For hotel groups & collections · 10+ properties

You know where you rank on Google. You have no idea what AI tells travellers about your hotels.

Millions of trips are now planned in ChatGPT, Gemini, Claude and Perplexity, and each answer recommends three to five hotels. Sightline is a monthly audit of what these tools say about every property you run: how often you appear, what they get wrong, and exactly which sources to fix.

Nothing to install · No access to your systems required · First audit within ten working days

EVIDENCE RECEIPT · STORED TRANSCRIPT
2026-07-22 06:41 UTC · run 3 of 3
PROMPT  “is The Lanyon Bath worth staying at”
CHATGPT SAID  “Rooms start at around £260 per night, and note the hotel has no on-site parking.”
THE HOTEL’S WEBSITE SAYS  Rooms from £189 · 22-space private car park, £12/night
CITED SOURCE  kayak.co.uk · stale 2023 listing · cited in 7 of 9 sampled answers
STATUS  Live 34 days · fix pack ready
3–5 hotels
is all an AI answer recommends. There is no page two.
Four engines, four rulebooks
Gemini reads hotel websites. ChatGPT leans on directories. Perplexity trusts TripAdvisor. Claude favours editorial coverage. Each needs different fixes.
Booking.com
is where AI answers often send your guests, even when the AI is recommending your hotel.
How it works

We ask. We verify. You decide.

1

We ask what your guests ask

Sixty questions about each hotel (“is it dog friendly”, “does it have parking”, “is it worth staying at”) plus twenty market searches your comp set competes in (“best spa hotel near Bath”). Every prompt goes to ChatGPT, Gemini, Claude and Perplexity, three times each, every month. That is 960 stored, timestamped answers per property.

60 hotel + 20 market prompts × 4 engines × 3 runs
2

We check every claim

Facts are verified against your own website: parking, policies, dining, spa. Rates are different, they change daily, so we never judge an exact price. Instead, we flag every public rate claim an engine makes and test it against your published from-rate, or a minimum rate you give us at setup.

Ground truth: your site, crawled weekly, plus your rate floors
3

You get the audit, and the ammunition

A monthly report your board can read in two minutes, plus a fix pack for every problem: the corrected schema, the drafted correction email, the listing rewrite. Your team executes; we never touch your site or listings.

Full sweep monthly · rotating subset weekly for same-day alerts
The dashboard

Every answer, every error, every receipt

This is what your team sees: live scores across every engine, every property, and your named competitors measured on the very same prompts. Click an error to see its receipt and fix pack.

SightlineWhat AI sees
Kingsbridge Hotel Group · 12 propertiesSweep 19 Jul · 11,520 answers stored
Alert · today 06:41New critical error: ChatGPT misquoting The Lanyon’s rates (+£71/night). Fix pack ready · alert emailed to 3 recipients.
Portfolio AI Score
57
▲4 up vs last month
vs June
Comp set 61 · gap −4, narrowed from −9
ChatGPT
51%▲5 up vs last month
inclusion · Cites Bing index + directories
Gemini / AI Overviews
62%▲3 up vs last month
inclusion · Cites hotel sites most
Claude
41%▲6 up vs last month
inclusion · Web search, favours editorial sources
Perplexity
44%▲8 up vs last month
inclusion · Anchors to TripAdvisor

Properties

60 hotel + 20 market prompts × 4 engines × 3 runs
PropertyScoreVisibilityAccuracyLive errors6-mo
The Kingsbridge, Cardiff71 ▲6 up vs last month64%92%1
Harbour House, Bristol66 ▲4 up vs last month55%96%clear
The Lanyon, Bath58 ▼2 down vs last month61%74%3
Gwynfor Hall, Brecon54 ▲9 up vs last month38%98%clear
The Copperworks, Swansea47 ▲1 up vs last month42%68%4
St Fagans Court, Cardiff43 ±0 no change vs last month29%88%1

Comp set benchmark

20 market prompts · same engines, same runs
The Angel, Bathcomp
71%72
The Kingsbridge, Cardiffyours
64%71
The Lanyon, Bathyours
61%58
Peterstone Court, Breconcomp
58%68
Harbour House, Bristolyours
55%66
The Gainsborough, Bathcomp
52%64
Competitors you name answer the same market prompts, on the same engines, the same number of times. Bars show the share of market answers that include the hotel; the right-hand figure is the AI Score, which also counts accuracy. That is why The Lanyon appears often but scores 58. Comp set average 61 · your portfolio 57.

What drives recommendations

your market · citation share
tripadvisor.co.uk
34% · top driver on Perplexity
booking.com
19% · engines quote OTA pages, not yours
your own websites
14% · Gemini only · should be higher
itsyourwales.com
8% · one listicle drives 8%, a PR target
kayak.co.uk
6% · source of 2 live errors

Live factual errors

9 open
EVIDENCE RECEIPT · STORED TRANSCRIPT
2026-07-22 06:41 UTC · run 3 of 3
PROMPT  “is The Lanyon Bath worth staying at”
ANSWER  “Rooms at The Lanyon start from around £260 per night, and note that the hotel has no on-site parking.”
YOUR WEBSITE SAYS (crawled 17 Jul)  thelanyon.co.uk/rooms: “Rooms from £189” · /parking: “22-space private car park, £12/night”
CITED SOURCE  kayak.co.uk · stale 2023 listing · cited in 7 of 9 sampled answers
SCALE  live 34 days · “hotel bath parking” ≈ 2,400 UK searches/mo (Google keyword data, Jun)
FIX PACK · YOURS TO EXECUTE
Correction email → Kayak listings team
Rate FAQ block for /rooms (paste-ready)
Schema: priceRange + parking amenity JSON-LD
Intelligence only: Sightline never modifies your website or listings. Ground truth: your own site, crawled weekly, plus the rate floors you set. Every figure is measured from stored transcripts or attributed to a named third-party source, nothing modelled. Method: 60 hotel prompts + 20 market prompts per property × 4 engines (ChatGPT, Gemini, Claude, Perplexity) × 3 samples = 960 answers per property, per month. Full sweep monthly, rotating subset weekly for same-day alerts. All answers stored & timestamped.

The Sightline dashboard · demo group data

Inside the monthly audit

One score. Every receipt.

Measured

AI Score & comp set

One number per property, 0–100, blending visibility and accuracy, benchmarked against competitors you name and trended monthly.

Evidence

Live error receipts

Every false claim with the stored transcript, the contradicting line from your own website, and the source that caused it.

Diagnosed

Why you’re losing

For every search a rival wins, the autopsy: which sources carried them, and what they have that you don’t.

Measured

How AI describes you

The words engines actually use about your properties, set against the brand you’re paying to build.

Measured

Direct link share

When AI recommends you, does it link your site, or Booking.com? Tracked monthly, against your comp set.

Ranked

What drives the answers

The sources that actually move recommendations in your market, ranked by citation share. Your PR target list, in effect.

Built the Porter way

No invented numbers. Just things we’ll put in writing

The Sightline promises · Porter
i.

Every number is measured or sourced. Each figure traces to a stored, timestamped transcript or a named third party. Nothing modelled, no “estimated reach”, no guessing dressed as data.

ii.

We never touch your website or listings. Sightline is intelligence, not access. We hand your team the exact fix; you stay in control of every channel.

iii.

No ranking guarantees. Nobody controls what an AI recommends, and anyone promising placement is selling you a guess. We guarantee you’ll know, and that errors get found.

iv.

Nothing to install, nobody to onboard. We need a list of your properties and the competitors you care about. That, in its entirety, is the setup.

v.

Leave whenever you like. Monthly rolling as standard. Your data and every stored transcript are yours to take with you.

Pricing

One price. Per property.

£199
per property, per month · minimum 10 properties
  • Full monthly audit across ChatGPT, Gemini, Claude and Perplexity
  • Every error documented with evidence, plus its fix pack
  • Comp-set benchmarking against competitors you name
  • Same-day alerts when something new appears
  • First audit delivered within ten working days of signing
Book 20 minutes
Billed monthly, cancel anytime, or take 15% off billed annually.
Questions

Asked in the boardroom

Can you make ChatGPT recommend our hotels?

Nobody can promise that, and we won’t pretend otherwise. What we can do is show you exactly which sources drive recommendations in your market, prove where you’re absent or misdescribed, and hand your team the precise fixes. The engines decide; we make sure they’re deciding on accurate, complete information.

How do you check rates when prices change every day?

We never judge an exact nightly price, because dynamic pricing means there is no single right answer. Instead, we flag every public rate claim an AI engine makes about your hotels and test it against your published from-rate, or a minimum rate you give us at setup. If ChatGPT tells travellers rooms start at £260 and your floor is £189, that is a documented error with a fix pack. Pricing above your floor is never flagged.

Exactly how many prompts do you run, and how often?

Per property: 60 questions about the hotel itself (rates, parking, pet policy, whether it is worth staying at) plus 20 market searches your comp set competes in (like “best spa hotel near Bath”). Every prompt goes to four engines, ChatGPT, Gemini, Claude and Perplexity, three times each. That is 960 stored, timestamped answers per property, per month, with a rotating subset re-run weekly to power same-day alerts.

Do you need access to our systems?

None. No PMS integration, no website access, no analytics connection. We work entirely from the outside, which is the same view your guests and the AI engines have. Setup is a property list, your named competitors and, if you choose, your minimum rates.

Who fixes the problems you find?

Your team, or your agency, using the fix packs we supply: paste-ready schema, drafted correction emails, listing rewrites, step-by-step claims for unclaimed profiles. Most fixes take minutes once you know exactly what to change and where.

How is this different from our SEO reporting?

Google rankings and AI answers barely overlap: the sources AI engines cite are largely different from what ranks organically, and each engine weighs them differently. Your SEO tells you about the ten blue links. Sightline tells you about the one answer that now replaces them.

Why Porter?

Porter has built for hotels for years, including booking partnerships with 300,000 properties in a previous life. We know how hotel data flows around the internet, which is precisely the plumbing that decides what AI says about you.

Twenty minutes. We’ll bring what AI says about you.

Before the call, we run a first sweep on your flagship property. You’ll see the real answers, the real errors, and the real sources, then decide if you want the full picture, every month.

gary@tryporter.co.uk · First audit within ten working days