The question your category is avoiding.
Brand Strategy
Category Audit
Positioning
Differentiation
Visual Identity
Brand Strategy
Category Audit
Positioning
Differentiation
Visual Identity
.webp)
Thirty AI brands homepages, cropped to the hero and left in the order found.
AUTHOR
Alia Sidorskaya
READ
7 min
TOPIC
Category Audit
We opened thirty AI brands websites in one afternoon. Twenty-six of them made the same three moves — here is what that says about standing out in a market where everyone has already said everything.
Last year we took on a brand identity for Arkane Group, an AI consultancy in Sydney. Before drawing anything, we did what we do at the start of every identity project: opened every competitor/ AI brand site we could find and read them.
Usually this produces nuance. A spectrum, a few outliers, some room to move. This time it produced a template. Twenty-six of the thirty made the same three moves: a gradient, a network of nodes, and a photograph of someone pointing at a screen.
All thirty talked about capability. Not one of them said what happens if the project fails. That absence turned out to be the most useful thing we found, and it is not specific to AI consultancies. Every saturated category develops a question it has quietly stopped answering.
{01}
CONVERGENCE
Convergence is not laziness
Those thirty firms are not lazy. Category conventions exist because they work. A gradient and a network diagram tell a visitor in about a second what kind of company this is. Legibility is a real asset and abandoning it costs you something.
The problem is that legibility and distinction pull in opposite directions, and most companies resolve that tension without noticing there was a choice. They look at the category, absorb its conventions, reproduce them, and end up reading as competent and forgettable at the same time.
{02}
INHERITANCE
You inherit the category's reputation, whether you earned it or not
Arkane was founded in 2024. No track record of its own yet, which we treated as a constraint until we realised it was the actual problem. A buyer evaluating a company with no history has to borrow one, and the one they borrow belongs to the category.
By 2025 that record was poor in a way anyone could look up. MIT's NANDA report put the share of pilots returning nothing measurable at 95%. Abandonment of AI initiatives ran to 42% that year, up from 17%. Arkane's buyer had usually not read those numbers. They did not need to — they had been sold AI once already.
So the problem was never visibility. Every one of those thirty sites was perfectly visible. The problem was that being seen changed nothing, because being seen meant being filed alongside the other twenty-nine.
Your brand's first job is not to introduce you. It is to separate you from what your buyer already believes about companies like yours.
{03}
COLOUR
Why nineteen of them were blue
There is nothing wrong with blue. It is the correct colour for financial services, enterprise software and healthcare, for reasons that go back decades and are mostly sound.
But a default is only safe when you already have credibility to spend. If you lead the market, adopting the category’s visual language reads as confidence. If you are the newcomer, the same choice reads as membership — you have volunteered to be judged on the category’s terms, and those terms include its failure rate.
Dominant hero colour of each of the thirty, ordered by hue. The marked run is nineteen of them.
For Arkane, blue was not a bad choice. It was an unavailable one.
{04}
LIMITS
Where this stops working
A brand can name the risk. It cannot remove it. If Arkane says the thing about finishing, and then a project does not finish, they have failed against a standard they published themselves. The category's silence on risk is cowardly. It is not stupid — saying nothing is genuinely safer.
We think the trade is worth making. We are also aware we are not the ones carrying it. The studio walks away at handover and the client lives with the promise for years. Anyone selling you positioning should be asked what happens to them if the positioning turns out to be wrong.
{05}
METHOD
Run the audit yourself
Three hours, a browser and a spreadsheet. List twenty-five to thirty competitors, including the ones you lose to. Record the dominant colour, the hero image type, the first claim and the typeface. Then open a column for what none of them mention — that column is the point of the exercise.
The column on the right stayed empty for the first ten sites, then filled quickly.
Count. Anything above roughly 60% is a convention. Anything at 100% is either a category requirement or a category blind spot, and working out which one is the whole job.
→
WHAT TO DO NEXT
01
Download the sheet above and block three hours this month. Twenty-five to thirty competitors, including the ones you lose to.
02
Count before you brief anyone. Anything above 60% is a convention you are choosing to join; anything at 100% is the conversation.
03
Ask whoever is selling you positioning what happens to them if the positioning turns out to be wrong.
*
SOURCES
MIT NANDA, State of AI in Business 2025 · Koto, Huntington Bank rebrand
About Us
Slick Prjct is a brand strategy and visual identity studio for companies that need to compete at a higher level.
Brussels, working worldwide.