For 18 months the dominant client objection to AI visibility work has been the same sentence. "ChatGPT cites us but no one clicks through."
That sentence has a published-data counter as of May 19, 2026.
Profound published research that morning documenting a single OpenAI deployment, shipped May 7 with no announcement, that doubled ChatGPT referrals to brand websites in the monitored-brand basket. Homepage URLs are now embedded inline as clickable hyperlinks inside ChatGPT responses. Every brand mention is now a hyperlink.
The numbers from the Profound monitored basket are the clearest empirical pivot point in AI search since the AI Overviews rollout in May 2024.
Profound — ChatGPT branded-link rollout, May 7, 2026
ChatGPT homepage embeds went from edge case to default in one day
Profound monitored-brand basket. Pre-rollout (April baseline) vs post-rollout (May 7–19, 2026).
Daily OpenAI referral lift by vertical
Largest absolute gainer
Steady lift, regulated buyer queries
Product flows route through shopping surfaces
Three measurements that moved together on May 7
AI mentions are no longer a perception metric. They are a referral channel.
If you run B2B SaaS marketing, the May 7 change is the largest single-day shift in ChatGPT traffic value your category has seen. Daily referrals to monitored B2B SaaS sites are up over 200%. This post breaks down what shipped, why the lift concentrated in B2B SaaS, and what to change in your measurement stack this week.
What OpenAI actually shipped on May 7
OpenAI did not publish a release note. The change surfaced in Profound's monitoring data first, then was documented twelve days later in the Profound research blog.
What changed in the response surface
Before May 7, ChatGPT mentioned brands inline in conversational responses but only embedded brand URLs in a small subset of cases, usually when the user explicitly asked for a link or when ChatGPT was running a web search subflow. URL embedding sat at roughly 4.5% of responses that mentioned a brand.
After May 7, brand homepage URLs became the default inline embed for branded entity mentions. Mention rate of "Notion" did not change. The rate at which "Notion" rendered as notion.com jumped to 20 to 24% of brand-mention responses.
What changed in homepage share of clicks
Homepage URLs were ~4% of the OpenAI referrals Profound tracked into monitored sites in April. By the second week of May they were ~24%. The same homepage that received 1 in 25 ChatGPT click-throughs now receives 1 in 4.
What did not change
Citation rate did not change. ChatGPT did not start citing more brands. ChatGPT started letting users click on the brands it was already mentioning. The improvement is not on the retrieval side. It is on the link-rendering side.
Why B2B SaaS gained the most: five reasons
The lift is not evenly distributed. B2B SaaS daily referrals climbed over 200%. Financial Services climbed 60%. E-commerce and retail were essentially flat. Five structural reasons explain the concentration.
Reason #1: B2B SaaS buyers ask brand-named questions
The B2B SaaS buying funnel runs on named-brand evaluation. "Is Notion better than Coda for product specs." "How does Linear pricing scale." These prompts force ChatGPT to mention specific brand homepages. When mentions become hyperlinks by default, B2B SaaS captures the lift.
E-commerce queries route differently. A user asking "what is the best running shoe" gets routed into ChatGPT shopping flows, not the conversational brand-name flow. The homepage-link path bypasses retail.
Reason #2: ChatGPT mentions correspond to homepage URLs, not product pages
The May 7 change embeds homepage URLs, not deep links. B2B SaaS homepages double as product landing pages and pricing entry points, so a homepage click converts close to a product click. Retail homepages are storefront index pages with low product-discovery intent, so a homepage click has weak commercial value relative to a product-listing click.
Reason #3: B2B buyers already qualify themselves before the click
By the time a B2B buyer asks ChatGPT a branded comparison question, they have already self-qualified on the category. The click is a stage-7 click, not a stage-1 click. Knotch's audience journey tracking inside Conductor's 2026 AEO/GEO benchmarks puts AI-referred visitors at 2x the conversion rate and 1/3 the session count of other sources. The new hyperlinks land qualified intent on the homepage with the engagement state already set high.
Reason #4: Financial Services is the second-strongest gainer for the same reason
Financial Services queries also concentrate around named brands, named products, and named compliance frameworks. The +60% lift reflects the same dynamic as B2B SaaS, attenuated by a smaller share of brand-named queries inside the total prompt mix. We expect this gap to close as Perplexity's equity-page analyst ratings and SEC filing tap-throughs train users to expect branded financial entity navigation across surfaces.
Reason #5: AI mentions had been a brand-recall metric, not a traffic metric, for B2B SaaS
The May 7 change reframes ChatGPT mentions as a decision-stage traffic channel rather than a brand-recall surface. B2B SaaS teams had been measuring share-of-voice in AI responses for 18 months without a meaningful click-stream attached. The new hyperlink layer turns the same upstream work into a directly attributable channel.
Get a ChatGPT brand-link audit for your B2B SaaS
We map every ChatGPT response that mentions your brand, score how often the URL embeds versus how often it should, and ship the structural changes that move the embed rate. The lift compounds against your existing AI visibility program.
Book a discovery callHow AI search measurement breaks under the new layer
A single-metric "citation share" dashboard now systematically understates ChatGPT's value to brand programs. Three measurement gaps opened on May 7.
Gap #1: Click-through rate is now a ChatGPT KPI
Pre-May 7, citation share was the citation surface. There were no clicks to track because there were no links. Post-May 7, two different brands can have the same citation share and very different click-through rates depending on whether ChatGPT renders the brand URL inline. Citation share without embedding rate is incomplete.
Gap #2: Embedding rate varies by query type
Branded-comparison queries embed at the high end of the 20 to 24% range. Open-ended category queries embed at the low end. Reverse-engineering "embed rate" by query type is the new audit layer. Most monitoring tools have not shipped this view yet.
Gap #3: Homepage telemetry sees AI for the first time
Before May 7, the homepage was a low-AI-traffic page. After May 7, the homepage is the top landing page for ChatGPT referrals. Analytics filters that excluded the homepage from AI-traffic reports are now systematically misreporting the channel.
Traditional SEO vs AI-search measurement
The hyperlink layer is the cleanest forcing function yet for retiring SEO-era measurement habits inside AI work.
Traditional SEO asks:
- •How many backlinks does the page have?
- •What is the page's organic ranking on the target keyword?
- •What is the bounce rate on that page?
AI search now asks:
- •How often does ChatGPT mention the brand entity in answers?
- •When ChatGPT mentions the entity, does it embed the brand URL?
- •What is the post-click engagement state of users who arrive from ChatGPT?
The middle question is the one that changed on May 7. Embed rate is now a measurable dependent variable, and it varies brand by brand.
Embed rate is to ChatGPT what featured snippet rate was to Google Search in 2016. A measurable surface that pays out in clicks.
Why this reverses the November 2025 collapse narrative
ChatGPT referral traffic collapsed roughly 52% from July 2025 to November 2025 according to eMarketer's analysis of SimilarWeb data. The cause was OpenAI rolling more answer content inline so users did not need to leave the response. The branded-link rollout reverses that trajectory for the slice of queries that mention brands by name.
This is not a return to zero-click being dead in general. It is a return to zero-click being dead for branded-entity questions, where the homepage now sits one click from the response.
The categories that benefit most are the ones where users ask "what does X do," "is X better than Y," "is X enterprise-ready." Those queries collapse cleanly into a single homepage embed.
Step-by-step actions for B2B SaaS this week
The Profound data is fourteen days old. The actions below capture the lift before the embed-rate variance compresses.
Audit your current embed rate
Pull a sample of 30 to 50 ChatGPT prompts that mention your brand. Score how often yourbrand.com appears as a clickable link inside the response. That is your starting embed rate. The current category benchmark is ~20%. Anything under 10% suggests structural issues with how ChatGPT resolves your brand entity to a URL.
Make your homepage the canonical brand URL
ChatGPT embeds the homepage that Wikipedia, Crunchbase, and your own structured data agree is canonical. If those three sources disagree, embed rate suffers. Reconcile your canonical brand URL across every authoritative property in the next two weeks.
Treat the homepage as your top AI landing page
The homepage moved from 4% to 24% of OpenAI referrals. Update your homepage hero, social proof block, and primary CTA against the assumption that the next visitor arrived from a high-intent ChatGPT branded comparison query. Most B2B SaaS homepages were written for search-led discovery and read as introductions. They should now read as decision-stage destinations.
Add CTR to your AI visibility dashboard
Citation share alone now under-reports the channel. Add an "embedded link rate" column to every brand visibility report, and an "OpenAI homepage referrals" line item to every GA4 traffic split. Both views were optional in April. Both are core in May.
Recheck your robots and crawler permissions
The May 7 change reinforces an old rule. If OAI-SearchBot cannot crawl your site, ChatGPT cannot mention your brand reliably, which means it cannot embed your URL. Verify ChatGPT-User and OAI-SearchBot are allowed in robots.txt this week.
Run your AI referral playbook through Cite
We rebuild B2B SaaS homepages and brand entity pages for the new ChatGPT hyperlink layer. Embed rate audit, canonical brand-URL reconciliation, and homepage decision-stage rewrites in one engagement.
Book a discovery callFAQ
Did OpenAI announce the May 7 branded-link change?
No. OpenAI shipped the change without a press release or changelog entry. Profound surfaced it through monitored-brand telemetry and published the analysis on May 19, twelve days after the deploy. There has been no official OpenAI confirmation as of late May 2026.
How long will the +200% B2B SaaS lift hold?
The structural change is permanent. Brand URLs are now embedded by default. The +200% number is a snapshot of the monitored-brand basket against an April baseline. The absolute lift will normalize as April becomes the prior-period comparison, but the channel itself has been reset to the new higher level.
Does the new embedding affect Perplexity or Google AI Mode?
No. The change is specific to ChatGPT. Perplexity already linked sources aggressively, so the same baseline applies. Google AI Mode renders citations as a panel rather than inline, so its referral profile is unchanged by the May 7 deploy. ChatGPT is now closer to Perplexity on link rendering and further from Google AI Mode.
What if ChatGPT does not embed my brand homepage at all?
Three possible causes. First, your brand entity may not resolve cleanly across Wikipedia, Crunchbase, and your own structured data. Second, the brand may be too new or too small for the ChatGPT entity layer to have a stable URL on file. Third, your robots.txt may block OAI-SearchBot. Run the AI visibility audit to identify which gap applies.
Should we still track citation share separately?
Yes. Citation share remains the upstream signal. The downstream signal is now embed rate and the resulting click-through. Track both. A high-citation, low-embed brand has a fixable structural problem. A low-citation brand still has the original visibility problem to solve first.
The mention-to-click bridge is finally built
For 18 months AI visibility has been a leading indicator with no committed traffic outcome. May 7 changed the contract.
ChatGPT mentions now compound into homepage clicks for branded queries. B2B SaaS is the largest beneficiary by category and the most exposed if the work is not in place. Citation share without embed rate now understates the channel. The homepage is the highest-value AI landing page in a B2B SaaS site, not the lowest.
The teams that audit their embed rate this week and rebuild their homepage for decision-stage traffic will capture the lift while the gap is open.
Continue the brief
Has ChatGPT Search Demand Already Peaked?
ChatGPT search demand hit 14,800/mo Jan-Mar 2026 then dropped 33% in April. AI overview and AI search queries are absorbing the demand.
Does AI Search Traffic Convert 4.4x Better Than SEO?
ConvertMate's 8,000-domain study found AI search traffic converts 4.4x better than organic. Here is what the data actually says and how to act on it.
Is Your AI Visibility Too Reliant on ChatGPT?
Microsoft feared OpenAI dependence per trial testimony this week. Most B2B AI visibility plans share the same single-vendor risk. Here is the fix.
Framework
Learn the CITE framework behind our GEO and AEO work
See how Comprehend, Influence, Track, and Evolve turn AI visibility into an operating system.
Services
Explore our managed GEO services and AEO execution model
Audit, prompt discovery, content execution, and ongoing monitoring tied to AI search outcomes.
GEO Agency
See what a managed GEO agency should actually do
Compare real GEO operating work against generic reporting or tool-only approaches.
Audit
Start with an AI visibility audit before execution
Understand prompt coverage, recommendation gaps, source mix, and where competitors are winning.
