Recruitment agency lead generation case study: Hoxton Circle
Case study ·
The problem
Hoxton Circle is a recruitment firm that places finance leaders, most of them chief financial officers. Like every search firm, it wins work in two places at once: from the companies that need a finance leader, and from the people who might fill the role.
On the company side, the signal is public. A company that advertises for a controller or a finance director is telling the market it has a gap, and a firm that reaches it while the advert is fresh has a reason to call. The work is in the reading: the adverts are spread across job boards and company career pages, many are for roles, industries or places the firm does not work in, and the advert never names the person who owns the role.
On the candidate side, the firm's vacancies sit on its own website with the company left out, because the companies it works for are confidential. A job seeker with a question has nobody to ask until they apply. And the firm's newsletter of featured jobs reaches only the people already on its list, while turning each issue into something worth posting on LinkedIn is one more job nobody has time for.
What we built
Three agents, each built around one part of that work, and all three shown to the firm on one private page.
New business. An agent that reads the job ads, keeps the companies that fit the firm's own profile, researches each one and finds the people behind the role. It works in the order below, and it is the agent behind recruitment agency lead generation.
Job seekers. A chat that answers questions about the firm's open roles from its vacancy pages alone, never names the company behind a role, and sends people to the apply form: an AI recruitment chatbot.
LinkedIn. An agent that turns each issue of the firm's featured-jobs newsletter into a LinkedIn carousel in the firm's own style, for the firm to edit, approve and download: an AI LinkedIn carousel generator.
Step 1
Read the job ads
Adverts for the finance roles the firm places, from the job boards and company career sites, posted in the last seven days.
Step 2
Keep the ones that fit
Each company is checked against the firm's own profile: its size, where it is and its industry.
Step 3
Research the company
Revenue, headcount, industry and stage, each with where it was read and the date it was checked.
Step 4
Find the people behind the role
The finance and talent leaders most likely to own the vacancy, with their title and LinkedIn.
Step 5
Say how sure it is
An address is marked Verified when it was checked, and Predicted when it was built from the company's format.
Step 6
Draft the first email
A draft for a person to read, change and send. Nothing is sent from the page.
One run of the new-business agent, from the job ads it reads to the draft a person sends.
Safeguards
It never guesses a number. A figure it could not find stays blank on the page, and every figure it did find says where it came from and the date it was checked.
Companies the firm rules out stay out. A company the firm removes from the list is skipped by every later search, so the same wrong lead never comes back.
An address says how sure the agent is of it. Verified when it was checked, Predicted when it was built from the company's format, so nobody mistakes a pattern for a confirmed inbox.
Nothing is sent on its own. The agent writes the first draft; a person at the firm decides whether it goes, and sends it.
The chat keeps the firm's confidences. It never names, hints at or confirms a company behind a role, never guesses a salary, and never judges whether someone fits. Those calls are left to the firm's consultants.
Every carousel waits for the firm. The firm can change any slide, leave out any role and rewrite the caption before it approves, and the post goes out by a person's hand.
The firm stays the one that decides: who to call, what to say, and what goes out under its name.
The result
All three agents are built and in front of the firm on one private page. On its first run, up to 1 November 2026, the new-business agent found 34 companies with open finance roles, each one researched against the firm's profile and with the people behind the role found. The chat answers from the firm's own vacancy pages, and the first carousel was made from an issue of its newsletter. What the agents save the firm in time is the next thing this page will report.
A job ad, as foundbefore
Company
Named in the ad
Size
Not stated
Fits the profile?
Unknown
Who owns the role
Not stated
Email
None
The same ad, researchedafter
Company
Named in the ad
Size
Revenue and headcount, each sourced
Fits the profile?
Checked against size, place and industry
Who owns the role
Names, titles and LinkedIn
Email
Verified or Predicted, and a draft
What one row holds once the agent has finished with it. The ad on its own names a company and a role, and nothing else.
Frequently asked questions
How can a recruitment agency use AI for lead generation?
By letting an agent do the reading. Companies advertise the roles they need filled; an agent can read those adverts, keep the companies that fit the agency's own profile, research each one and find the people who own the role, so the agency's time goes on the calls rather than the search.
Where do the leads come from?
From job adverts posted in the last seven days, on job boards and company career sites. A fresh advert for a finance role is a public signal that the company has a gap, which is why it is worth a call.
Does the agent send emails on its own?
No. It writes a first draft for each company, and a person at the firm reads it, changes it and decides whether to send it. Nothing is sent from the page.
Can a recruitment chatbot answer job seekers without naming the company behind a role?
Yes. This one answers only from the firm's own vacancy pages, describes a company only the way the listing does, and never names, hints at or confirms who it is, even when asked directly.