Customer story: TestGorilla

Automating content intelligence and content production with Claude

How 2,831 sales calls and 230 weighted recruiter signals gave a new product launch a content strategy built on what the market was actually asking for, and a system that kept running after the engagement ended.

Client

TestGorilla

Industry

HR technology, skills-based hiring

Region

Amsterdam, global customer base

Contacts

Mehak Chowdhary, Head of Marketing. Estee Chaikin, Head of Brand Engineering.

Engagement

Buyer research, AEO and SEO content strategy, and an AI visibility monitoring system for the launch of a new Sourcing product

Engagement status

May to September 2026

Headline result

A content strategy built from market signals scraped fortnightly, turning what recruiters are actually asking into a validated, prioritised content pipeline the marketing team runs on its own.

Challenge

A new product, a narrow window, and no way to hear the market

TestGorilla was taking a new Sourcing product into a global talent market it sizes at $500 billion, on a compressed launch schedule. That was a real opportunity for a fresh story, built on what the market was asking for right now. Three things had to exist first, and none of them did.

A way to hear the market directly. A corpus of 2,831 recorded sales calls and public forum discussion, both untapped as sources of content insight.
A way to stay ahead of where the market was heading, rather than answering the questions it was asking six months ago.
A way to see what was actually working. Without AI visibility insight there was no way to know whether content was paying off, or where to focus next.
And a back catalogue nobody could see. 66 sourcing articles were already published. An audit found 64 of them generating zero organic traffic. Without a page-by-page diagnosis, the default plan was to publish more of the same.

How topics were chosen before

Source of truth
What the team assumed recruiters cared about.
Evidence
None. 2,831 sales calls unopened, forums uncollected.
Validation
None. Topics went straight into writing.
Measurement
Google Analytics only.
Result
66 articles published. 64 with zero organic traffic.

How topics are chosen now

Source of truth
What recruiters actually said, in their own words.
Evidence
Market signals scraped fortnightly from sales calls, Reddit, LinkedIn, G2 and Capterra.
Validation
A hard gate. No demand and difficulty data, no topic.
Measurement
Live performance dashboards tracking visibility across Google, ChatGPT, Gemini and Perplexity.
Result
40 validated pieces, sequenced by what ranks fastest.

Solution

One connected system, built to be handed over

Three layers, each answering one of the three gaps, designed from the first week to be run by TestGorilla's own team rather than by a consultant indefinitely.

A research layer that hears the market directly

AI-assisted mining of 2,831 sales calls, 94 of them confirmed to carry sourcing pain, plus scraping of Reddit, LinkedIn, G2 and Capterra. Everything normalises into one live signal dataset that grows with every fortnightly run, weighted by an explicit formula so themes rank on the strength of the evidence behind them.

A content layer that turns pain into validated topics

Ranked pain themes translated into the queries recruiters actually type and the longer questions they ask an AI assistant, every one validated for real demand and checked against what competitors already rank for. Output: four content pillars and a prioritised pipeline of 40 pieces, 34 new and 6 refreshes, sequenced across three waves.

A monitoring layer that shows where to focus next

AI visibility tracked from a zero baseline across six AI assistants, combined with Search Console, GA4 and competitor publishing activity in a single read-only dashboard. The first competitive baseline was uncomfortable and useful: TestGorilla sat seventh for sourcing answers, behind two direct competitors.

The handover

The real deliverable was independence

The question that shaped the final phase was not about the research. It was how the marketing team could run all of this without me.

✓Packaged as an installable plugin: ten commands and four skills, config-driven so nothing depends on one person's machine or logins.
✓Scraping runs on a fortnightly schedule and writes to a shared signal sheet the skills read from.
✓Every new cycle deduplicates against what has already been collected and already been planned, so nobody pays twice to answer the same question.
✓Documented with an operating guide written for someone who has never seen the system, a testing runbook, and an explicit open-items list naming what is unfinished and who owns it.
✓Ten days after the final day, the team ran the workflows themselves, diagnosed a credential failure from the automated alert and reassigned ownership internally.

"Working with Natalia was smooth, efficient and refreshingly hands-off for us: she built the system carefully incorporating stakeholder feedback, gave data-driven insights on what to publish, and kept it current as our market moved."

Estee Chaikin, Head of Brand Engineering, TestGorilla

In closing

The system is theirs now

TestGorilla has a working system that keeps the team in contact with what the market is actually asking, and clear on how their own content is performing against it. Recruiter questions surface while they are being asked, and the monitoring layer shows which of them TestGorilla is already winning and which are still open. That is what makes it possible to answer a query first, while competitors are still deciding it matters.

"Natalia's AI build took our marketing and our AI understanding to new heights."

Mehak Chowdhary, Head of Marketing, TestGorilla

Want a system like this for your company?

I work with a small number of B2B technology companies at a time. If that sounds like you, let's talk.

Let's Talk