Something strange is happening on a lot of company websites. Rankings look stable. The pages that used to bring in leads are still sitting on page one. But traffic keeps sliding, and the contact form is quieter than it was a year ago.
If that sounds familiar, the problem probably is not your website. What changed is search itself. It used to be a place people passed through on the way to your site. Now it is increasingly a place where they get the answer and stop.
Here is what actually changed, what it means for a business that sells to other businesses, and what to do about it.
The search results page now answers the question for you
Google’s AI Overviews, the summary block that appears above the normal results, are no longer occasional. Conductor’s 2026 benchmark study analyzed 21.9 million Google searches and found that 25.11% of them triggered an AI Overview. Roughly one in four.
When that summary shows up, people click less. The Pew Research Center tracked the actual browsing behavior of 900 US adults and found that when an AI summary appeared, users clicked a traditional search result 8% of the time, compared with 15% when no summary was present. Only 1% clicked a link inside the summary itself. Twenty-six percent ended their browsing session right there, versus 16% on a normal results page.
Google has publicly disputed that study, arguing the query set is not representative of real search traffic. That is a fair caveat. But the direction of the finding matches what most marketing teams are watching in their own analytics, and it matches the broader numbers: B2B data compiled by Cognism puts organic sessions down 33.6% year over year and organic clicks down 23%.
The second shift matters just as much. A growing share of business research never reaches Google at all. Buyers open ChatGPT, Gemini, Perplexity, or Copilot and ask there. Conductor found that ChatGPT alone accounts for 87.4% of AI referral traffic, and that information technology companies are seeing the highest share of visits arriving from AI tools.
Your rankings can be perfectly healthy while your traffic dies
This is the part that confuses business owners, and it is worth being precise about it.
Ranking number one means you are the first blue link. It does not mean you are the first thing the buyer reads. If an AI summary sits above you and answers the question adequately, your position is intact and your visit is gone. The metric your agency reports and the metric your bank account cares about have quietly stopped pointing in the same direction.
So a monthly SEO report showing stable or improving rankings is no longer evidence that anything is working. It is evidence that one input is holding. That is not the same thing.
The goal has shifted from getting clicked to getting cited
When someone asks an AI assistant “who handles outsourced drafting for engineering firms” or “what does outsourced bookkeeping actually cost,” the model produces an answer built from sources it trusts. Some of those sources get named. Most do not.
Being one of the named sources is the new version of ranking. It does three things that a click never did:
- It reaches the buyer earlier, while they are still deciding what kind of solution they need, not which vendor to pick.
- It arrives with borrowed credibility, because the assistant is effectively vouching for you.
- It shapes the shortlist before you ever know the buyer exists.
The trade-off is that you often will not see the visit. Someone can read about your company, form an opinion, and never appear in your analytics until they type your company name directly. Which is why branded search volume is becoming one of the more honest health metrics a business can track.
What gets a company mentioned in AI answers
Nobody outside the model providers knows the exact mechanics, and anyone who claims otherwise is guessing. But the patterns in what does get cited are consistent enough to act on.
1. Answer the question in the first two sentences
AI systems extract answers. A page that opens with three paragraphs of throat-clearing before defining the term gives them nothing usable. Lead with the direct answer, then expand. This also happens to be better for human readers, which is usually a sign you are doing it right.
2. Be specific enough to be quotable
“Outsourcing reduces costs” is unquotable. “An outsourced AEC drafting team typically covers CAD production, Revit modeling, and as-built documentation, with turnaround measured in days rather than weeks” is quotable, because it contains actual information. Vague copy does not get cited. It gets skipped.
3. Get mentioned on sites you do not own
AI answers pull heavily from third-party sources: industry publications, review platforms, forums, directories, and comparison sites. Your own website is one input among many, and not the most trusted one. A single mention in a respected industry outlet can influence more AI answers than a year of blog posts on your own domain.
4. Keep your facts consistent everywhere
If your website says you serve architecture and engineering firms, your LinkedIn says construction technology, and a directory listing from 2022 says general BPO, you have given the model three different companies to reconcile. It will pick one, or trust none of them. Consistency across every place your business is described is unglamorous work that directly affects whether you get named.
5. Cover the whole question, not just the keyword
A buyer asking about outsourced accounting also wants to know what it costs, what stays in-house, how security works, and what happens if it goes wrong. A page that answers the full set is more useful to an AI assembling an answer than five thin pages that each answer one slice.
6. Publish things only you can publish
Your own data, your own process, your own results. A summary of publicly available information adds nothing an AI model does not already have. An account of how your team actually handles a workflow, with real numbers, is genuinely new information. That is what gets pulled.
What to stop doing
Three habits are actively working against most companies right now.
- Judging content by traffic alone.
If the only number you look at is sessions, you will cut the content that is quietly shaping buying decisions off-site, because it will look like it failed.
- Publishing for volume.
Twelve generic posts a month built around keyword targets produced modest results in 2019 and produce almost nothing now. Four genuinely useful pieces will outperform them.
- Gating everything.
If your best material sits behind a form, AI systems cannot read it, cannot cite it, and will name a competitor who published openly. Gate the tools and templates. Do not gate the thinking.
A twenty-minute check you can run today
Before you plan a response, find out where you actually stand.
- Open ChatGPT, Gemini, and Perplexity. Ask each one the three questions your best customers ask before they buy. Not “who is the best provider,” but the real questions, like “how much does outsourced bookkeeping cost for a 40-person company” or “what work can an architecture firm outsource.”
- Write down which companies get named. If yours does not appear in any of the nine answers, that is your baseline.
- Ask each tool directly what it knows about your company. Check whether the description is accurate, current, and matches how you describe yourselves.
- In Google Analytics, filter referral traffic for ChatGPT, Perplexity AI, and Gemini. Whatever the number is, watch whether it grows.
That exercise takes less than half an hour and tells you more about your visibility than most monthly SEO reports.
The most direct fix, and sometimes the correct one. If you are trying to hire a senior accountant at a band you set in 2022, the market has moved past you.
The real obstacle is not knowing what to do. It is having someone to do it.
Most of what is written above is not conceptually hard—it is just relentless: auditing business descriptions across dozens of sites, rewriting pages to lead with answers, producing citable content, pursuing industry mentions, and re-checking visibility monthly as answers shift.
This continuous workload often falls on small marketing teams already stretched thin by events, website management, social media, and sales support. Asking them to absorb an entirely new discipline is why the work quietly gets postponed.
You have three realistic options:
- Outsource dedicated support: A specialist executes your process continuously at a viable cost
- Hire internally: Slow and expensive for an evolving function.
- Retain an agency: Great for high-level strategy, but less suited for daily execution.
At Guided Outsourcing, this is the gap our marketing teams are built to close. Our Philippines-based specialists work as an extension of your in-house team on content production, SEO and AI search visibility, website updates, and the ongoing maintenance work that gets postponed when everyone is busy. You keep the strategy. We supply the hands to execute it consistently.
FAQs
- Is SEO dead?
No, and be skeptical of anyone who says it is. AI systems draw on the same underlying signals search engines use, so technically sound, well-structured, authoritative content still matters. What has changed is the measure of success. Position on a results page is no longer a reliable proxy for visibility.
- How is this different from the SEO we already pay for?
Traditional SEO optimizes for a ranking position on a page you control. This work optimizes for being accurately described across sources you mostly do not control. There is real overlap, but the off-site and consistency work tends to sit outside a standard SEO retainer. Worth asking your current provider directly what they are doing about it.
- How long before we see results?
Expect a slower and less linear curve than traditional SEO. AI models update on their own schedules, and the same question can return different answers week to week. Treat it as a three- to six-month exercise, measured by whether you appear in relevant answers at all, not by a traffic line.
- Does this apply to a small B2B company, or only to large brands?
It often matters more for smaller companies. Large brands are already widely documented across the web, so AI systems have plenty to draw on. A smaller firm with a thin footprint is easy to overlook entirely, which means the gap between being cited and being invisible is wider.
Where to start
Run the twenty-minute check this week. If your company does not appear in the answers your buyers are reading, you now have a specific, measurable problem rather than a vague sense that traffic is down.
Fixing it is not complicated. It is just more work than most in-house teams have room for, and it does not pause while you figure out the staffing.
If you would like to talk through what it would take to run this properly alongside everything else your team is already doing, we are happy to have that conversation.
