Highlights
- The dataset: This study is an analysis of over 210,000 AI source citations using prompts specific to the design agency industry. It analyzes AI responses from four AI models run in 12 countries over the course of Q2 2026.
- Clutch.co is top-5 for all four models: And the #1 cited domain for Grok (9%), Microsoft Copilot (5%), and Google AI Overview (5%).
- Reddit makes up 7% of ChatGPT's citations: Narrowly ahead of Clutch, giving ChatGPT the most community-driven profile of the four.
- Clutch.co alone makes up 9% of Grok's citations: Its single most-cited source by a wide margin.
- 67% of Google AI Overview's citations are from commercial domains: Versus 46% for Grok, which sends 40% to user-generated content instead.
- 21% average overlap between any two models: Roughly 80% of the top sources one model cites are absent from another's top tier.
Which sources should you target to get cited as a design agency brand?
A shared directory layer sits beneath four otherwise very different top-source lists.
| Model | Top cited sources | Citation character |
|---|---|---|
| Grok | clutch.co (9.3%), designrush.com, sortlist.com, themanifest.com, awwwards.com | Directories + listings |
| ChatGPT | reddit.com (6.7%), clutch.co, agencycluster.com, en.wikipedia.org, blissagency.it | Community + reference + directories |
| Microsoft Copilot | clutch.co (5.4%), themanifest.com, sortlist.com, designrush.com, cieden.com | Directories + agency sites |
| Google AI Overview | clutch.co (5.2%), google.com, blissagency.it, sortlist.com, designrush.com | Directories + search + agency sites |
ChatGPT is the outlier at the top: Reddit edges out Clutch as its most-cited domain, and Wikipedia appears in its top five, neither shows up high for the others. Grok is the most concentrated on directories, with its top four all being agency-listing platforms. A design agency optimizing for one AI model is still absent from roughly 80% of the domains the other models prefer, but getting rated on the major directories is the rare move that pays off across all four.
How have AI source rankings changed over time in the design agency industry?
We split the observation window at its midpoint (early May 2026) and compared the most-cited domains in each half, ranking domains by their within-half citation share rather than raw counts.
The story is a shift in citation character: global directory aggregators (TechBehemoths, agencies.semrush.com, Awwwards, GoodFirms) gave up large amounts of share, while individual agency homepages (madeagency.com.au, deep.co.uk, folklore.mx) and community/search sources (Reddit, Google) climbed. Clutch held #1 in both halves; the anchor directories did not move, but the second tier rotated underneath them. With a single split, we read this as suggestive of a move toward individual agency sites rather than proof of one.
What type of content do AI models cite for design agencies?
We classified every cited domain into content categories. Commercial sites, agency homepages and for-profit directories, dominate the category as a whole, as expected for a B2B service, but the mix differs sharply by model.
ChatGPT is the most diverse: it cites the most reference content by far (8.2%, driven by Wikipedia). Grok is the least commercial of the four, only 45.5% commercial, but a category-leading 40.2% UGC, reflecting how the domain registry classifies the large agency-listing and review directories it floods answers with, plus its Reddit and LinkedIn usage. Google AI Overview and Microsoft Copilot are the most straightforwardly commercial, sending two-thirds or more of citations to agency and directory homepages.
Do AI models cite local-language content for design agencies?
The country drives local sourcing far more than the model does, and even the most local markets stay below half. Using each source's country-code top-level domain (ccTLD) as a proxy for local-language content, we measured the share of citations pointing to a local-ccTLD source in each non-English market.
Across all non-English markets, 33.4% of citations point to a local-ccTLD source, and no market exceeds 42%. This is a more English- and .com-dominated category than most: the big agency directories (clutch.co, sortlist.com, designrush.com) live on generic TLDs, so even a French or German prompt frequently lands on a .com directory rather than a national site. The proxy undercounts genuinely local content published on .com, so treat these figures as a floor. The takeaway holds regardless: in design agencies, global directories crowd out local-language sources more than in most service industries.
Which AI model relies most on local sources for design agencies?
Holding the market constant, the models differ sharply in how local they go. Google AI Overview cites local-domain sources far more often than Grok in almost every language.
Google AI Overview is the most aggressively local where it has coverage, 72% in Dutch, 68% in Italian, reflecting its habit of surfacing national directories and local agency sites. Grok is the clear globalizer: it lands at or below 31% in every language and dips to just 10% in Dutch, because its citations are dominated by big global directories. The country still drives localization more than the model does, but within a country Grok reaches for global aggregators while Google AI Overview reaches for local ones. (Google AI Overview produced no French citations in scope.)
Should design agencies optimize for each AI model separately?
Not much, with one exception. When the four models answer the same design-agency questions, their top-cited domains barely converge. We compared the 20 most-cited domains for each model and measured how much each pair holds in common.
That overlap is higher than in most service categories, and the reason is structural: a small set of B2B agency directories, Clutch, Sortlist, DesignRush, The Manifest, anchors every model's list. The spread between the highest- and lowest-overlapping pairs is not statistically distinguishable at this scale, so we read all four models as similarly low-overlap once you look past the directories.
How many sources does each AI model cite per answer for design agencies?
The models differ enormously in how many sources they cite per answer. Grok is by far the most source-hungry; ChatGPT the most economical.
Grok cites roughly 4.5x as many sources per response as ChatGPT. For an agency, that cuts two ways: Grok offers far more citation slots to compete for, and many of them are directory listings, which are reachable by getting rated, but each individual citation carries proportionally less weight. ChatGPT cites few sources, so earning one of its six is both harder and more valuable.
Context
This analysis draws from Temso's AI visibility monitoring platform, which tracks how brands appear in AI model responses across ChatGPT, Microsoft Copilot, xAI's Grok, and Google AI Overview. It covers how these four systems cited web sources when answering design-agency questions between March 11 and June 29, 2026, across 12 markets, Argentina, Australia, Canada, France, Germany, Italy, Mexico, the Netherlands, Spain, Sweden, the United Kingdom, and the United States, and seven languages, comprising 21,128 responses and over 210,000 cited source citations.
The findings describe which sources AI models cite, not which agencies or services are best. A high citation share means a domain is frequently surfaced by a model, not that it is authoritative or accurate. Citation behavior is also a moving target: models update their retrieval and ranking continuously, so a snapshot like this captures a period, not a permanent state.
Methodology
How we measured this
We analyzed every AI response and its cited sources for the design-agency category over the observation window: 21,128 responses and over 210,000 cited source citations. A citation is a web source a model explicitly surfaced in support of an answer; we count only sources flagged as cited. For each model we ranked domains by cited-source count and reported the top five with each domain's share of that model's citations. Cross-model overlap was measured on each model pair's top-20 most-cited domains. Domain content categories (Commercial, UGC, Editorial, Reference, Institutional) were assigned from a global domain registry. Temporal analysis split the period at its midpoint and compared domain rankings by within-half citation share.
Localization is measured by a country-code-domain (ccTLD) proxy, the source records carry no detected content language, so those figures are directional and exclude English-language prompts, which sit overwhelmingly on generic domains. The proxy undercounts local content published on .com domains, which are especially common among design-agency directories, so the localization rates are conservative. The temporal split is a single midpoint cut and model coverage varied across the observation window, so the movement findings are suggestive rather than conclusive.
Frequently asked questions
Do different AI models cite the same sources for design-agency questions?
Mostly no, with one big exception. Any two models share only about 21% of their top-20 most-cited domains on average, so roughly 80% of the sources one model relies on are absent from another's top tier. The exception is the major B2B directories, Clutch, Sortlist, DesignRush, The Manifest, which anchor all four models' lists.
Which single source matters most across all the AI models?
Clutch.co, without question. It is in the top five for every model and the most-cited domain for Grok, Microsoft Copilot, and Google AI Overview. If there is one must-win source in this category, it is Clutch, followed by the other big agency directories.
Which sources does each AI model favor beyond the directories?
They diverge sharply. ChatGPT leans on community and reference, Reddit is its #1 at 6.7%, and Wikipedia appears in its top five. Grok floods answers with directory and listing platforms at high volume. Microsoft Copilot and Google AI Overview send most of their citations to commercial agency and directory homepages, with Google AI Overview the most likely to surface a local agency's own site.
Does language matter for getting cited by AI in design?
Yes, but less than in most industries. In the most local markets, Sweden and Italy, about 42% of citations go to local-ccTLD sources; in Argentina and Spain, only about 21–27% do. Because the dominant directories live on .com domains, even local-language prompts often land on global sites, so these figures understate genuinely local content.
Which AI model is most likely to cite a local agency's own website?
Google AI Overview. It reaches 68–72% local-ccTLD rates in Italian and Dutch and surfaces individual agency sites more than the directory-heavy models. Grok is the opposite, it stays on global aggregators and dips to just 10% in Dutch.
Which AI model cites the most sources, and does that help my agency?
Grok, by a wide margin, about 28.3 cited sources per response, roughly 4.5 times ChatGPT's 6.2. More sources mean more slots to compete for in Grok, and many of those slots are directory listings you can earn by getting rated. ChatGPT's economy of citations makes each of its few slots harder to win and more valuable.
Did the sources AI models cite change over the period?
They shifted underneath the directories. Clutch held its #1 position throughout, but in the second half global listing aggregators like TechBehemoths and agencies.semrush.com lost large amounts of share while individual agency sites (madeagency.com.au, deep.co.uk, folklore.mx) and community/search sources (Reddit, Google) climbed. We read this as a suggestive trend toward individual agency websites, not a proven one, given the single time split.
What should a design agency do with these findings?
Win the directories first, then optimize per model. Getting listed and well-rated on Clutch, Sortlist, DesignRush, and The Manifest is the one move that pays off across all four AI systems. After that, treat each model as a separate channel: community presence (Reddit, Wikipedia) helps with ChatGPT, sheer directory coverage helps with Grok, and a strong, locally hosted agency website helps with Google AI Overview and Microsoft Copilot.
