45% of a travel section’s search clicks, from a standing start
45%of the travel section’s Google Search clicks
In eleven weeks BILD Reisen went from no articles to 109. Those pages accounted for 45% of every Google Search click across its travel section, and the brand became measurable in AI answers for the first time.
- 584KGoogle Search impressions on new pages
- 45%share of the section’s Search clicks, 45.0% of 14,898
- 3 → 9mentions in AI answers, one stable panel
Serge HeitmannGeschäftsführer, Axel Springer Teaser Ad GmbH
Building a travel library that search engines and assistants could both use
BILD Reisen is BILD’s booking-oriented travel platform, operated by Axel Springer Teaser Ad GmbH. It sells package holidays, hotels, short breaks and cruises, and it needed a travel magazine that could earn discovery in its own right. When the engagement began there was nothing to measure: on 14 May 2026, before a single article existed, the tracked prompt panel returned no mentions of the brand at all, and it sat 17th of the 18 brands on that panel.
We started by measuring, then built a content operation around the client’s own bookable inventory, with the compliance rules of a German newsroom enforced inside the production system rather than checked at the end. From late June, every hotel an article could name had to exist in the catalogue the client actually sells from.
By 31 July, 109 articles were live. They earned 583,914 Google Search impressions and 6,706 Search clicks from a corpus that did not exist before 26 May, and accounted for 45% of all Search clicks across the entire travel section. On the AI panel, mentions went from none before launch to three by early June, and tripled again to nine over the following three weeks.
- Client
- BILD Reisen
- Operator
- Axel Springer Teaser Ad GmbH
- Industry
- Booking-oriented travel platform
- Market
- Germany · German
- Scope
- AI visibility, content production and reporting
- Measured period
- 14 May – 31 July 2026
- Published in period
- 109 articles · ~399,000 words
- Figures re-derived
- 17 August 2026
No conversion, booking or revenue data was in scope. Every figure on this page is a visibility measurement, and every search figure covers Google Search only.
A brand everyone knows, that AI never named
Assistants answering German travel questions recommended competing tour operators and booking portals. BILD Reisen was not among them, and there was no library of its own for an assistant to draw on.
The constraints made it harder. Every article had to satisfy the German press code on truthfulness, diligence, the separation of advertising and editorial, and personal rights; state EU air-passenger compensation tiers correctly; and never send a reader to a competing operator or booking portal. Recommendations could only point at hotels the client was able to sell.
Starting point · 14 May 2026
- 0articles live
- 0Google Search impressions
- 0brand mentions in AI answers
- 17th of 18rank on the tracked panel
Compliance built into production, not bolted onto the end
A newsroom cannot ship travel content that invents a price, cites a dead source, or recommends a hotel the business does not sell. We moved those rules inside the production system, so drafts reached a human editor having already passed them.
Step 01
Understand your customers
We turned the travel desk’s audiences, voice and hard editorial limits into one governed Brand Kit — from press-code obligations down to which words may not describe a hotel.
Output: six governed source files
Step 02
Find the questions customers ask
We recorded a pre-launch baseline on a tracked panel of German travel prompts, then rebuilt it in July as 45 non-branded booking questions weighted by demand.
Output: versioned prompt panel
Step 03
Publish content that sells
109 articles, produced under a rule requiring three to five authoritative sources and blocked from naming a competing operator. From late June, hotel recommendations were gated against the client’s bookable catalogue.
Output: traceable, catalogue-safe content
Step 04
Update to stay relevant
We diagnosed the live corpus and refreshed all 81 articles with a rewrite that could not drop or invent a price. It ran on every page at once with no holdout, so we report the refresh as an operation, not as a result.
Output: fact-locked refresh
Step 05
Report and improve
Monthly reporting with independent verifiers re-deriving every published number. When a figure turned out to be a measurement artefact, we corrected the report and removed the comparison.
Output: figures that survive checking
May 2026
Brand Kit, baseline recorded at zero, first article live on 26 May.
June 2026
Bookable hotel catalogue wired into generation; 49 articles published.
July 2026
Booking-intent prompt panel introduced; 39 articles published.
From August 2026
Monthly reporting, every published figure re-derived by independent verifiers.
Three results, each with its baseline and its scope
583,914 Search impressions · 45.0% of the section’s Search clicks · 3 → 9 mentions in AI answers
Each number below is stated against what it was before, and inside the scope it is true within. Where a comparison would need two different instruments to be joined into one line, it is not made.
Google Search impressions
583,914 on pages that did not exist
Impressions on articles that did not exist before 26 May 2026, alongside 6,706 Search clicks. Between June and July impressions rose 147% as the library grew from 70 to 109 pages, and average position improved from 9.7 to 8.2.
The section as a whole also improved its average position over the same months, so the position gain is not isolated to our pages.
Search impressions on pages that did not exist in April
Monthly Google Search impressions on the articles we produced
Google Search Console · property bild.de/reisen · Google Search only, Discover and News excluded · 14 May – 31 July 2026 · final data.
Share of the section’s Search clicks
45.0% of every click in the section
Of the 14,898 Google Search clicks earned across the whole travel section in the period, 6,706 went to pages we produced. The section’s other 382 measured pages took the remaining 8,192. On impressions our share is 51.5%.
Our share is a floor rather than an exact figure: the rule excluding URL-variant rows removes a far larger portion of our rows than of the rest of the section. It is a descriptive share of one period, not a per-page comparison — our pages were new and the section’s were established.
109 new pages against the section’s established 382
Google Search clicks across bild.de/reisen in the measured period
Google Search Console · one page-level pull, so both sides sit on the same measurement basis · 14 May – 31 July 2026.
Mentions in AI answers
From absent, to measurable, to tripling
On 14 May, before anything was published, the brand was mentioned in none of the tracked AI answers and sat 17th of 18 on the panel. Within the first stable measurement panel it reached three mentions. On the panel that ran through June it went from three mentions to nine, and from 10th to 9th, with a visibility score of 0.35 out of 100.
That score is reported on its own scale on purpose. Under half of one percent is a small number and we state it as one: the result is that a brand absent from this category of answer became measurable in it, not that it became dominant.
| Tracked panel | Window | Mentions | Visibility (0–100) | Rank |
|---|---|---|---|---|
| Panel A · 45 prompts | 14 May – 9 June | 0 → 3 | 0.00 → 0.07 | 17th → 14th |
| Panel B · 25 of those prompts | 10 – 30 June | 3 → 9 | 0.14 → 0.35 | 10th → 9th |
| Panel C · 45 booking prompts | 1 – 19 July | 20 → 17 | 0.36 → 0.32 | 10th → 9th |
The tracked panel was narrowed on 10 June and replaced outright on 1 July, so movement is only ever reported within a stable panel and the three are never joined into one trend. Changing the panel changes the denominator: mentions rose from 9 to 20 overnight on 1 July because the panel was replaced, not because anything improved. Panel C was built around the client’s own bookable inventory, so it is a purpose-built instrument and its opening reading is not a baseline. Measured on 56 days between 14 May and 19 July, with no readings between 15 and 25 May. Panel sizes are 18 rows at the start and 19 at the end, including BILD Reisen itself. Model coverage was ChatGPT, Google AI Overview and Perplexity.
Where does your brand stand when buyers ask AI?
Every engagement starts the same way this one did: with a reading taken before anything is published, on a panel of the questions your buyers actually ask. Run the free Health Check on your website, or book a call with our team.
Figures re-derived 17 August 2026 from Google Search Console, the AI-visibility snapshot series and the client content database, and independently reproduced by separate reviewers. Search results are scoped to Google Search; Google Discover and Google News are excluded and were the larger surfaces in this period. AI-visibility figures are reported per tracked panel and never joined across panels or measurement instruments.
