Wealth-Tech Platform Wins the Generative Search Layer: A 6.4x Lift in AI Citations
Client identity protected under NDA. Details available under mutual sign-off in a discovery call.
Composite company profile
The client is a US-headquartered wealth-tech platform serving independent RIAs and mid-market wealth management firms. Series C, roughly $42M ARR at kickoff, ACV band $28K to $180K depending on AUM tier. Core buyer: Head of Digital Wealth, CTO, or COO at firms managing $500M to $8B in client assets. Their category is regulated adjacent (SEC and FINRA touchpoints), so every marketing asset ships through a compliance review with a 5 to 7 business day SLA.
The problem
Inbound demo requests had plateaued for three consecutive quarters. The marketing team had a respectable classic SEO footprint (domain rating in the low 60s, 4,200 ranking keywords), but pipeline attribution kept surfacing a strange pattern. High-intent buyers were arriving already educated, often citing product comparisons and category definitions that the client had never published. Sales calls opened with, “I saw you were the top choice for portfolio rebalancing automation, is that right?”
When the RevOps lead ran the buyer survey, the answer became uncomfortable. Forty-one percent of new opportunities said they had first heard the client mentioned inside ChatGPT, Perplexity, or Google’s AI Overviews. Yet a manual audit showed the client was cited in fewer than 8 percent of the 60 highest-intent generative queries in the space. Competitors with weaker classic SEO were showing up more often in the AI answer layer. Prior efforts had failed because the in-house team was optimizing for the ten blue links, not for the passage extraction, entity clarity, and citation-friendly formatting that large language models actually reward.
What our team diagnosed
Our SEO and analytics team ran a Generative Engine Optimization audit across five surfaces (ChatGPT, Perplexity, Gemini, Claude, and Google AI Overviews) and mapped citation share for 78 category-defining prompts. Three root causes emerged that were not obvious to the client:
- Entity ambiguity. The client’s brand string collided with a similarly named crypto product. LLMs were hedging or defaulting to the more established (and off-topic) entity. Their Wikipedia stub was thin, their Crunchbase entry was outdated, and their sameAs graph in schema was incomplete.
- Passage structure penalized extraction. The best product pages were built as long-form marketing narratives. Great for humans, poor for LLM chunking. Definitions, use-case boundaries, and comparison tables were buried inside paragraphs instead of surfaced as extractable answer units.
- Missing category coverage. Twenty-two high-value informational queries had no dedicated URL. Competitors owned the answer because the client had no page to cite.
Strategy MV3 shipped
Vance oversaw the engagement and our SEO, analytics, and content teams executed a 90-day Generative Engine Optimization sprint layered on top of the client’s existing SEO program. Services engaged: GEO Audit, technical SEO remediation, content production (composite persona bylines with reviewed-by credentials from real practitioners at the client), and llms.txt authoring. We deliberately did not touch paid media, because inbound quality was the constraint, not volume.
Key insight we sold in: winning the generative layer is not the same problem as winning classic SERPs. It requires entity disambiguation first, extractable passage engineering second, and citation-friendly content architecture third. Ranking movement is a lagging byproduct.
Implementation
Over 14 weeks our team produced and shipped:
- Full entity cleanup: rewrote and expanded the client’s Wikipedia stub through proper editorial channels, updated Crunchbase and G2, standardized the Organization schema with a complete sameAs array, and added Product schema to every pricing tier.
- 22 new category pillar pages, each built to a strict passage template: 40 to 60 word Quick Answer block above the fold, definition, use-case boundaries, comparison table, FAQPage schema, and a “reviewed by” byline pointing to a real subject matter expert at the client.
- Retro-conversion of 34 existing product and comparison pages to the same passage template. This was the heaviest lift and required tight coordination with the compliance team.
- A hand-authored llms.txt file mapping the client’s canonical documentation surface, published at the root and referenced from robots.txt.
- Weekly generative citation tracking across the five LLM surfaces with a Supabase-backed dashboard so their team could watch citation share move in near real time.
Outcomes
Measured at week 16 against a baseline snapshot taken in week 0:
- AI citation share: from 8% to 51% across the tracked 78 high-intent prompts. Primary metric. A 6.4x lift.
- Demo requests self-reporting an AI-surface origin: from 41% to 63% of new opportunities.
- Total inbound demo volume: up 74% quarter over quarter, with no incremental paid spend.
- Sales cycle length on AI-sourced opportunities: compressed from a 71 day median to 48 days, because buyers arrived pre-educated on the client’s positioning against the two main competitors.
- Blended CAC on new logos: down 29% versus the trailing two quarters.
The classic SEO footprint improved as a secondary effect: ranking keywords grew from 4,200 to 5,880 and 14 previously page-two terms moved into positions 3 through 7. Not the goal of the engagement, but a welcome side effect of the passage engineering.
Timeline
Kickoff to first citation share lift: 6 weeks. Kickoff to primary outcome measurement: 16 weeks. Total engagement including a 4-week stabilization tail: 20 weeks.
Composite testimonial
“We had been treating AI search as a novelty. When 4 out of 10 new opportunities started telling us they’d found us through ChatGPT, we realized the answer layer was already the top of our funnel, and we owned almost none of it. The GEO work moved the number that actually mattered.”
Priya, VP of Marketing
NDA framing
Client identity protected under NDA. Composite profile, verified numeric outcomes, and full methodology available under mutual sign-off in a discovery call.
Ready to win the generative layer in your category?
If your buyers are showing up pre-educated by AI answers you didn’t write, the answer layer is already your top of funnel. Book a discovery call or start with a GEO Audit to see your current citation share across ChatGPT, Perplexity, Gemini, Claude, and Google AI Overviews.
