Site index for AI agents: /llms.txt. Documentation pages under /docs are also served as markdown at the same URL plus .md (e.g. /docs/geo-audit.md) or via an Accept: text/markdown request.

Case studiesB2B software

Employee rewards platform, US · Managed engagement

A rewards platform grew its US AI visibility 23% on fixed buyer questions

An employee rewards and recognition platform selling to US enterprises grew its AI visibility 23% on a fixed set of 33 buyer questions in seven weeks, and moved from 7th to 5th among 14 tracked brands. The gains landed in the three product clusters AirPulse prioritised, after an audit, two fix packs and a crawler-access fix.

+23%
AI visibility, 4.4 to 5.4 on 33 fixed US questions
7th to 5th
Market position among 14 tracked brands
87%
Of own-site citations from the 3 clusters we targeted

The challenge

The platform sells employee rewards, gift cards and perks to US enterprises, where review sites and listicles out-cite vendor sites about ten to one. It was rarely named when US buyers asked AI for a rewards platform.

What we found

On the same 33 US buyer questions, AI visibility rose from 4.4 to 5.4 (+23%) and brand mentions from 5.9 to 7.2 per daily run, Aug 4 to 10 vs Sep 22 to 28, 2026.
Market position improved from 7th to 5th among 14 tracked brands over the same windows.
87% of the citations to its own site came from the three clusters picked in the Jul 23 plan: employee perks and discounts, corporate gifting, and rewards.

What we did

2026-07-01: GEO and AI-search visibility audit with a scored remediation workbook.
2026-07-21: Applied the client team's 33 US buyer queries as the tracked prompt set.
2026-07-23: US analysis and page plan that put reward distribution and employee discounts first.
2026-07-29: Fix pack handed over (pricing-page structured data, security headers, an extended llms file, entity markup, specs for three content pages), with each item re-checked on the live site.
2026-08-04: Wikidata entity guide; the client created the entity from it on 2026-08-09.
2026-08-05: Second fix pack with schema refinements and blog-migration guardrails to keep cited pages stable.
2026-08-19: Re-audit found a new bot-protection challenge blocking non-browser clients, including AI crawlers, and canonical tags on 31 comparison pages pointing away from the pages. By 2026-09-15 the client had fixed the challenge and 30 of the 31 pages.
2026-09-15: Roster review recommending buyer-shaped rewrites of the 21 prompts that had never named the brand.

What worked

  1. 1Lock the prompt set and competitor roster before reading a trend. Swapping 5 broad prompts for 33 long-tail ones cut visibility from about 15 to about 3 on 2026-07-21 with no change in the market.
  2. 2Aim at clusters where a product page exists. Discounts, gifting and reward distribution earned 87% of own-site citations; definition prompts for adjacent categories earned almost none.
  3. 3Re-audit the live site after every release. A bot challenge and a canonical template bug can hide pages from AI engines without anyone noticing.
  4. 4In this category, review sites and listicles out-cite the brand's own site about ten to one; being listed and reviewed there moves many prompts at once.

Results

4.4 to 5.4
AI visibility
2026-08-04 to 2026-08-10 vs 2026-09-22 to 2026-09-28; 33 US prompts; ChatGPT, Perplexity, Gemini, Google AI Overviews
5.5% to 10.4%
Share of voice
Same windows and scope; competitor mentions fell 37% over the period
7.0 to 5.0
Market position
Among 14 tracked brands, 1 = best; same windows and scope
0.52% to 0.69%
Brand share of citations
Same windows and scope; 0.50% to 0.67% excluding Google AI

Where AI finds its answers in this category

Reddit0.5%
YouTube1.0%
Wikipedia0.1%
Review and marketplace sites6.3%
Listing portals2.2%
News and media1.7%
How we measured

The platform sells employee rewards, recognition, gift cards and perks to enterprises and tracks US buyer questions. On 2026-07-21 its team replaced 5 broad global prompts with 33 US long-tail queries, and on 2026-08-01 the competitor roster grew from 4 to 13 brands. This study starts on 2026-08-04, the first full run on both.

33 tracked buyer questions, 54 daily runs, Aug 4, 2026 to Sep 28, 2026, on ChatGPT, Perplexity, Gemini, Google AI Overviews. Published without the brand's name.

  • Share of voice rose from 5.5% to 10.4%, but most of that came from competitor mentions falling (99.4 to 62.5 per run), so visibility is the headline.
  • Google AI runs on 2026-09-12 to 2026-09-28 were hit by a scraper outage; the end window has 6 runs, and 2026-09-27 had none.
  • Share of voice and position are relative: competitor mentions fell from 99.4 to 62.5 per run, so they overstate the brand's own gain. Visibility, the brand's own mention rate, is the headline for that reason.
  • July figures (5 global prompts until 2026-07-20, a 4-brand roster until 2026-07-31) are not comparable with this window. Measurement lapsed from 2026-07-28 to 2026-08-03.
  • The client's team shipped the site changes; AirPulse audited, specified and verified them. The gain coincides with those fixes and new comparison pages; the data does not prove cause.
  • Source mix covers 2026-08-29 to 2026-09-28, all engines, US. 'Other' is mostly vendor product pages, analyst and research firms, and Google search links.

See where AI recommends you.

Book a free strategy call and we run your buyers' questions on the engines while you watch.