Blue Light Card, AIR SEO case study
Whitworth SEO, supporting evidence for 2026 award entries
GSC and GA4 via SEO Gets
AI ready, not AI first
Answering the call. AI Ready SEO for the UK’s key workers
One AIR SEO strategy, run through two lenses, took Blue Light Card from absent in AI answers to number one AI share of voice. The same work nearly doubled organic sessions and more than quadrupled organic sign-ups. No separate GEO budget. Just SEO done properly, made AI ready.
The campaign at a glance
Sign-ups grew nearly four times faster than sessions
1 January to 31 August 2026 against the same period of 2025 unless labelled. Figures in blue are AI search.
The client and the challenge
If “SEO is dead” were true anywhere, it would be true here
Blue Light Card is the UK discount platform for the NHS, emergency services, social care, teaching, carers, and armed forces communities. Millions of verified members, and a business model that runs on two things, new member sign-ups and members returning to redeem offers.
Organic performance was steady, but it was built almost entirely on brand, TV, offline, and social. Brand terms already sat at an average position of 1.53, so brand was never the problem. The commercial opportunity was non-brand, across thousands of “[brand] discount code” queries where voucher aggregators with two decades of authority ranked and Blue Light Card did not.
Then AI moved in. A discount code query is close to the perfect use case for an AI answer. The user wants one fact, whether a code exists and whether they qualify. AI Overviews, AI Mode, ChatGPT, Perplexity, Gemini, and Claude are built to answer that without a click, and early prompt audits confirmed the aggregators were the answer, not Blue Light Card.
The client was understandably cautious about investing in non-brand SEO, because the size of the opportunity had never been mapped. Our research phase mapped it, from NHS and emergency service terms to thousands of partner brand discount code terms, and forecast the traffic and revenue impact using our methodology fused with Blue Light Card’s own value model.
Objectives against results
Every target beaten, on one retainer
Targets were agreed with the client against real baselines. Sign-ups were the primary KPI, because Blue Light Card runs on members, not sessions.
| Objective | Target | Result | Outcome |
|---|---|---|---|
| Double organic sign-up events, Jan to Aug | +100% | +347% | 2.2x target |
| Grow non-brand clicks, August | +50% | +79% | Beaten |
| Triple LLM referral sessions, Jan to Aug | +200% | +1,389% | 4.7x target |
| Grow AI sign-up events tenfold, Jan to Aug | +900% | +9,041% | 9x target |
| Top three AI share of voice on priority prompts, from absent | Top three | #1 | Every prompt set |
| AI referral value covers the monthly retainer by year end | December | July | Five months early |
Value measured on Blue Light Card’s own model for members and logins. No media spend, no AI tooling budget, and no separate GEO line item.
The approach
AI search is a layer, not a channel
AIR SEO treats AI search as a retrieval layer sitting on top of organic search. Weak foundations produce weak AI visibility. Strong foundations are already most of the work. So we ran one plan through two lenses, from the same brief, with every deliverable serving both surfaces.
Traditional search
9,365 keywords mapped by intent, clustered, and prioritised for non-brand. Search volume, difficulty, relevance, and weekly rank tracking across a keyword universe of 3,000+ intent clusters.
AI retrieval
Fan-out queries, passage level extractability, entity depth, and prompt level visibility across AI Overviews, ChatGPT, Gemini, Perplexity, and Claude, informed by Waikay and Keyword Insights.
Three decisions carried the strategy
Eligibility is the moat
Any aggregator can publish a code. None can verify a key worker. Commercial pages were built around verified eligibility, the one answer competitors structurally cannot give, with an eligibility block written for retrieval.
Honesty is the differentiator
Where no code exists, the page says so and explains the Shopping Card route instead. It satisfies the query honestly, which is what E-E-A-T measures, and gives AI an answer it can retrieve without contradiction.
Fan-out reshaped the architecture
AI splits a “[brand] discount code” prompt into four intent clusters, and none of them is NHS. So member offers moved inside mainstream code intent rather than sitting on separate NHS pages. That one finding reversed the site’s existing approach.
Blue Light Card wanted to move fast, with AI generated pages at scale across the keyword universe we had built. Being that aggressive in 2026 carries real algorithmic risk. Our counsel was to scale safely instead. Clusters were prioritised on volume, difficulty, relevance, and AI search potential, and the AIR SEO brief became mandatory input to the client’s page automation, with human review, cannibalisation control, and genuine template variation. Where a partner code didn’t exist yet, our data became market intelligence Blue Light Card used to build those partnerships, while Shopping Card pages captured the demand in the meantime.
What we did
Delivered through RARER, Whitworth’s methodology
Every technique below is classic SEO. What changed is the lens. Each one was specified, briefed, and QA’d for extractability and retrieval as well as rankings.
Find the gaps AI exposes
- Prompt audits across ChatGPT, Gemini, Perplexity, and Claude surfaced five sitewide gaps. No offer hub, no freshness or verification cues, no redemption block, no social proof, and weak audience clarity.
- Those five became the template specification.
- Monthly cannibalisation audits consolidated fifteen orphaned NHS variant URLs.
Fix the foundations at scale
- Schema rebuilt across templates. Product replaced with Offer and OfferCatalog, plus MobileApplication, Speakable, and sameAs.
- URLs with valid schema up 460%.
- 3,072 orphaned URLs triaged, 2,551 legacy redirects mapped, 18 hubs reviewed, and a 174 URL internal link plan built.
- AI citations checked against orphans before any redirect shipped.
Brief for people and passages
- A 62 piece content programme mapped to customer moments, with 41 AIR SEO briefs delivered.
- Each pairs keyword intent with fan-out themes, direct answer leads, entity depth, and E-E-A-T.
- Passages written to stand alone when retrieved. Human led throughout.
- 58 new pages proposed from cluster data.
One implementation stream
- SEO and AI search shipped together, never siloed.
- The AIR SEO brief made mandatory input to page automation, with human review.
- A blog template specified with TL;DR blocks, Written By and Edited By credentials, sticky contents, and a one tap AI summary bar.
Judged on sign-ups
- Looker Studio, GA4, Search Console, Sitebulb, and a weekly visibility index.
- LLM referrals reported as their own GA4 channel.
- Prompt level share of voice used to steer, never as the KPI.
- The May core update was diagnosed as flux. The strategy held and clusters recovered in full the following month.
Performance, the SEO story
Organic nearly doubled as a channel, and converted harder as it grew
Growth, Jan to Aug 2026 vs 2025
Growth outpaced traffic rather than riding it. That rules out the “they just got lucky with demand” read.
Organic took the acquisition load
Organic’s share of every sign-up on the site more than doubled (to July). Brand clicks also grew 10.8% in August, so non-brand growth was additive, not a mix shift.
| Growth, Jan to Aug YoY | Change |
|---|---|
| Organic sign-up events | +347% |
| Organic impressions | +284% |
| Organic clicks | +200% |
| Visibility index | +137% |
| Organic sessions | +93% |
| Non-brand clicks, August | +79% |
| Brand clicks, August | +10.8% |
| Before and after | 2025 | 2026 |
|---|---|---|
| Sign-ups per organic session | 2.29% | 5.30% |
| Organic share of all site sign-ups (to July) | 7.6% | 16.4% |
| Average position | 33.5 | 9.3 |
| Non-brand average position, August | 18.0 | 4.98 |
| Visibility index | 1.79 | 4.25 |
| Top three keywords, of 9,365 tracked | 4,392 |
Topic clusters moved from page two or three to the top five
| Topic cluster (August) | Clicks YoY | Position 2026 | Position 2025 |
|---|---|---|---|
| Travel & Transport | +102% | 4.8 | 12.8 |
| Holidays & Accommodation | +142% | 4.5 | 22.5 |
| Home & DIY | +282% | 4.6 | 22.8 |
| Restaurants & Takeaways | +88% | 5.9 | 13.9 |
| Supermarkets & Groceries | +330% | 5.8 | 26.0 |
| Beauty & Health | +204% | 5.6 | 29.3 |
Search Console topic clusters where fan-out led architecture and AIR SEO briefs moved average position into the top five.
Pages the campaign built now lead the account
| Query or page | Clicks YoY | Detail |
|---|---|---|
| railcard discount code | +111,200% | Dedicated page built from cluster data, now position 3.56 |
| chester zoo discount code | +103,000% | Days out, one of the strongest partner verticals |
| ocado discount code | +63,300% | Backed the new Supermarkets & Groceries master category |
| travelodge discount code | +49,900% | Part of the Holidays & Accommodation cluster |
| john lewis discount code | +19,783% | Briefed through the AIR SEO content planner |
| dunelm discount code | +7,218% | Briefed through the planner, top non-brand query at position 2.12 |
| premier inn discount code | New | From a standing start to position 3.55 |
| Eligible professions page | +3,228% | Nine and a half minute average sessions |
Query level growth for July 2026 against July 2025 from Search Console, with the railcard figure updated to August.
Research finds it, Adapt structures it, Refine briefs it, Execute ships it, Report proves it and finds the next one.The railcard cluster, 76,140 monthly searches, was ranking through a generic savings page. A dedicated Travel & Transport architecture turned it into one of the fastest growing queries on the account.
Visibility, the AI search story
Number one in the AI answer, and it converts
The thesis was that the same work would move the retrieval layer. It did, and the timing is the evidence. LLM referral growth tracked the non-brand SEO curve, not a separate GEO workstream, because there wasn’t one.
From visibility to members, Jan to Aug YoY
Visitors who sign up, not curiosity clicks
LLM referral sessions grew 1,389% and AI sign-up events 9,041%, so AI visitors are converting faster than they are arriving. ChatGPT drives the large majority of sessions, with Gemini, Copilot, Perplexity, Brave, and Claude in the long tail.
On the client’s own value model, AI referrals covered the full monthly retainer in July, five months ahead of the year end target.
What this proves
SEO isn’t dead. The prize for doing it properly just got bigger
- AI search is a layer, not a channel. One holistic AIR SEO strategy beat the idea of a separate GEO workstream, in the sector most exposed to AI summarisation.
- SEO fundamentals drive AI visibility. Schema, architecture, extractable content, and E-E-A-T took Blue Light Card to number one AI share of voice on the highest volume terms in the category.
- Revenue is the KPI. Organic sign-ups up 347% and AI sign-ups up 9,041%. That is the headline, not citation counts.
- Honesty wins. Telling key workers the truth about codes gave Google and LLMs an answer aggregators cannot copy.
“The prudent strategy is AI ready, not AI first.”Charlie Whitworth, on stage at Search ‘n Stuff London 2026, with Blue Light Card as the proof
And this is only month nine. Site structure, schema, and template changes are still in the development queue, the new blog template is yet to ship, and reporting has already surfaced the next opportunity, a carers head term at 90,500 monthly searches with demand up 308% year on year.
The team
Seasoned consultants, doing the work themselves
Charlie Whitworth
Founder, strategy lead
16+ years in SEO, with a journalism background and senior roles including Head of SEO at Missguided and SEO lead at Halfords. Creator of AIR SEO and RARER, organiser of HiveMCR.
Rachel Deer/Whitworth
Content strategy and delivery
15+ years of ecommerce SEO and content strategy. Leads AIR SEO brief delivery and monthly reporting on the account.
Harrison Lemon
Technical SEO
8+ years of technical SEO at agencies including MediaCom. Leads the schema, architecture, and crawl programme.
From client work to conference stage
We argued it on stage before the data was in
At the HiveMCR AI Search Masterclass, Charlie argued the retrieval layer means AI search should not be treated as a channel. At Search ‘n Stuff London, he walked through the exact techniques used here. The Blue Light Card charts on those slides are not illustrations. They are this campaign.
- How Poor Migrations Are Destroying AI & SEO VisibilityArticle, Aug 2026
- What We’ve Learned From 6 Months of AI Ready SEO CampaignsArticle, Aug 2026
- The Upcoming Cloudflare Changes, And How They Impact SEO & AI SearchArticle, Jul 2026
- The Top SEO Techniques That Are Still Very Much Alive in 2026, Search ‘n Stuff LondonTalk, Jun 2026
- AI Tools Are Killing Creativity in SEOArticle, May 2026
- AI Search Optimisation Masterclass, HiveMCR 26Talk, May 2026
About the figures
Search Console and GA4 data via SEO Gets, comparing 1 January to 31 August 2026 with the same period of 2025 unless labelled. Organic sessions are search engines with medium “organic”. Sign-up events are the GA4 sign_up_button_click event, a registration start rather than a completed membership. LLM referrals are sessions from ChatGPT, Gemini, Perplexity, Copilot, Claude, and other AI assistants, reported as their own GA4 channel. AI share of voice and knowledge score via Waikay, AI Overview tracking across 1,000 keywords, and rankings across 9,365 tracked keywords. Shared with Blue Light Card’s permission for award judging.
