Home / Case studies / roadsurfer
Case study 01 — analyst story — roadsurfer, US blog

Growth exposeda harder problem.

I owned SEO and content for roadsurfer's US blog from March 2025. Every figure here carries its window and source. The ledger at the end lists them.

Scope: strategy, keyword research, briefs, drafting and refreshes, WordPress publishing and schema, internal linking, GSC and GA4 reporting, localisation screening, local SEO for stations.

+91%
non-brand organic clicks
+219%
non-brand impressions
20×
keywords ranking top 3 · 22 → 449
1,358
AI citations tracked
01The measure

The growth is real.
So are its boundaries.

Same 13 weeks, one year apart, US searchers only: non-brand clicks 1.24K → 2.37K, impressions 89.3K → 285K, average position 14.9 → 9.5. See the capture ↗

NON-BRAND CLICKS JUN–AUG 2025JUN–AUG 2026 NON-BRAND IMPRESSIONS JUN–AUG 2025JUN–AUG 2026 1,2402,370 89.3K285K

Google Search Console · Country = United States · pages /blog/ · brand queries excluded · 1 Jun – 31 Aug 2025 vs 1 Jun – 31 Aug 2026.

The rankings moved up the page.

MAY 2025AUG 2026 POSITIONS 1–322 POSITIONS 4–1076 POSITIONS 11–20178 44987166

Ahrefs · roadsurfer.com/blog/ · US · monthly snapshot · May 2025 vs Aug 2026. Top-three peak: 622, June 2026. Bands below 20 omitted (see limits). See the capture ↗

Most of the blog now ranks on page one.

Pages with a top-ten ranking: 16 in February 2025, 75 in August 2026, out of 94 published posts (71 of them written since March 2025). Twelve of today's 15 highest-traffic pages had no ranking at all in May 2025.

PAGES WITH AT LEAST ONE TOP-10 RANKING · US · MONTHLY 94 PUBLISHED POSTS FEB 2025MAY 2025AUG 2026 16 26 75
TWELVE OF TODAY'S 15 HIGHEST-TRAFFIC PAGES · NO RANKING IN MAY 2025

rv-classes · rv-rental-prices · class-b-vs-class-c-rvs · seattle-to-san-francisco · pet-friendly-rv-rental · salt-lake-city-to-yellowstone · 24-best-rv-gift-ideas · route-66-road-trip-itinerary · pop-up-camper-rentals · 7-day-colorado-to-california · rv-insurance · one-way-rv-rental

One refresh · /rv-rental-vs-hotel/The guide went from 4 to 397 estimated monthly visits and 17 to 107 keywords, and ranks #3 for “rv vs hotel calculator”.
4 → 397
est. monthly visits
17 → 107
keywords
#3
for “rv vs hotel calculator”

Ahrefs · US · pages ranking top 10 · top pages 1 Sep 2026 vs 1 May 2025 · WordPress post inventory 11 Sep 2026.

02The finding

Better rankings did not
produce a better click rate.

Average position improved by 5.4 places while CTR fell from 1.4% to 0.8%, a 43% decline. That divergence was the useful finding.

AVERAGE POSITION (SOLID, LEFT AXIS INVERTED) · CTR (DASHED) JUN–AUG 2025JUN–AUG 2026 TWO LINES THAT SHOULD MOVE TOGETHER, MOVING APART AVERAGE POSITIONIMPROVED 5.4 CTRDOWN 43% 14.99.5 1.4%0.8%

Google Search Console · US · matched 13-week windows · brand queries excluded. See the capture ↗

The diagnosis 205K impressions inside Google's generative-AI features for US blog pages in the six months to 11 September 2026, the RV rental prices guide alone 36.8K. The report launched after the anomaly, so it confirms the diagnosis after the fact rather than proving it in advance.

Caveat: query mix can also change. A future programme could separate branded, informational and page-level anomalies.

03The response

Optimise for the answer layer,
not only the blue link.

Clearer entities, extractable answers, stronger evidence, source-worthy page structure. By 24 August 2026: 1,358 citations across all tracked platforms, 281 roadsurfer pages cited.

Ahrefs Brand Radar · US Brand Overview · all platforms (six platforms, eight indexes) · 24 Aug 2026. Prompt-only subset: 646. Definitions in the ledger.

CITATIONS TRACKED · AUG 2026 APR 2025SEP 2026 MONTH-END VALUES · ALL PLATFORMS 1,328

Ahrefs Brand Radar · US · all platforms · month-end · exported 11 Sep 2026. The drawdown happened with no change to the pages. See the capture ↗

PAGES CITED ROADSURFER PAGES APPEARING AS A SOURCE TOP GUIDE — “YOURS” VIEW: CITED IN 362 OF 740 SIX PLATFORMS TRACKED CHATGPT · AI OVERVIEWS · AI MODE GEMINI · PERPLEXITY · COPILOT · GROK PAUSED 281

Ahrefs Brand Radar · cited pages · 27 Aug 2026. Grok paused; Claude not in the dataset. See the capture ↗

The most-cited source on RV rental prices in US AI answers.

Ask an AI assistant what an RV costs to rent in the US and /rv-rental-prices/ is the page it cites most: 664 responses, ahead of every competitor. The cited pages are almost all blog pages; the product pricing page gets 23.

RESPONSES CITING THE PAGE · ALL-DOMAINS RANKING 42 ROADSURFER URLS SIT AMONG THE 400 MOST-CITED PAGES IN THIS PROMPT SET roadsurfer — RV rental pricesRVshareHomeGuide Cruise AmericaEl Monte RVOutdoorsy 664429349 231197182

Ahrefs Brand Radar · US · cited pages by responses · 27 Aug 2026. Competitor figures are public Ahrefs data. The “Yours” view counts the guide in 740 responses, cited in 362. 'Yours' view ↗

Where the citations sit.

CITATIONS BY PLATFORM · PROMPT-ONLY SUBSET · 570WINE = GOOGLE SEARCH'S AI OVERVIEWS AND AI MODE · 319 OF 570 · 56% AI Overviews167 AI Mode152 Gemini82 Copilot59 ChatGPT55 Perplexity55

Ahrefs Brand Radar · prompt-only subset · US · 11 Sep 2026. Google Search's AI Overviews and AI Mode carry 56%: that is the answer layer. See the capture ↗

A citation is not a visit, and platform coverage changes. I use it beside rankings, impressions, clicks and sessions, not as a replacement for them. How the citations were tracked ↗

04The operating system

Sixteen workflows,
with people at the gates.

Sixteen workflows in production, from a library of 32, across eight blog locales: research, briefs, drafting, refreshes, QA, CMS formatting, localisation screening, site checks. About seven hours saved per article. People kept three decisions: topic, QA verdict, publish.

RESEARCHDRAFTINGQAPUBLISH KEYWORDS · BRIEFSREFRESHESLOCALISATIONCMS · SITE CHECKS WINE BLOCKS ARE HUMAN DECISION GATES — TOPIC, QA VERDICT, PUBLISH 16 OF 32 WORKFLOWS IN PRODUCTION · 4 STAGES · 8 BLOG LOCALES

One run of the pipeline, measured at day 21.

“12 Camping Gadgets Actually Worth Packing”, live 20 August 2026. Brand queries excluded.

8.3
“camping gadgets” · 815 impressions · 4% CTR
9/10
top queries by clicks, on page one
5.3%
CTR at avg position 9.7 · 190 queries
68%
engaged · 1,909 organic users · 21 days
QUERYIMPRESSIONSPOSITION 815255621671198 11.8 10 — PAGE ONE BOUNDARY camping gadgetscamping gadgets 2026best camping gadgets 2026 best camping gadgetscool camping gadgetscampervan gadgets 8.36.85.39.78.5

Google Search Console · queries · page = /blog/camping-guides/camping-gadgets/ · brand queries excluded · 3-month window, page live from 20 Aug 2026. Six of the ten highest-click queries shown, of 190 total. See the capture ↗

SESSION SOURCE — DIRECT EXCLUDED 67.6% ENGAGED AI ASSISTANTS — FIRST REFERRALS NINE USERS — A CITATION SIGNAL, NOT TRAFFIC google chatgpt.com 5 · bing 2 · perplexity.ai 2 1,909

GA4 exploration · page path × session source · 20 Aug – 9 Sep 2026 · direct excluded, unattributed. See the capture ↗

What one run shows Re-cut when a 90-day window exists; my access ended before day 90. If the head term slips, this block will say so.

GSC totals ↗ · Read the article ↗ · See the page ↗

The same discipline, applied to the CMS itself.

One workflow in the library moved published posts from the classic editor to the block editor. I built and ran it for the Italian and Spanish blogs. The German, French and Dutch posts were migrated separately by a colleague and are shown for scale, not claimed. Each post was snapshotted before any change and checked eight ways after it — text length, images, captions, anchors, one H1, empty headings, editor label — before it counted as migrated.

168
verified migrations, all markets · 152 by my workflow (IT + ES) · 16 Jul – 13 Aug 2026
9.9
posts per active day · 17 active days
38% → 51%
posts on the block editor · 1,328 posts · 8 markets
−75%
Italian backlog 53 → 13 posts · 12–26 Aug
VERIFIED MIGRATIONS BY MARKET ITALIANSPANISHGERMAN FRENCHDUTCH UNDATEABLE IT + ES · MY WORKFLOW DE, FR, NL · COLLEAGUE, SEPARATE PROCESS 77759 61 20

WordPress · revision history · a classic → block transition in a post's own revisions dates its migration · 16 Jul – 13 Aug 2026. Open the capture ↗

ITALIAN BLOG · CLASSIC-EDITOR POSTS REMAINING 12 AUG18 AUG26 AUG −75% IN 14 DAYS · MEASURED ON THE DATES MARKED 532513

WordPress · REST headcount on 12 Aug, daily run logs after · counts by stored content, not editor label. Open the capture ↗

A migrated post is a post, not a visit. The Italian and Spanish counts say the workflow held at volume; the checks say nothing broke on the way.

Counting “modified since date” overstated US migrations 85 to 1, so I built the revision audit before reporting a number. About 45 minutes saved per post.

Handed over.

Nine of the workflows were packaged, validated against company skill governance and handed over to the team with a stage-gated runbook and how-to pages written for a non-specialist successor.

05What I would do differently

Instrument earlier.
Escalate sooner.

Track citations before optimising, with stable prompts and definitions. Push for specialist tooling earlier. Separate query-mix effects from page-level CTR anomalies from day one.

Version from day one. The pipeline began in Copilot Studio; moving it to versioned workflows surfaced four structural bugs. The governance rules came from those scars.

06Sources & limits

Every number,
with its window.

The charts above are drawings of this data. Each evidence tag opens the unedited export or screenshot it came from.

Claim Source Window Evidence
Clicks +91% · 1.24K → 2.37K GSC · Country = US · page contains /blog/ · query excludes roadsurfer 1 Jun – 31 Aug 2025 vs 1 Jun – 31 Aug 2026 · 13 weeks First-party ↗
Impressions +219% · 89.3K → 285K GSC export and filters Same matched windows First-party ↗
Position 14.9 → 9.5 · CTR 1.4% → 0.8% GSC export and filters Same matched windows First-party ↗
Same claims, original window: clicks +111% · impressions +334% · position 15.4 → 9.4 · CTR 1.9% → 0.9% GSC · all countries · page contains /blog/ · query excludes roadsurfer 8 May – 8 Sep 2025 vs 2026 · 4 months First-party ↗
Top 3: 22 → 449 · positions 4–10: 76 → 871 Ahrefs · roadsurfer.com/blog/ · prefix · US · monthly snapshot May 2025 / Aug 2026 Tool data ↗
1,358 AI citations · all platforms, prompt + search-query datasets Ahrefs Brand Radar · US Brand Overview · citations 24 Aug 2026 Tool data ↗
281 pages cited · one guide in 362 of 740 responses · rv-rental-prices guide cited in 664 responses vs RVshare 429 Ahrefs Brand Radar · cited pages 27 Aug 2026 Tool data ↗
Citation series · month-end 28 → 1,328 · peak 1,421 on 9 Aug · −22.8% drawdown · platform split 570 (prompt-only subset) Ahrefs Brand Radar · US Brand Overview · all platforms · daily export; subset via API 30 Apr 2025 – 11 Sep 2026 Tool data ↗
Pages on page one 16 → 26 → 75 · 94 published US posts, 71 since Mar 2025 · 12 of top 15 new since May 2025 Ahrefs · pages with ≥1 top-10 US ranking · top pages by est. traffic · WordPress REST post inventory Feb 2025 / May 2025 / Aug 2026 · 1 Sep 2026 · 11 Sep 2026
168 migrations · 188 upper bound · IT 77 · ES 75 (my workflow) · DE 9 · FR 6 · NL 1 (colleague, separate process) WordPress · revision history · classic → block transition per post 16 Jul – 13 Aug 2026 First-party ↗
Italian backlog 53 → 13 classic posts · −75% WordPress · REST headcount + daily run logs 12 – 26 Aug 2026 First-party ↗
205K generative-AI impressions · rv-rental-prices 36.8K · 120 pages GSC · generative-AI features beta · Country = US · page contains /blog/ 18 May – 11 Sep 2026 · 6 months First-party ↗
273K generative-AI impressions, all countries GSC · generative-AI features beta · page contains /blog/ 27 May – 25 Aug 2026 · 3 months First-party ↗
Specimen: head term at 8.3 · 5.3% CTR · 190 queries · 9 of 10 top queries on page one GSC · page filter, brand excluded Page live 20 Aug 2026 First-party ↗
Specimen: 1,909 organic users · 67.6% engaged · 9 AI-assistant users GA4 exploration · session source 20 Aug – 9 Sep 2026 First-party ↗
About seven hours saved per article cycle · itemised: research and brief ~1.5 h · draft ~1.5 h · fact check and QA ~1 h · CMS draft build ~2 h · linking, tagging and reporting ~1 h Per-stage estimates from the workflow documentation; draft build ~2 h, migration ~45 min per post and progress report ~30 min per request are documented, the rest is my estimate Programme period Itemised estimate

Limits. GSC and Ahrefs are both US only. Headline GSC figures are US only and 13 weeks; the original four-month, all-country reading (+111% / +334%) is kept in the ledger with its capture, since 16-month retention has since aged out its start date. Ahrefs traffic figures are estimates. Citations: a citation is an AI response that used a roadsurfer.com URL as a source; the headline and series use the full Brand Radar view (six platforms, eight indexes), the platform split uses the prompt-only subset (646 on 24 Aug 2026); the index is revised backwards, so every figure carries its export date. Ahrefs stopped reporting positions below ~50 from late 2025, so keyword totals are not comparable across the period; top-10 counts are. RV demand is seasonal; autumn readings sit below the summer peak. Direct traffic on the specimen page is excluded. Not claimed: revenue, bookings, share of voice, colleague quotes.