Siavak Analytics

Case study · Design licensing · United States

Described well, and almost never named

An independent designer licenses her artwork to brands and sells design files. The question was whether AI assistants send buyers her way.

CategoryDesign licensing and design files
MarketUnited States
Where we lookedAI assistants and Google's AI answers
FieldworkAugust 2026
1%of the assistants' answers named herWhen buyers looked for a designer without a name in mind
0%of Google's AI answers named herThe same searches, on Google
97%of what the assistants said was rightWhen they could read about her

01What she believed

She thought the problem was how she was described.

She believed there was little public material about her.

She believed the brands she had worked with kept quiet about working with her.

And she believed that if she had a problem with AI assistants, it was in how they described her.

02What surprised her

Her description was never the problem.
Getting into the answer was.

Brands she had worked with credit her in public, and some of that material was more substantial than she knew. The study found it before she did.

Almost never in the answer

Share of the assistants' answers that named her, when buyers looked for a designer

Each square is 1%. On Google's AI answers, the grid stays empty.

Source: study data, August 2026

03What we found

Right about her when they can read her.
Absent when they cannot.

Six findings, each with the exhibit that shows it.

Finding 1

Buyers looking for a designer to license from almost never heard her name. She was named in 1% of the assistants' answers and 0% of Google's AI answers.

71%

of the praise for her offer was an echo

One assistant repeating her own description word for word. The assistants that read her offer themselves described it differently.

Finding 2

When she was named, she was described well. Where an assistant could read about her, it got her right 97% of the time.

Reading her beats remembering her

Share of claims about her that were wrong

An assistant that read her site3%
An assistant working from memory45%
Another assistant working from memory51%
The assistants do not dislike her. The ones that can read her are right; the ones that cannot, guess.

Source: study data, August 2026

Finding 3

Her two kinds of buyer meet different winners. People buying design files are sent to a marketplace. Brands looking to license are sent to named individual designers, who take 25% of those recommendations against 19% for marketplaces. One fix would not serve both.

Two buyers, two winners

Two questions, two winners. The one she sells into rewards being a person.

Who gets recommended when brands look to license

Share of all recommendations, by kind of supplier

Named individual designers25%
Marketplaces19%
Licensing agents14%
Design studios13%
Stock libraries11%
Trade shows7%
The designer herselfunder 1%
Designers take the largest share, spread across many names. No one owns it yet.

Source: study data, August 2026

Finding 4

The assistants repeated an old fact about her that was no longer true. The biggest source was her own site: the words on her about page said one thing, and the data behind the page said another.

One page, two stories

What the page says

dctlpcgytk

What the data behind the page says

qquxyxriwf

What the assistants repeated

todxtntkty

The assistants believed the part of her page no visitor sees.

Finding 5

One assistant invented at least eight different people under her name.

8+

different people invented under her name

By one assistant working from memory

Finding 6

Most of the praise for her offer was an echo. 71% of it was one assistant repeating her own description word for word. The assistants that read her offer themselves described it differently.

Praise, or an echo?

Where the positive statements about her offer came from

71% her own words, repeated29% the assistants' own reading
The assistants that read her offer described what it says, not what the marketing says.

Source: study data, August 2026

04What she is doing

Every change made.
Every month measured.

She got a prioritized and clear action plan: strategy, content strategy, positioning strategy, messaging and copy, a technical improvement and schema kit, site navigation restructuring, and more, and had made all the changes. We're now monitoring the results every month, with a new and updated, adjusted action plan per month.

  • Strategy
  • Content strategy
  • Positioning strategy
  • Messaging and copy
  • Technical improvement and schema kit
  • Site navigation

Your brand

What do the assistants say about you?

Every brand's picture is different. A free assessment shows you yours.