold for the building.
The question to ask any onboarding flow
Which of these did you measure, and which did you guess?
If a product cannot separate those, every confident sentence in it is a hypothesis you are about to mistake for research.
What this does not do
It is a hypothesis built from one paragraph, so the personas are guesses in the shape of personas. Their value is as something to argue with on day one, and a business still running on them in month three has mistaken a scaffold for research.
Common questions
Can a tool really know my market from a paragraph?
No, and this one says so. What it can do is produce a structured starting point far faster than a blank page, with every list labelled as a hypothesis rather than a finding.
Which parts should I overwrite first?
The problem map and the objections, because those are the two you already know better than any model — and they are the parts that drive the copy.
Why does the same description always give the same answer?
Because there is no randomness and no clock in it. Reproducibility means that when the picture changes you know your description changed, rather than the dice.
Does it invent social proof?
Never. No testimonials, no reviews, no performance metrics. Those are the fabrications that do real damage, and a fake review is a consumer-protection matter rather than a growth tactic.
Related
What are your customers actually telling you?
Reviews, surveys, support messages and sales calls say the same three things, and nobody reads them. How clustering differs from inventing.
What should you write about next?
Content calendars are usually guesses with dates on them. The seven factors in a published formula, and why being able to re-weight them matters more than the score.
How many people have you actually asked to buy?
No customers has a cause, and rarely the one that gets blamed. The one number that tells you which problem you have, and why tools hide it.