How to Get Found and Booked as a Tour & Activity Operator in an AI World
We tracked 530 prompts across ChatGPT, Gemini, Google AI Overviews, Google AI Mode, and Perplexity for more than 6 months and analyzed 406,373 AI responses. Here’s what the data says about your organic traffic, AI visibility, and what you need to do with your website now.
Here are 5 quick key highlights if you don't read the complete report
This is the short version of 530 tracked questions and 406,373 AI
responses across ChatGPT, Gemini, Google AI Overviews, Google AI Mode,
and Perplexity. The conclusion is not "keep doing good marketing." The
old SEO playbook no longer covers how customers discover, compare, and
choose an experience. Your business now has to be known across the web
and knowable to machines and customers.
Discovery clicks have already collapsed.
Across the 52 operator websites we tracked, total Google
search visits per site fell roughly 41% in 2 years. Clicks from people searching for a business
by name rose 12% year over year while top-of-funnel discovery clicks
fell 48%. Most OTAs have lost more than half their organic search traffic. Demand did not disappear. More of the research now happens
before the customer reaches your website, so do not wait for your own
traffic to collapse before adapting.
A #1 Google ranking does not secure a place in an AI
answer. Pages in Google's top 3 were roughly 5x more likely to
be cited than pages ranking 11-20, so rankings still matter. But 83% of
the exact pages AI cited did not appear in the top 20 for the broader
Google search we tested. And when Google's AI mentioned an operator, it
displayed that operator's Google Business Profile 96% of the time.
Treat your ranking and your Business Profile as separate assets, and
stop treating first place in the blue links as the finish line. See why ranking #1 is no longer enough.
Be known: build a brand AI can recognize and
repeat. The useful mentions
connect your company and named products to a place, an activity, a
customer, and a reason to choose you. Operators with more independent mentions appeared in 43% of tracked answers, versus 30% for
those with fewer mentions. See what the mention data showed. Use the same distinct brand and product names
everywhere. Give DMOs, DMCs, local bloggers, travel publications,
news outlets, partners, and customer communities specific details worth
repeating: the captain's history, a signature route, specialized
equipment, a named experience, or another fact a competitor cannot
claim. In AI discovery, differentiation is the
context that helps a machine understand where you belong and when to
recommend you.
Be knowable: answer the question before AI has to ask
it. A basic product description is no longer enough. AI needs enough
current information to judge fit for every relevant traveler, group,
occasion, ability, schedule, transportation need, weather condition, and more.
Operators with an AI-optimized page appeared roughly 120% more in AI results than those without.
Each page combines core business and product facts
with up to 300 questions and answers drawn from the operator's products
and website, patterns in customer reviews, and local destination
knowledge. The goal is to make every decision-changing fact retrievable
in the exact context a customer may ask about.
Your website has a new job: become the deepest source on
your own business. You need more content: the most complete, current record of what can be booked, who each
experience fits, why it is different, and what could stop someone from
booking.
Backlink counts, more blog posts, and more schema types did not predict greater AI visibility. Build depth around each experience:
prices, schedules, meeting points, inclusions, exclusions, policies,
accessibility, weather, transportation, group types, and the real
differences between options. The new unit of content is a customer decision that both AI and your
website can answer accurately. See how your website's job has changed.
The rest of the report shows the evidence and limits behind this
model. But the operating principle is simple: be known enough to enter
the answer, and knowable enough to win the decision. See how the two work together.
Watch a preview of this
research
This preview session, recorded live for the Tourpreneur community, walks through a few of the data points as a first introduction to the research.
Letter from the CEO
At
the end of 2025, I pulled our profitable 6-figure SEO package off the
shelf because the 'old' way of doing SEO no longer works.
Not because customers stopped buying it.
Not because it lost money.
But because traditional SEO can no longer deliver the outsized ROI
that it once did.
For years, Gondola offered our Foundational SEO Package with a 100%
money-back guarantee. If it didn't generate more bookings across 12
months, we would give you your money back.
That was not a clever marketing trick. We knew with 99% certainty
that it would generate more bookings for you.
And it consistently did its job. We never had to issue a refund for
that service.
Unfortunately, over the last 2 years, the relationship between
ranking, traffic, and bookings broke.
You would still rank. Sometimes you would rank even better than
before, but a #1 ranking lost its associated traffic and bookings.
Concerned customers started calling us.
"Why is my traffic down?"
"Why is my website not performing like it did before?"
"Why are bookings down if I still show up #1 on Google?"
I kept hearing versions of that conversation, and I did not have a
good answer for them.
That was a problem.
I created Gondola for the sole purpose of getting you more direct
bookings. If I could no longer deliver on that promise, we weren't doing
a good job as a company.
So we stopped what we were doing and invested over 18 months into a
complete product overhaul.
That's why this research exists... and why I've put on 20 pounds.
🙈
Back when I ran my own sailing charter business, I would have climbed
the tallest palm tree on the island for research like this... anything
to help me get found, compete, and grow online.
Today, Gondola works with hundreds of tour and activity operators
across many markets.
We host your websites, see how customers reach you, and have access
to first party data that give us unique insight into the market that
literally no one else has.
I would have climbed the tallest palm tree on the island if I could
have gotten access to this kind of research when I was operating.
We tracked 530 prompts about 34 real businesses across ChatGPT,
Gemini, Google AI Overviews, Google AI Mode, and Perplexity. From
December 1, 2025 through June 9, 2026, Peec recorded 406,373 AI
responses. The frozen citation source contains 3,388,833 citations
across those 530 prompt IDs. We analyzed the pages AI cited, the
businesses it mentioned, the Google clicks operators gained and lost,
the outside websites that discussed them, and what happened when their
own websites changed.
This is not a claim that SEO is dead (although that would be some
good click-bait material).
Our own data shows that ranking high still helps in its own ways.
It is also not an argument that websites no longer matter.
It turns out that websites are actually becoming even more important,
but in a different way.
This is an attempt to answer a more useful question:
How does an independent tour or activity operator get found and
booked in today's rapidly changing AI environment?
This report includes many of our learnings from over 6 months of
dedicated research and 5 years of building websites for hundreds of tour
and activity operators.
We are publishing it because you deserve to see what we're seeing.
Because our customers need us to adapt. And because Gondola needs to
build around the way people discover and book now, not the way they did
5 years ago.
I have opinions about what the evidence means. I have tried to label
those opinions honestly and show the underlying data that supports
them.
You should challenge both.
Vitaliy Levit Founder & CEO, Gondola
Part 2 of 5
What has changed
Your customers still research their options. AI now does much more of that work for them.
Your
customers are asking AI which businesses to choose
The person looking for your business may still start in Google. That
part has not disappeared.
What changed is the work that happens next.
To keep this report concrete, we are going to follow 1 real example.
Morgan Hill Wine Trolley is 1 of the operators in our study, and it
appears in roughly 61% of the AI answers we tracked for its business. 1
of the questions came from a group planning a birthday:
AI
Prompt sent to AI
My friends and I want to celebrate a birthday with wine tasting near San Jose. What are the best tours in Morgan Hill?
A few years ago, that question would have created a pile of work for
the customer. Search Google. Open a dozen tabs. Read local guides. Check
Yelp and Tripadvisor. Visit operator websites. Compare public and
private tours, prices, pickup points, winery stops, group policies, and
availability.
Now the customer can ask the whole question at once. ChatGPT, Gemini,
Perplexity, Google AI Overviews, or Google AI Mode can search, read, and
compare the options. The customer gets a short list with reasons and
sources, then decides what to check or book.
The research still happens. The customer just does less of it by
hand.
The research still happens. The customer just does less of it by hand.
1 question
can replace a whole afternoon of clicking
Google explains that AI Mode can split 1 question into smaller
searches, gather information from different websites, and assemble the
findings into 1 response. Google calls this “query fan out.” Google
explains how AI search gathers information.
The customer no longer needs to search “wine tours near San Jose,”
then “Morgan Hill wine trolley,” then “private wine tours,” and then
open 8 websites to compare the details. 1 question can trigger all of
that research.
The answer will not always be good. AI can miss businesses, mix up
details, and rely on information that is old or wrong. But the basic
capability is already here. The customer asks once. The machine does the
searching, reading, and sorting.
And this can happen inside the same Google your customers already
use. They do not have to become ChatGPT users for their research process
to change.
Different
questions pull from different parts of the web
AI answers usually show the pages they used as sources. We call those
pages citations.
The birthday question above is already close to a booking. That
matters. The 3 pages cited most often for that question all belong to
Morgan Hill Wine Trolley:
The exact order can change as AI produces new answers.
The larger pattern is what matters. Broad questions
lean on destination organizations, government pages, guides, and other
outside sources.
As the question becomes more commercial and specific, the operator's
own pages become more useful.
Same business. Different question. Different sources.
AI can assemble the old pile of browser tabs before the customer
visits a single operator website. Which tabs it chooses depends on the
job the customer is asking it to do.
Your
customers aren't clicking through to your website like they used to
If direct bookings, website visits, or the number of customers saying
“I found you on Google” feel softer than they used to, you're not
alone.
Maybe your ranking slipped. Maybe a new competitor entered the
market. Maybe demand changed. Maybe the economy, weather, or another
factor outside your control changed with it. We cannot explain what
happened to every individual business.
However, Gondola hosts and measures hundreds of operator websites...
and when we zoom out across that network, the overall direction is
clear.
Organic search traffic is down well over 40% in the last 2 years.
Google search visits per operator website, comparing the same
months in 2024, 2025, and 2026. By April 2026, Google was sending about
half as many visits as it did 2 years earlier.
The large experience marketplaces move in the same direction. Viator,
Klook, Tripadvisor, and Tiqets all sit well below their 2024 organic
search peaks.
The major experience marketplaces have lost a large share of
their Google traffic too. Viator is down 57% from its peak, Klook 48%,
Tripadvisor 27%, and Tiqets 77%. These are modeled Ahrefs estimates, so
read the direction rather than the exact level.
The
biggest drop in clicks comes from top of funnel discovery terms
People who search for your exact business name still click about 12%
more year over year.
People who are searching to learn about things to do at the top of
funnel and discovery searches fell 48%. These include terms like “best
wine tasting tours in Morgan Hill” or “things to do near San Jose.”
Searches from people who already know a business name grew.
Searches from people still deciding what to do or which business to
choose fell sharply.
Think about those searches through the Morgan Hill operator we are
following:
“Things to do near San Jose” comes from someone who may not know
they want a wine tour at all.
“Best wine tours near San Jose” comes from someone who knows the
activity but not the operator.
“Best private wine tours in Morgan Hill” comes from someone who has
narrowed the area and the kind of experience but not the business.
“Morgan Hill Wine Trolley tours” comes from someone who already
knows the business.
“Morgan Hill Wine Trolley private tours” comes from someone who
knows the name and wants a specific experience.
The further that customer sits from knowing your name, the more
likely the click has disappeared.
That makes sense when you realize that many AI recommendations do not
send the customer directly to an operator website. The customer may hear
your name from AI, then search for you by name when they are ready to
look closer.
This is the point you need to pay attention to. The customer who
already knows you can still get back to you. The harder problem is
getting introduced to the customer who does not.
AI
is contributing to this shift, but it's not solely responsible
We cannot look at every missing Google click and say AI took it. The
data does not support that claim.
Google results now include Maps, ads, booking sites, videos, forums,
product carousels, and AI answers. Demand changes. Competitors improve.
Rankings move. Algorithms change. Any of those things can affect
traffic.
What we can say is simpler. A customer can now complete much more of
the research without visiting an operator website. AI did not cause
every lost click, but it makes the website visit less necessary during
research.
Organic
clicks to your website show up in your Google Analytics. AI
recommendations do not.
If someone clicks from Google to your website, your analytics records
the visit. If an AI system reads your website, uses the information in
its answer, and the customer never clicks, your analytics records
nothing.
Both events can influence a booking. Only 1 looks like traffic.
Google search visits per operator website keep falling while
direct visits from AI tools rise from almost nothing to about 8 a month.
The AI line is still tiny next to Google. It also measures only the
people who click, not everyone who saw an AI recommendation.
This is why direct AI referrals create such a misleading picture. The
visible traffic remains tiny. The invisible research can be much
larger.
We cannot see every choice a customer makes inside an AI answer. We
can measure whether a business appears, which pages AI cites, and which
sources repeatedly support the recommendation. That is the evidence
behind the rest of this report.
Analysis reference: traffic change
2-year cohort: 52 operator websites with
uninterrupted analytics in the same comparison months of 2024, 2025, and
2026. Mean Google search visits per site fell about 41%.
Search Console cohort: 131 sites with comparable
data from January through April in both 2025 and 2026. Total clicks fell
25%.
12-month site distribution: 126 active sites with
at least 100 Google visits in the earlier period, comparing May 2025
through April 2026 with the prior 12 months. Retired sites and sites
with no recent data were excluded.
3-year sensitivity cohort: 20 sites with complete
data from May 2023 through April 2026. Mean search visits per site ended
35% below the starting period.
Search-type cohort: 94 sites with 16 complete
months of Search Console data. The classification uses the top queries
Google reported, not every long-tail query, so the contrast between
branded and discovery searches is more reliable than either exact
decline estimate.
Interpretation limit: these observational
comparisons measure changes in clicks. They do not isolate AI as the
cause.
Your website's job
description has changed
The website used to handle almost every stage. It helped the customer
discover the business, understand the experience, compare options, build
trust, and book.
Now AI may handle part of the discovery and comparison before the
customer arrives. The website gets the customer later, when they want to
check the recommendation, understand the details, see whether the
business feels legitimate, check availability, and book.
That does not make the website less important. It gives the website 2
clearer jobs.
First, give AI accurate information it can use. Second, give the
customer a trustworthy place to verify the recommendation and act.
But the website cannot do either job if the business never makes the
short list.
So that is where we go next: what gets your business found when AI
decides which operators belong in the answer?
Part 3 of 5
Known: How your business gets found
Ranking high improves your odds of being cited. But most AI sources come from elsewhere, and independent brand mentions explain the visibility gap better than the metrics operators usually track.
Ranking #1 on
Google search isn't enough anymore.
Your website is ranking #1 for an important keyword.
That's awesome!
Ranking first still has real value.
It can put your business in front of more people and send qualified
visitors to your website. It also improves your odds of appearing as a
source in an AI answer.
In our research, pages in Google's top 3 are about 5x more likely to
be cited than pages ranking 11-20.
But here's the part that can make you feel a little crazy...
You can rank first on Google, ask ChatGPT or Google AI Mode a closely
related question, and still find no trace of your business in the
answer.
How is it possible that you rank at the top of Google search but
you're not included at all in the AI recommendation if the AI agents are
doing the searching?
Put simply: AI uses the top results sparingly and includes many other
sources in its answer.
Ranking
high on Google improves your shot at being cited
A page in Google's top 3 is about 5x more likely to appear among the
leading AI sources than a page ranking 11-20, but it is still only a
fraction of the total sources AI references.
Pages in Google's top 3 appear among the leading AI sources about 26%
of the time. That falls to 14% for positions 4-10, then to just 5% for
positions 11-20.
Pages in Google's top 3 appear among the leading AI sources about
26% of the time, compared with 14% for positions 4-10 and 5% for
positions 11-20. A top-3 ranking gives a page roughly 5x the citation
probability of a page ranking 11-20.
Ranking well helps, but does not guarantee your place in the AI
response.
83%
of the pages AI cited did not appear in Google's top 20 for the broader
search
But here's the rub...
Only 17% of the exact pages AI cited appeared in the top 20 for the
broader Google search we tested.
The other 83% did not show up in Google search top 20 at all.
Only 17% of exact AI-cited pages appeared in the top 20 for the
broader Google proxy search. Another 22% came from a domain with a
different page in the top 20. The remaining 61% came from domains absent
from that top 20.
This does not mean Google had never indexed or ranked the other
pages. It means the exact cited page did not appear in the broader
comparison search. AI can reach a page through narrower searches and
other retrieval routes that this benchmark does not observe.
Analysis reference: Google rankings and AI citations
We started with a random sample of 200 tracked AI questions and
found a useful broader Google topic search for 157 of them.
Most Google searches in this analysis are close topic matches, not
word for word copies of the AI question. A question about a romantic
private boat tour may map to a broader search for boat tours in that
city.
The AI side includes the 10 most cited pages for each question.
Sources below that threshold are not part of the rank comparison.
We could not observe every query or retrieval route the AI systems
used internally. The broader Google search is a consistent benchmark,
not a trace of the path AI followed.
Google rank and AI citation may share some of the same underlying
advantages. A well known brand, a useful page, and a strong website can
help with both. The analysis shows a relationship. It does not prove
that moving from position 5 to position 2 causes AI to cite the
page.
Google
AI cites your Google Business Profile... almost always.
Google uses 2 primary data sources to find info about your
business.
The first source it uses is the familiar '10 blue links' list of
ranked web pages.
The second source is the Maps Place Card powered by your Google
Business Profile, which holds your category, location, hours, phone
number, reviews, photos, and other basic facts.
A real search for a tracked operator, Morgan Hill Wine Trolley. Google shows the ranked web results and the Maps Place Card from the operator's Google Business Profile side by side. These are 2 separate data sources.
These 2 surfaces do not share the same ranking system.
Google's local Maps results are mainly determined by relevance,
distance, and prominence.
The familiar web results use a separate search-ranking system.
A #1 webpage ranking and a #1 Maps listing are two totally different
things.
When Google's AI (Gemini, AI Mode, or AI Overviews) mentions an
operator in its results, it shows the Google Business Profile (Google
Maps Card) 96% of the time.
Breakdown of Google-sourced AI citations by surface type: Maps
Place Card (~96%), Google Search results (~2.5%), Maps direct (~0.4%),
other (~1%). Key takeaway: Google's AI citations are almost entirely
powered by the Maps Place Card, not search results or your website as
accessed through Google. Your Business Profile is your direct feed into
Google's AI.
That makes your Google ranking and your Business Profile 2 different
assets.
Your ranking helps a particular page become a source.
Your Business Profile gives AI Google's own record of the
business.
Do not optimize this like a directory listing. Optimize it like the
compact answer Google can use to understand what customers can book,
where they can do it, and why your business fits the question.
Other
websites mentioning your brand increases your chances of being
recommended by AI
A competitor with fewer backlinks, fewer reviews, and a smaller
following can appear in an AI answer while you do not.
That may feel random and unfair, but they likely have one thing that
you don't...
More mentions about their brand across the web.
AI reads local guides, visitor organizations, news coverage,
directories, review platforms, YouTube, Facebook groups, Reddit, and
pages you may have never seen.
Those sources connect a business name with a place, an activity, and
a reason it belongs among the options.
Analysis reference: sources cited for broad location questions
Tripadvisor, Google, Yelp, and Reddit lead broad location
questions. Individual operator websites still appear surprisingly high
for their size.
When we counted the independent websites that mention each operator
by name, the pattern was clear: operators mentioned across more websites
appeared in more AI top-of-funnel answers.
Across 30 active operators, the number of independent websites
mentioning a brand has a clear positive relationship with AI visibility.
Bigger operators tend to have more of both, so this is a strong signal,
not proof that a specific number of new mentions will cause a specific
lift.
A mention contains language and context. It can tell AI that you run
wine tours in Morgan Hill, kayak tours in San Diego, rent ATVs outside
Moab, or operate an escape room in Nashville.
It connects the name to the place and experience.
A larger Ahrefs
study of 75,000 brands finds the same general pattern in Google AI
Overviews: brand mentions have a much stronger relationship with
visibility than backlink volume.
Analysis reference: independent mentions and AI visibility
There is an important limit. Bigger brands get more searches,
reviews, coverage, links, and mentions. After we account for overall
site strength, the mention result remains positive but becomes less
conclusive at this sample size. We cannot promise that 10 new mentions
will produce a specific increase in AI citations.
What we can say is that independent mentions explain the visibility
gap far better than backlink count does.
Operators mentioned on many outside websites appeared in 43% of
tracked AI answers, compared with 30% for operators mentioned on few
sites. Across 4,000 resampled versions of the operator set, mentions
out-predicted backlinks every time. The broad-footprint, low-visibility quadrant identifies operators
whose outside recognition is stronger than their current AI citation
rate. Mention breadth separates the higher- and lower-visibility
groups, 43% versus 30%. Average source prestige does not, 38% versus
37%.
Sample: 30 active operators.
Outcome: visibility across Gondola's fixed set of tracked prompts,
not every AI question customers may ask.
Result: independent website mentions have a positive relationship
with AI visibility; backlink count does not.
Sensitivity check: mentions beat backlinks across all 4,000
resamples.
Limitation: after adjustment for overall site strength, the mention
estimate remains positive but is not conclusive at this sample
size.
We
analyzed the typical SEO-era factors... none of them move the needle in
AI visibility
We tested the numbers operators are routinely told to grow. None
explained how often an operator appeared in our tracked AI answers:
Backlinks: total link count, referring-domain
count, link growth, source types, authority tiers, and anchor-text mix
did not separate the operators AI cited often from the ones it
skipped.
Reviews: after accounting for business size, review
count and number of review platforms had no clear relationship with AI
visibility across 29 operators.
Owned social media: follower count, posting
frequency, platform breadth, and overall social presence did not predict
visibility.
To be clear, this does not mean reviews or social media are
useless.
Reviews still build customer trust and can affect a business's
position inside Google Maps and marketplaces.
Social media can introduce customers and show what an experience
feels like.
Useful articles can rank, earn attention, and explain a
destination.
The data says the totals on those scorecards do not predict AI
visibility. It does not say that these channels have no value. These are
still useful assets across the web, they just do not directly correlate
to AI visibility in our study.
Analysis reference: metrics with no detected relationship to AI visibility
Backlinks
Raw backlink count has no detectable relationship with AI
visibility at this sample size. Referring-domain count shows the same flat relationship. Overall domain strength has a modest positive lean, but it is a
broad property of a developed site rather than a link-count
target. More-visible and less-visible operators have similar link-source
mixes. Anchor-text quality is descriptive context, not a visibility
predictor in this analysis. The full link-profile family remains flat after correcting for
multiple comparisons.
1 operator has almost 290 linking websites and appears in about 9% of
tracked answers. Another has fewer than 70 linking websites and appears
in about 55%. Link count points in the wrong direction for that pair and
does not explain the cohort overall.
Reviews
The raw review count leans positive because bigger businesses
tend to have more reviews and more visibility. After adjustment for
business size, the relationship is flat.
An operator's own Tripadvisor, Yelp, or Google review profile
accounts for about 0.8% of its citations. Roughly 93% of citations from
review websites go to broader list and ranking pages. Reviews may help
an operator rise within a platform, which may then affect those list
pages, but this study did not test that full chain.
Social media
Follower count does not predict how often AI cites an
operator. Followers, activity, platform breadth, and composite social
scores all sit near 0 relationship with visibility. Google's AI products cite social content far more often than the
other engines in the study. The social sources AI cites include several content and account
types. Different engines favor different social platforms; YouTube and
Instagram are especially prominent in Google's AI products. When account ownership is identifiable, about 90% of social
citations come from someone other than the operator. Social citations concentrate in discovery questions and nearly
disappear from booking-intent questions. The measurable volume of outside social citations does not
predict overall visibility either. Social mentions make up about 1% of the wider off-site mention
footprint measured in the independent-mentions analysis.
Account ownership could be identified for about 45% of social
citations, so the owned-versus-outside split is incomplete. The result
does not prove that 1 customer or influencer video creates a visibility
lift.
The
rest of the internet makes your brand known. Your website makes you
knowable.
Here's the simplest way I've found to think about this...
Local articles, customer videos, visitor guides, Reddit threads, and
platform pages make your business known.
They connect your name to a place, an activity, and a reason to
consider you.
Your website makes the business knowable.
It explains the products, prices, schedules, tradeoffs, limitations,
and booking details that help AI and customers decide whether you
actually fit.
You don't control everything the internet says about you, but you do
control whether your own explanation is specific, useful, and
current.
Part 4 of 5
Knowable: How your business gets recommended
The closer a customer gets to choosing, the more AI needs the details only you can provide.
Getting
found isn't enough. AI still needs enough information to recommend
you.
Getting your name into an AI answer feels like a win, and it is.
But then the customer asks a follow-up:
Which option is best for a birthday group?
What does the price include?
Is it private?
Will it work for a 5-year-old?
Is there availability on Saturday?
Is it good for a romantic date?
To answer those questions, AI needs to understand what you sell, who
it fits, what it costs, where it starts, when it runs, and what makes
you different.
Being found gets you into the conversation. Being understood keeps
you there.
A
broad question and a specific question need different sources
Ask AI for fun things to do in a destination, and it leans heavily on
broad articles and lists from neutral sources that provide insight and
an original perspective.
Ask which operator is the perfect fit for a specific customer, and
the source mix changes quickly.
We grouped our tracked questions into 3 levels because each one asks
the AI to do a different job. Look at the sources that AI cites for each
of these prompts/questions:
Level 1 · Broad
Local discovery
Asks about the destination before the customer has chosen an activity.
Best-of lists lead broad discovery with 34% of citations. Business websites of any kind supply 27.6%, while the tracked operator's own site supplies 1.4%.
Level 2 · Narrower
Specific activity
Names the kind of experience but not the customer's situation.
Prompt analysis
AI
What are the best wine tasting tours in Morgan Hill?
The source mix changes quickly. Once the customer names an activity, business websites supply 51.6% of citations and the tracked operator's own site supplies 12.2%. Category, product, and homepage content all take a larger role.
Level 3 · Ready to choose
Very specific
Adds the people, occasion, preferences, or constraints that determine fit.
Prompt analysis
AI
My friends and I want to celebrate a birthday with wine tasting near San Jose. What are the best tours in Morgan Hill?
When the customer adds a real-life constraint, business websites supply 53.2% of citations and the tracked operator's own site supplies 11.8%. Product, category, and homepage content now share the work.
The change happens as soon as the customer names an activity.
"Best-of" lists account for 34% of citations during local discovery,
then fall to 11.7% for specific-activity questions.
At that same step, product pages rise from 2.4% to 19.8%, while
homepages rise from 4.8% to 17.3%.
Very-specific questions do not push the numbers in one perfect
direction. They keep category, product, and homepage content in a much
larger role than those pages had during broad discovery. The important
change is not a smooth climb. It is the jump from “What can I do here?”
to “Which version of this activity fits me?”
This chart describes the kind of page AI cited. It does not tell us
who published it. A business page may belong to the tracked operator,
another operator, or a different local business. That distinction
matters, so we kept the publisher view in the deeper analysis below.
Based on 510 questions with available prompt text from the frozen
December 1, 2025 through June 9, 2026 study window. Those questions
cover 99.94% of citations in the 530-prompt source file. We calculated
citation share inside each question before averaging across
questions. Deeper analysis: how AI chooses sources as questions narrow
Source mix across 510 questions with available prompt text from
the frozen December 1, 2025 through June 9, 2026 study window. Citation
shares were calculated within each question and then averaged, so every
question has equal weight.
Business websites collectively supply 27.6% of citations during local
discovery, 51.6% for specific-activity questions, and 53.2% for
very-specific questions. The tracked operator's own site is a smaller
part of that total: 1.4%, 12.2%, and 11.8%, respectively. “Business
websites” should not be read as “the operator we tracked.”
This view includes the 71 “where to book” questions with at least
one positive citation to the tracked operator's own site during the
frozen December 1, 2025 through June 9, 2026 study window. Shares were
calculated inside each question before averaging across
questions.
When the tracked operator's site does enter a where-to-book answer,
its homepage and product pages do most of the work. Together they
capture 80.6% of the operator's citation share. Its articles and guides
capture 0.2%.
Page type crossed with publisher type for 71 “where to book”
questions during the frozen December 1, 2025 through June 9, 2026 study
window. Cells sum to 100%. The full-window URL classifier leaves 5.1% of
citations in “Other,” and those citations remain visible instead of
being forced into a cleaner category.
The cross-section is messy in a useful way. Booking answers pull from
operator sites, other businesses, official tourism sources, platforms,
and pages we could not classify cleanly. AI is not following one
universal source hierarchy. It is assembling an answer from whichever
pages can support the question.
Very specific questions from the frozen December 1, 2025 through
June 9, 2026 study window. The chart includes information needs
represented by at least 15 tracked questions. A question may contain
more than one need, but every row shows its own page-type mix and sums
to 100%.
Different details lead AI to different pages. Group-format and
transportation questions pull harder toward product and homepage
content. Timing questions lean more toward guides. That is why “just
make a page for AI” is not a complete strategy. The useful page depends
on what the customer needs to know.
These patterns describe Gondola's tracked questions. They do not
represent real-world query volume or prove that a page type caused a
recommendation or booking.
We
built 1 new page specifically for AI and it increased citation rate by
121%
This was one of our biggest findings in our study that influenced AI
visibility directly.
It's called the AI Start Here page and it's a
technique used by many leading websites already.
Operators with an AI Start Here page appeared in 53% of tracked
questions, compared with 24% for operators that did nothing.
It is a public HTML reference page that gathers information Gondola
already has about an operator at 1 stable URL. The page itself gets
regenerated automatically any time that an operator changes their
products, pages, or new reviews get imported.
Page builders appeared in 53% of tracked questions versus 24% for
the no-intervention group. After adjusting for site strength, the
estimated lift tied to the page itself was 7 percentage points.
The version we tested was assembled from 5 layers:
Business identity
Pulled from the operator's Gondola company and website
settings.
Includes the company name, phone number, domain, booking link, and
active LinkedIn, Twitter, Instagram, Facebook, YouTube, and WhatsApp
profiles.
Published website content
Starts with the live website's HTML structure so the page remains
part of the operator's own site.
Pulls every available WordPress page and includes its title and
published content.
Tours and activities
Pulls directly from the operator's product records in Gondola.
Includes descriptions, sales pitches, highlights, itineraries,
inclusions, exclusions, product FAQs, locations, and durations.
Customer evidence
Imports the latest qualifying reviews from the sources the operator
has enabled, including Google, Tripadvisor, Airbnb, Yelp, and
Trustpilot.
Uses the operator's configured minimum star rating and keeps the
review text, author, date, source, and rating together.
Questions and local knowledge
Generates up to 100 Q&As from the operator's products and
website pages, 100 from patterns in customer reviews, and 100 from the
destinations the operator serves.
Organizes those answers into products and services, customer
questions, and destination guides so AI can retrieve the relevant detail
quickly.
It is not a secret landing page or a replacement for the pages
customers use to compare and book.
Think of it as a reference library for the business: a single place
where AI can retrieve product details, pickup rules, itineraries,
cancellation policies, and pretty much everything else it would want to
know without having to piece together the entire record from
scratch.
The page has evolved several times since this experiment. Gondola
customers today get a more advanced version of it.
Across 14 page builders and 20 comparison operators, the builders
appeared more often at every question level: 10.8% versus 5.5% for local
discovery, 61.8% versus 41.3% for specific activity, and 59.8% versus
46.2% for very specific questions.
The page was much more useful once the customer had an activity in
mind. At that point, AI needs real details about the operator's
products, policies, and what makes each experience different. That is
exactly what the AI Start Here page brings together.
Deeper analysis: results by AI engine
The page builders were about 1.4x to 1.5x as visible on every AI
engine in the wider descriptive comparison. Page builders received 36% to 50% more citations on every engine
in the wider descriptive comparison.
These platform charts compare all operators that had the page with
operators in the other test groups. They are useful for seeing whether
the pattern was confined to 1 engine, but they are not a clean
page-versus-no-page treatment estimate.
AI
directly cited this new 'AI Start Here' page 5,784 times during our
experiment
The group comparison is encouraging, but the URL data is even more
useful because we can see AI using the pages themselves.
13 of the 14 operators with an AI Start Here page had that exact page
cited.
Together, the pages received 5,784 citations during the study, and
the most-cited page earned 1,387 on its own.
13 of the 14 AI Start Here pages were cited directly. The range
runs from 1,387 citations for the leader to 0 for 1 page that never
entered the index.
That proves something narrower than a booking result: AI used these
pages as sources. It does not prove that the pages caused a
recommendation or a booking.
Deeper analysis: what AI cited from these pages
AI Start Here pages earned 3,683 citations for very-specific
questions and 1,889 for specific-activity questions, compared with only
209 for local discovery.
This page seems to help after AI has a reason to consider the
operator and needs exact details about the business or activity. It
doesn't replace the outside sources that introduce businesses during
broad research, and it shouldn't try to.
Specific product descriptions, Q&As, and local details
attract more citations than generic boilerplate.
What happened when
2 operators left Gondola
I'm a little embarrassed to show this because we lost customers
during the experiment.
Still, it taught us something we could not have learned from the
operators who stayed.
One business closed and its website went down completely, while
another moved to the GoDaddy Airo website platform. In both cases, AI
kept mentioning the business by name while citations to its own website
fell drastically.
So the AI kept mentioning the brand because it was mentioned across
the web, but it failed to pull specific details and citations to the
site itself collapsed.
After each website change, brand mentions held comparatively
steady while citations to the operator's own website fell sharply. The
dashed line shows the median visibility of operators that stayed on
Gondola.
Two isolated cases don't prove that leaving Gondola would cause a
decline. I wouldn't trust that claim if a software company made it about
a competitor, and you shouldn't trust it from us either.
But the narrower lesson is still useful: AI can remember a brand
while losing a current source for its products and details.
Moving platforms isn't the problem by itself. The risk is removing
useful URLs, skipping redirects, or replacing detailed pages with thin
or stale information.
If you move your website, make sure you preserve or improve the
information and the URL routes so that AI has somewhere current to
point.
90%
of Gondola-powered operators hold the top 3 citations across all AI
systems for the prompts we tracked
[SHAMELESS PLUG WARNING]
While we were in there, we wanted to understand how Gondola website
operators compared to their immediate competition.
This was a very pleasant surprise.
For 23 of 31 (74%) tracked brands, Gondola-powered operators were the
most-cited business site in its market.
For 28 of 31 (90%) tracked brands, Gondola-powered operators ranked
in the top 3 across all AI prompts.
It means that, while the businesses were on Gondola, their own
websites usually supplied AI with more useful pages than any competing
operator website in the market.
For 23 of 31 active operators, the operator's own domain earned
more citations than every competing business domain in its market. 28 of
31 ranked in the top 3.
That is the golden part of this finding. Your website does not have
to be the biggest source on the internet. It needs to be the clearest
and most useful source about your own business, and for most of our
operators, it was.
What
we tested that did not meaningfully change AI visibility
The useful result is not only what worked. We also tested 3 common
ideas about operator websites that did not hold up well enough to build
a strategy around:
Adding more schema types: Keep structured data
complete and correct, but more types did not meaningfully improve AI
visibility.
Publishing more blog posts: Post count and total
blog volume had no positive relationship with visibility. Only about 2%
of article citations went to the operator's own blog.
Writing longer pages: The small content-depth
analysis did not produce a defensible word-count target.
The bias here is that Gondola websites are already schema-rich.
80% of the sites in our original cohort shipped the same rich 12-type
structured-data bundle, including FAQPage, Organization, TravelAgency,
BreadcrumbList, WebPage, and WebSite.
We keep that information complete because it helps search engines
understand the page and can support richer search results. It just did
not explain the wide differences in AI visibility we observed.
Deeper analysis: what did not meaningfully change AI visibility
Schema is useful
hygiene, not a growth strategy
Adding more schema types did not meaningfully separate the pages AI
cited most from the pages it cited least. Across 139 rival operators,
overall structured-data quality had a small positive relationship with
visibility, but the adjusted result stayed below the threshold we set
for a material finding.
Completeness carried a faint signal. Validity and the number of
schema types were essentially flat. So, yes, keep structured data
complete and correct. Then improve the information underneath it. Schema
can label a price or an FAQ, but it cannot invent a useful answer.
Across 139 rival operators, stronger structured data came with a
modest raw lift in visibility. After adjustment, the result stayed below
the threshold we set for a material finding. Schema completeness has a faint positive relationship with AI
visibility. Validity and the sheer number of schema types are
essentially flat.
Publishing
more did not make the operator's site more useful
Blog post count, total blog word volume, and the number of posts
tailored to specific questions had no positive relationship with
visibility in this sample. Only about 2% of article citations went to
the operator's own blog, while about 87% went to articles published by
other people.
About 2% of article citations go to an operator's own blog,
compared with roughly 87% for articles published by others.
That is not an argument to stop publishing useful articles. It is a
warning against using archive size as the goal. The AI Start Here pages
received about 10x as many citations per page as the typical operator
blog post because they concentrated useful business information instead
of simply adding another URL.
A separate, small content-depth analysis did not give us a defensible
word-count target either. Longer AI Start Here pages leaned toward fewer
citations, but the sample included only 14 pages and the uncertainty was
too wide to call that a reliable effect. The fair conclusion is that we
found no useful length target, not that shorter pages automatically
win.
Outside sources help AI learn that your business belongs in the
conversation. Your own website gives it the current details it needs
when the customer asks which option fits, what the price includes, or
where to book.
That is the difference between being known and being knowable.
Neither one guarantees a recommendation or a booking. Together, they
give AI better evidence and give the customer a clearer path from an
answer to a decision.
Part 5 of 5
Conclusion and key takeaways
The role of the website has changed. The final model is simpler than another AI checklist.
To get
booked, your business has to be known and knowable
The platforms will keep changing. Google will add another answer
format, ChatGPT will change how it searches, and someone will invent a
new acronym and sell a checklist before lunch. You cannot rebuild your
strategy every time that happens.
The evidence in this report points to a simpler model: to get booked,
your business has to be known, and it has to be knowable.
Known and knowable: Outside sources help your business enter the conversation. Specific information on your website helps AI and customers understand, trust, and book it.
The searches that used to introduce operators to new customers are
producing fewer website visits. Search visits per site in the
uninterrupted operator cohort fell roughly 41% in 2 years, while AI can
now search, read, and compare sources without sending a visit that
appears in ordinary analytics.
The customer did not disappear. More of the research now happens
before the customer reaches your website. That makes 2 conditions
especially important: AI has to include your business among the options,
then both AI and the customer need enough current information to choose
it.
Your
business has to enter the conversation before it can win the
booking
Known does not mean famous. It means the web connects your business
with the right place, activity, and customer need often enough for AI to
consider it. Visit
Morgan Hill's feature about Morgan Hill Wine Trolley is 1 example:
an independent local source connecting an operator to wine tasting in a
specific destination.
Google rankings still matter because pages near the top have a better
chance of earning an AI citation. Google Business Profiles matter too.
When Google's AI mentioned an operator in our study, it showed the
operator's Google Business Profile 96% of the time.
But the strongest pattern was broader than either one. Businesses
mentioned across more independent websites generally appeared in more AI
answers. Most of the pages AI cited did not rank in Google's top 20,
almost 87% of article citations went to articles published by other
people, and most identifiable social citations came from accounts the
operator did not own.
The public record is bigger than your website. Your site can support
it, but it cannot create the whole record by itself.
Once AI
finds you, it still needs a reason to choose you
When a customer asks for general ideas, AI leans heavily on lists and
guides. Best-of lists receive 34% of citations during local discovery.
Once the customer names an activity, product pages rise from 2.4% to
19.8% of citations and homepages rise from 4.8% to 17.3%. That is a
page-type shift across all publishers, not a claim that the tracked
operator owns every business page.
When the tracked operator's own site does enter a where-to-book
answer, its homepage and product pages capture 80.6% of its citation
share. Those are the pages most likely to explain whether a trip fits,
what it includes, where it starts, and how to book it.
The AI Start Here experiment was the strongest knowable result in the
report. Operators that built the page appeared in 53% of tracked
questions, compared with 24% for operators that did nothing. After
adjusting for site strength, the estimated lift tied to the page was 7
percentage points. Thirteen of the 14 pages were cited directly, showing
that AI used them. That does not prove the pages caused a recommendation
or a booking.
Your website now has 3 jobs
Fewer customers may use your website for the entire journey from
first idea to booking. That does not make the website less important. AI
can read it before the customer visits and use its details while
comparing options.
The website now has to do 3 things at once:
Give AI a reliable source of current business
facts. Publish the products, prices, schedules, meeting points,
policies, limitations, and differences that determine fit.
Let the customer verify the recommendation. When
someone clicks through, the page should confirm what AI said and give
them enough context to trust it.
Make the booking easy and safe. Show current
availability, explain what happens next, and make the direct-booking
path clear.
Those jobs belong across the site. The homepage explains the
business. Product pages explain whether a specific experience fits this
customer. Policies and FAQs handle the questions that can stop a
booking. The AI Start Here page is 1 useful reference inside that
system, not a substitute for it.
The useful details are usually the ordinary ones operators already
explain on the phone: whether a trip is private, how many people can
join, what is included, where it starts, what happens in bad weather,
and how 2 similar options differ. The work is getting those facts out of
phone calls, messages, and people's heads and onto pages that stay
current.
Use 2 questions to
judge every new tactic
When a new AI tactic appears, ask what job it performs:
Does this help the right customers and systems find the
business?
Does this help them understand, trust, and book it?
An independent article can help people and AI find you. A complete
product page can help them understand you. An accurate Google Business
Profile can support both, while a current booking page turns that
confidence into action. If a tactic does neither job, it deserves a much
harder look.
The 2 questions do not guarantee a recommendation or a booking. They
give you a way to judge the next tactic without rebuilding your strategy
around the latest tool.
Analysis reference: known and knowable
What changed: Search visits per site fell roughly
41% across the uninterrupted operator cohort, while direct AI referral
traffic remained too small to represent the research AI performs before
a customer visits.
Known across the web: Google rankings improve a
page's chance of earning a citation. When Google's AI mentioned an
operator, the Google Business Profile appeared 96% of the time.
Independent website mentions had a much stronger relationship with AI
visibility than backlink count, while review count after adjusting for
business size and owned-social metrics were essentially flat.
Knowable on your own terms: Product and homepage
citations rose sharply once a customer named an activity. When the
tracked operator's site was cited on a where-to-book question, its
homepage and product pages captured 80.6% of its citation share. AI
Start Here page builders appeared in 53% of tracked questions versus 24%
for comparison operators, with a 7-point adjusted estimate.
What did not provide a shortcut: More blog posts
did not improve visibility, schema behaved like useful hygiene rather
than a demonstrated growth lever, and the content-depth analysis did not
produce a defensible word-count target.
Important limit: The study measures citations,
mentions, rankings, traffic, and observed website changes. It does not
prove that completing a fixed checklist causes an AI recommendation or a
booking. Known and knowable is Gondola's strategic interpretation of the
measured findings.
Your customers
still need a reason to choose you
AI can combine what other people say with the current facts you
publish. It cannot manufacture independent trust, witness the trip you
ran this morning, or fill in an important detail you never put
online.
Outside sources help your business enter the conversation. Useful,
current pages help AI answer the specific question and give the customer
a place they can trust when they are ready to book.
Your customer may ask AI to do the research. They still need a reason
to choose you.
How we did the research
The short version: we tracked 530 real customer questions about 34
independent tour and activity operators for a little more than 6 months.
We recorded which businesses appeared, which pages AI cited, how Google
traffic changed, what the wider web said about each business, and what
happened when websites changed.
The report measures answers, citations, mentions, rankings, and
traffic. It does not measure every AI question in the market or prove
which source created a booking. If you want the exact study design,
tools, sample sizes, limits, and Gondola disclosure, they are all
below.
Study design, tools, data, limits, and Gondola disclosure
The questions,
businesses, and AI systems
The main AI answer and citation study covers 34 independent tour and
activity operators that use Gondola. They include tours, rentals,
attractions, and other bookable experiences across North America and the
Caribbean.
From December 1, 2025 through June 9, 2026, Peec.ai monitored 530
distinct questions about those businesses across 5 AI platforms:
ChatGPT
Gemini
Google AI Overviews
Google AI Mode
Perplexity
The set includes broad location questions, activity and offer
questions, and detailed questions based on what each business actually
sells. Prompt text is available for 510 of the 530 prompt IDs. Those 510
prompts cover 99.94% of all citations in the frozen source.
Peec recorded 406,373 AI responses across those questions and
platforms. We round that to “more than 400,000 answers” in the
report.
The frozen citation source contains 3,388,833 citations across the
530 prompt IDs. These are source citations, not website visits,
bookings, or unique pages. The same URL can be cited many times and can
appear in more than 1 question.
What we mean by AI
visibility
AI visibility is how often a tracked business appears in the recorded
answers to its fixed set of questions.
That makes it useful for comparing operators, comparing groups, and
watching what happens after a change.
It does not represent every question real customers ask. It does not
measure the total volume of AI searches in the market. A different
question set could produce different percentages.
The other data behind the
report
The AI answer and citation data sits alongside several other
sources:
Google Search Console for search queries and clicks
First party website analytics from Umami
Ahrefs for rankings, backlinks, referring domains, and modeled
search data
Google search results collected for specific comparison
analyses
Public websites, articles, platform listings, reviews, and social
sources cited in the tracked answers
Gondola website and server data used to validate pages and website
changes
The broader website experiment contains 35 sites. 1 control site has
website data but no Peec visibility record, so the AI analysis uses the
other 34. Most of the remaining analyses use between 29 and 34 operators
after removing businesses with missing data, closures, or a mismatch
between the sources. Narrower comparisons use 14 to 18 operators. We
show the smaller sample next to the claim when it materially changes how
strongly the result should be read.
Study limits
The question set leans toward finding and comparing experiences. It
is not a sample of every real-world question asked across all AI
systems.
A citation shows which page supports an answer. It does not prove
that the page caused the recommendation or created a booking.
Most relationships in the report are observational. When we ran an
experiment or examined a business change, we say so and preserve the
limits of the comparison.
Some null findings may reflect the sample size. “We do not detect a
relationship” does not mean the factor can never matter for any
business.
The platforms changed during the study and will keep changing after
publication.
Google Search Console and website analytics measure different
things, so their traffic totals should not be expected to match
exactly.
Direct AI referral traffic is difficult to isolate. The roughly 0.6%
figure in this report should be read as an upper bound, not a precise
count of every visit from a chat tool.
A note about Gondola
Gondola funded and conducted this research. Every operator in the
main study cohort is a Gondola customer, and several analyses examine
pages, websites, or interventions Gondola created.
That relationship is a strength and a source of potential bias.
It is a strength because a standardized network of real
small-business websites lets us compare changes that would be nearly
impossible to observe across unrelated enterprise brands.
It is a source of potential bias because Gondola sells websites and
will use these findings to shape its product. We have tried to handle
that plainly: show the results that contradict our assumptions, separate
raw comparisons from adjusted estimates, label our interpretations, and
avoid claiming that Gondola already solves every problem in the
report.
The research points toward a product direction. It does not turn that
direction into proof.
Peec.ai provided the AI monitoring platform that made daily tracking
across 5 surfaces possible. Gondola performed the analysis and remains
responsible for the claims in this report.
