Challenge · Brand discovery
The brand questionnaire nobody fills in
Every engagement starts with a document that assumes the client has three uninterrupted hours. They do not, and the work starts anyway.
20 August 2026 · 6 minute read
The brand questionnaire fails for a structural reason: it asks the busiest person in the room to write an essay about something they know so well it is hard to articulate. The fix is to invert it. Gather everything already written down, extract what it answers, and only then ask about what is genuinely missing, in conversation rather than in a form. In practice this cuts the questions from forty to about eight, and the eight that remain are the ones only a human can answer. A brief built this way arrives in days rather than weeks, and it is more accurate, because most of it came from evidence instead of recollection.
Why does the questionnaire always come back thin?
Three reasons, and none of them are that the client does not care.
- It asks for articulation, not information. "What is your brand personality" is a genuinely hard question to answer cold, even for someone who has run the brand for a decade.
- It is sent to whoever is available rather than whoever knows, because the person who knows is the founder and the founder is the busiest.
- Most of it is already answered somewhere. Being asked to restate what is on your own website is the fastest way to make a form feel like homework.
What should you read before you ask anything?
More than you would think. Most brands have documented far more than they realise, just not in one place.
| Source | What it usually answers |
|---|---|
| The website | Positioning, product language, the words they already use in public |
| Product catalogue | Range, hero products, price architecture, what the business actually sells |
| Old decks and pitches | Origin story, mission, the numbers they are proud of |
| The last year of social | Actual voice, as opposed to intended voice |
| Reviews and app store feedback | What customers value, in their words rather than the brand's |
| Press coverage | How the outside world describes them, which is often not how they describe themselves |
Read all of it first. Then list what you still do not know. That list is your questionnaire, and it will be short.
What is left that only a person can answer?
In our experience, about eight things. They are worth naming because they are consistent across brands.
- Why the business exists, in the founder's own words rather than the website's.
- Which customer they would rather have more of, and which they would quietly rather not.
- What they have tried that did not work, which never appears in any public document.
- What they will not do, even if it would sell.
- Who they consider a competitor, which is frequently not who the category data says.
- What has to be true in twelve months for this year to have been worth it.
- The thing they are worried about that nobody has asked them.
- Which words make them wince.
Ask the last one
The words-that-make-you-wince question produces more usable output per minute than anything else on the list. It is concrete, it is easy to answer, and it directly generates the forbidden words list that stops future work sounding off-brand.
How do you keep the answers from rotting in a folder?
By writing them into the place the work actually reads from, rather than into a document about the work. A brief that lives in a shared drive gets read once. The same answers, written into the voice specification and the audience definition that every subsequent piece of work is built from, get used every time.
This is the difference between research that informs a kickoff and research that informs everything afterwards. Most brand discovery is the former, which is why it gets redone every eighteen months.
The best question in brand discovery is the one you did not have to ask because you read the answer first. Every question you ask that the material already answered costs you a little credibility with the person answering.
If you are an agency
This method shortens onboarding from weeks to days, which is the part of a new account that clients judge you on before any work has shipped. It also changes the tone of the kickoff: arriving having already read everything and asking eight sharp questions positions you very differently from arriving with a form. If you run several accounts, the read-first step is also the most delegable part of discovery, while the eight questions are where senior time is actually worth spending.
How we do this ourselves
The Brand AI Assistant does the read-first step automatically, then asks only about the gaps across seven areas of the business. What it learns is written back into Brand DNA as coverage improves, so the answers land where the work reads from rather than in a transcript. It can also search the web mid-conversation when a question turns out to be a research question, and every message is screened for personal information before processing.
How the Brand AI Assistant works