Case studies

Proof, with the
numbers attached.

Two kinds of evidence: client engagements we ran, and independent teardowns of categories we do not work in. Both use the same method.

Client work

Engagements where we ran the program and can show the before-and-after.

Teardown

Independent audits of public categories, sampled the same way we sample our own clients.

Client work B2B SaaS · Analytics

Northwind Analytics

Northwind: 6.4× AI citations in 120 days

B2B analytics platform was losing AI answers to two competitors. We rebuilt its entity layer, published a quotable corpus, and grew citations 6.4× in 120 days.

6.4×: Weekly AI citations across six engines
41%: Share-of-answer on 60 core buyer prompts
18%: Of trial signups now AI-referred

Client work DTC · Home goods

Loom & Ledger

Loom & Ledger: 34% lower CPA, 2.3× headroom

A DTC home goods brand scaled Google Ads on inflated platform ROAS. We rebuilt measurement, stripped branded-search credit, and found profitable growth.

34%: Reduction in blended CPA
2.3×: Increase in profitable spend headroom
94%: Of spend tied to a verified conversion path

Teardown B2B SaaS · CRM

CRM category

Teardown: five CRM brands in ChatGPT

We asked ChatGPT 40 buying questions across five CRM brands. Three took almost every mention, review sites took the citations, and no vendor published schema.

3 of 5: Brands named in any answer
61%: Citations pointing at review aggregators
0 of 5: Vendors publishing Organization schema

Teardown B2B SaaS · Payroll

Payroll category

Teardown: what Perplexity cites for payroll

Perplexity payroll teardown, 30 prompts: eight sources supplied most citations, aggregators outranked vendors, and one review site decided the winner.

8: Sources supplying 80% of citations
4 of 10: Citations from a single aggregator
1: Vendor page cited more than twice

How we measure

No cherry-picked screenshots.

Every number in these case studies comes from the same measurement spine we install for clients: a fixed prompt panel sampled weekly, citations logged by source domain, and revenue reconciled against the CRM.

Teardowns are labeled as such. They are snapshots of public data, not claims about client results, and they intentionally include findings that are unflattering to the category.

Read the measurement framework

Your turn

What would your numbers look like?

Bring us a category and a commercial target. We will baseline your visibility and tell you what is realistically winnable.

Run the GEO scan Book a strategy call [email protected]

Proof, with thenumbers attached.