Search engine optimization (SEO) is not dead, but the click economics behind it have shifted. Pew Research found that when an AI summary appears, only 8 percent of visits produce a click on a traditional result, against 15 percent when no summary appears. Fewer people click, so San Antonio businesses need to be the source the answer is built from rather than the tenth blue link.
Every few years somebody announces that SEO has died, and this time the obituary has better evidence behind it than usual. AI answers now sit above the results on many searches, and they are absorbing clicks that used to reach websites.
That is a genuine change, not a scare headline. At Texas Web Design, we watch this play out across client accounts every month, and the pattern is a shift in which searches pay rather than an end to search itself.
Here is what actually shifted, what still works, and where a local business should put its effort now. Contact us today if you want a straight read on how much of your own traffic is exposed to this.
What’s Actually Changing in SEO?
The mechanics of ranking have not been rewritten. What changed is what sits between a ranking and a visitor.
Answers are appearing above the results
Google now generates a written answer at the top of many searches, assembled from sources it selects. The user reads it, and often that is the whole interaction.
Pew Research studied 68,879 unique Google searches from 900 US adults and found that about 18 percent produced an AI summary in March 2025. That is close to one search in five, which puts the exposure well past a niche problem confined to a few query types.
On client accounts we check this first, because the share of a site’s queries that trigger an answer decides how much of a decline is even fixable. A business whose traffic comes mostly from definitions sits in a very different position than one ranking for hiring terms.
Clicks are down, and the numbers are specific
Pew found that when an AI summary was present, 8 percent of visits produced a click on a traditional search result, compared with 15 percent when no summary appeared. This is where the argument stops being theoretical.
Ahrefs looked at it from the ranking side across 300,000 keywords and found that the presence of an AI Overview correlated with a 34.5 percent lower average click-through rate for the top-ranking page. Two different methods, same direction.
Neither study says search stopped sending people to websites. Both say the same impression count now buys fewer visits than it used to, which is a change in price rather than a death notice.
Citation replaced position as the thing worth winning
Most coverage skips the next number. Pew also found that only 1 percent of visits produced a click on a link inside the AI summary itself.
So being cited does not reliably send traffic either. What it does is put your business name in front of someone at the moment they are deciding, which is a branding win rather than a click win, and it needs to be valued accordingly.
We treat it the way a business would treat a mention on a local news segment. It is worth having and worth tracking, but it will not show up as a session in an analytics report.
Okay, So What Do I Actually Do?
The response is not a new discipline. It is a reweighting of things that were already on the list.
Go after searches that still convert
Not every search is equally exposed. Informational queries that begin with “what is” or “how does” tend to be the ones answered above the fold, while searches with buying intent still send people to websites because a summary cannot book an appointment.
For a San Antonio service business, that means the pages worth defending are the ones about hiring, pricing, comparing, and booking. Those are also the pages that were always worth the most.
The practical move is to sort your top queries by intent before touching anything else. Losing a batch of informational impressions matters far less than losing one page that ranked for a hiring term.
Write so a machine can quote you
If an answer engine is assembling a response from sources, the practical goal is to be the source it lifts from. That rewards a specific writing shape: state the answer plainly in the first two sentences of a section, then support it.
Vague opening paragraphs that circle the topic before saying anything are now expensive. They were always weak, but they used to still rank.
The test we use is simple. Pull any single paragraph off the page, read it with no surrounding context, and see whether it still answers a question a buyer would actually ask.
Make the local signals unambiguous
Map results and near-me searches appear less disrupted by generated answers, because the user wants a business, not an explanation. Local visibility looks like the steadiest part of the mix right now.
Consistent business details, real review activity, and location pages that say something true about each area continue to work much as they did.
It is the least glamorous part of the work and the part we would cut last. A generated answer can describe a service category in general terms, but the map pack still drives who gets the call in Alamo Heights or Boerne.
The Clicks You Lost Were Not the Clicks You Wanted (Mostly)
There is a detail buried in the click numbers worth sitting with. The searches most likely to trigger a generated answer tend to be the ones with the least commercial intent.
Somebody asking what a service generally costs, or how a process works, was rarely close to calling anyone. Losing that visit usually costs an impression and a traffic number, not a customer. The line on the chart looks worse than the month on the calendar.
What this does break is any strategy built on volume: publishing broad informational content to accumulate sessions and hoping some fraction converts later. That model has weakened, and the businesses that leaned on it tend to feel this hardest.
What About Small Businesses in San Antonio?
Local businesses are, oddly, better positioned than most publishers. A generated answer can summarize what web design costs, but it cannot show up, look at your site, and quote the job.
The searches that fill a local calendar are still transactional and still local. Somebody looking for a web designer in Stone Oak or a contractor in New Braunfels is looking for a business to contact, and that intent has not moved.
The real risk for a small business is different from the one in the headlines. It is having a site that no engine can read cleanly, so it is never the source an answer gets built from, and never the result someone clicks either.
That failure mode is quiet and common. A site can sit for years with a setting hiding its blog from search, or with every page sharing one description, and nothing about the situation announces itself in a monthly report.
Bottom Line
SEO is not dead; it has just stopped rewarding volume and started rewarding being the clearest, most quotable source on the questions your customers actually ask. At Texas Web Design, we build search engine optimization around the searches that end in a phone call rather than the ones that end in a summary.
If your traffic has softened and nobody has told you which part of it is exposed, that is worth knowing before you change anything. Call us today for a straight read on where your visibility actually stands.
Frequently Asked Questions About SEO and AI Search
Is SEO still worth it in an era of AI answers?
Yes, though the return now comes from different searches. Queries with buying intent still send people to websites, while general informational queries are increasingly answered on the results page. Businesses that shift effort toward hiring, pricing, and comparison pages are better placed to keep search delivering customers.
How much traffic do AI Overviews actually take?
Pew Research found that when an AI summary appeared, 8 percent of visits produced a click on a traditional result, against 15 percent without one. Ahrefs measured a 34.5 percent lower click-through rate for the top-ranking page across 300,000 keywords carrying an AI Overview.
Does being cited in an AI answer send traffic?
Rarely. Pew found only 1 percent of visits produced a click on a link inside the AI summary itself. Citation works as brand exposure at the moment of decision rather than as a traffic source, so it should be valued as visibility instead of sessions.
Which searches are least affected by AI answers?
Transactional and local searches. Someone looking for a service provider near them wants a business to contact, and a generated summary cannot quote a job or book an appointment. Map results and near-me queries remain the steadiest part of local search, which is one reason service businesses appear to have felt this shift less sharply than publishers.
Should I stop writing blog content?
Not stop, but change what you write. Broad informational posts published for volume have weakened, while content that answers a specific question a buyer asks before hiring can still earn both citations and clicks. Fewer, sharper pages now tend to outperform a high publishing count.
What is the difference between SEO and optimizing for AI search?
Less than the marketing suggests. Both reward clean technical structure, clear answers stated early, and consistent business information. Optimizing for AI answers mostly means writing so a passage can be lifted and understood without surrounding context, which is a writing discipline rather than a separate service.
Has Google confirmed that AI Overviews reduce clicks?
No. In August 2025 Google said total organic click volume from Search to websites had been relatively stable year over year and that average click quality had increased, a claim about Search overall rather than AI Overviews specifically, published without supporting figures. Independent studies from Pew Research and Ahrefs measured lower click-through rates when an AI Overview appeared.
How do I know if my site is losing traffic to AI answers?
Look for impressions holding steady while clicks decline in Google Search Console. That pattern points to your pages still appearing but fewer people clicking through, which is a common fingerprint of an answer being read on the results page instead. Check it query by query rather than sitewide.
