LuckyPaw: 20+ inquiries a month and #1 in ChatGPT, in under 3 months

LuckyPaw is dog boarding, daycare and dog walking in Wassenaar, run out of a family home. Boarding takes one family at a time, with a maximum of two guest dogs. That is the product, and it is a good one. The problem was that nobody could find it.

At the start there was no website, no search presence, a Google Business Profile that had just been created, and no reviews. Every trust signal a buyer looks for was missing.

Local pet care is also a crowded category, and the top of it is not owned by local businesses. It is owned by booking platforms. Search for a dog sitter in almost any Dutch town and you get marketplace listings first. Those same platforms are what AI assistants read when someone asks for a recommendation, because they are the pages that exist, are structured, and are linked to. A one-family boarding service with no website does not appear in that conversation at all.

So the job was not to outspend the platforms. It was to make LuckyPaw the most specific, best-described, most machine-readable answer to a narrow local question — and to do it before the business had any review history to lean on.

Everything below is live on luckypaw.nl and can be checked directly.

1. 01 Built the website, bilingual from day one Wassenaar has a large international population, so the site ships in Dutch and English as equal versions, not a translation bolted on. Every page carries a canonical tag and hreflang alternates for nl-NL, en and x-default.
2. 02 Gave every service its own page Group walks, private walks, daycare and boarding each have a dedicated page in both languages — eight service pages in total. Each one describes the format, the group size and what is included, rather than pooling everything into one services list.
3. 03 Published every price A full pricing page in both languages, with per-service prices, monthly plans and multi-day passes marked up as Offer and UnitPriceSpecification. Published prices are answerable; “contact us for a quote” is not.
4. 04 Built the entity in structured data A LocalBusiness node with a stable @id, postal address, geo coordinates, opening hours, telephone and email, an areaServed list covering Wassenaar and the surrounding towns and sub-areas, and an OfferCatalog holding the four services. Service pages, guides and articles add Service, FAQPage, BreadcrumbList, Article, BlogPosting and Person, all pointing back to the same business node.
5. 05 Wrote the local guides nobody else had written Seven local guides per language — off-leash areas in Wassenaar, the dog beach rules, dog parks and swimming spots, walking a reactive dog, moving to Wassenaar with a dog, the dog walking service comparison, and boarding options across Wassenaar and The Hague. These answer the questions people actually ask before they ever search for a provider.
6. 06 Launched a blog with real sourcing standards Six articles per language across four categories, plus a published editorial policy that states how articles are researched, how AI is used in drafting, what counts as an acceptable source, and where the line is on veterinary advice. The policy is a page on the site, not a claim in a footer.
7. 07 Made the site readable by AI crawlers on purpose An llms.txt file describing the business, the service area, every price, the key booking terms and the full page list. A robots.txt that names GPTBot, OAI-SearchBot, ClaudeBot, PerplexityBot, Google-Extended, CCBot and Bytespider and explicitly allows them.
8. 08 Wired up the technical foundations An XML sitemap with hreflang alternates on every entry, clean URLs, static pre-rendering on Vercel, and GA4 so inquiries and traffic sources are actually attributable.
9. 09 Connected the Google Business Profile to the site The Business Profile is linked from the site, so the map listing and the website reinforce one entity instead of competing as two.
10. 10 Made contact one tap, not a form WhatsApp, phone and email are exposed directly on every page. For a local service where the first question is usually “are you free on these dates”, a contact form is a delay, not a funnel.

We test the queries a buyer would actually type, logged out, so the answer is not shaped by an account history. Here is where LuckyPaw stands.

Google Search Console, web search, 28 June to 24 September 2026.

Clicks

102

Impressions

3.5K

Average CTR

2.9%

Average position

11.3

Through July and August daily impressions stayed low and flat, mostly under 50. From the end of August the line steps up and holds: in September it runs at roughly 60–130 a day.

An average position of 11.3 means the site sits around the bottom of page one and the top of page two on a typical query. That is the position a three-month-old site with no backlink history can reasonably hold, and it is already enough to produce inquiries at this volume in a market this specific.

Every number on this page comes from one of three sources, listed below. We publish the method so the figures can be judged on it.

Google Search Console

28 June – 24 September 2026, web search only. Clicks, impressions, average CTR and average position are taken directly from the property for luckypaw.nl, unfiltered. Discover and Google News are excluded. Average position is Search Console’s own impression-weighted average.

ChatGPT, logged out

Tested 24 September 2026. One query, run in a logged-out session so no account memory or personalisation shapes the answer. The screenshot above is the run. Generative answers are not deterministic: the same query can return a different answer on a different day, in a different session, or on a different model version. This is a single dated observation, not a ranking guarantee.

Website inquiries

20+ per month; 5 paying clients to date. An inquiry counts as one new person making first contact about a booking through a channel exposed on the website — WhatsApp, phone or email. Repeat messages from the same person in one conversation count once. Existing clients rebooking are not counted. A paying client is someone who completed and paid for at least one booking.

The full procedure we run when measuring AI visibility on a paid engagement — prompt-set construction, clean-room conditions, run counts and what the evidence does and does not support — is published as the GEO Measurement Protocol.