About PerceptionX
Candidates have stopped starting their job search on Google. They ask an AI. That AI gives them an answer about your company assembled from Glassdoor threads, review sites, press coverage and whatever's left of your careers page — and most employers have no idea what it says.
We built the measurement layer for that. Every quarter we run thousands of candidate-style prompts across the major AI platforms, capture what comes back, and score each employer on visibility, sentiment and relevance — broken down by market, job function and platform. Clients get a report that tells them exactly where they're invisible, where they're being described badly, and which specific sources are writing their story for them.
What you'd actually do
Produce the client reports. This is the core of the job. Each quarter, for each client, you'd query the database, verify every figure, find the story in the data, and write the narrative. You'd own the analysis end to end — not "pull the numbers for someone else to interpret." Our reports live or die on whether the insight is real and the number behind it is right.
Write our proprietary research. We publish original analysis on how AI is reshaping candidate research — which sources AI platforms actually cite, how that's shifting quarter over quarter, what it means for employers. Nobody else has this dataset. You'd get a byline on it.
Build the tooling. Report production is too manual. We want someone who reaches for Claude and builds the thing rather than doing it by hand a fourth time — query templates, verification scripts, automated checks that catch a wrong figure before it reaches a client. You don't need to be an engineer. You need to be someone who builds.
What we need
- You've used Claude (or similar) as a working tool, not a novelty — to write code, query data, or build something that exists.
- You can write. Clearly, in plain English, for a non-technical reader. Half this job is turning a number into a sentence a Head of Talent can act on.
- Obsessive about accuracy. Our entire commercial position rests on clients trusting our methodology. A figure that can't be traced back to a query is a figure that doesn't ship. If "close enough" is a phrase you use about data, this isn't the role.
Nice to have
- Supabase or Postgres specifically
- You've built and shipped something of your own using AI — a tool, a site, a product, anything real. We'd rather see that than a CV.
- Any exposure to employer brand, talent acquisition, or HR data
- Python, Charts/viz work, or experience turning analysis into decks
What you get
- Ownership from day one. You're the third person. Your work goes directly to clients whose names you'd recognise, with no layer of approval sanding it down.
- Your name on published research. You'd be the author on original analysis in a field where the body of knowledge is about two years old. That's a portfolio most analysts can't build in a decade.
- You'd get very good, very fast. AI-era measurement is a discipline that barely exists yet. Being early in it is worth more than the salary difference.
- Real scope to build. If you want to automate your own job, we'll thank you for it.
Pay: £15,000.00-£18,000.00 per year
Benefits:
Application question(s):
- Are you comfortable working in remote setting?
- Do you have any work you'd like to share with us of something you have built?
Work Location: Remote
