Salary
From £50,000 to £60,000 per annum.
Location
Enjoy the flexibility of remote work while collaborating with teams across the globe (must be UK based).
Build the AI and intelligence layer SGNL runs on — then use it to explain why revenue and margin move, and to put better decisions back into the business.
The business
SGNL is a high-volume financial-services lead generation and distribution business, principally in the US. Our Pingtree platform routes consumer enquiries from affiliates to lenders and buyers in real time, producing continuous data on partners, campaigns, pricing, routing, acceptance, revenue and margin. The job is not more reporting. It is to turn that stream into a durable operating system the business uses every day.
Why the role exists
Too much important work still starts as a one-off question: why an affiliate slipped, why a lender stopped buying, where margin leaked last week. Those answers vanish if they live in a spreadsheet or a Slack thread. This role owns both sides of the same loop:
• Build infrastructure. Trusted definitions, reusable models, exception detection, investigation tools and AI-assisted insight that last beyond the original question.
• Find the money. Use that layer to diagnose why revenue or margin is being lost, recovered or left on the table — and measure whether the resulting decision worked.
Commercial decisions stay with the business owner. You make them better informed, faster, and measurable afterwards.
What you will do
Diagnose commercial performance
• Work with the Operations Manager on the highest-priority commercial questions and turn unclear problems into structured investigations.
• Analyse affiliate, campaign, lender, pricing, routing, revenue and margin performance — emphasising why a metric moved, not that it moved.
• Typical questions: why a source or lender changed behaviour; where revenue or margin is leaking; which restricted sources should be reconsidered; whether last month’s optimisation actually worked.
• Present findings so non-technical colleagues can act: what the data shows, what it may mean, what is still uncertain, what should happen next.
Build the intelligence layer
• Establish trusted, documented definitions for the metrics the business actually runs on.
• Improve the data models and pipelines under analysis so tools sit on clean, reusable data rather than one-off extracts.
• Build Power BI (and related) surfaces around exceptions and decisions, plus automated alerts on material moves in volume, acceptance, price, route, revenue and margin.
• Keep a living record of optimisation decisions — restrictions, exclusions, price and routing changes — and close the loop so each change is measured and reviewed.
Make AI part of how the business operates
• Ship a small number of high-value capabilities into the operating rhythm: natural-language querying, automated investigation summaries, anomaly detection, assisted root-cause analysis, and forecasting only where it is reliable enough to act on.
• Use LLMs to accelerate investigation — then validate every material conclusion against the underlying data. Prototype fast, kill what does not create value, harden what does into shared infrastructure.
• Keep the AI layer maintainable: versioned workflows, clear data contracts, documented assumptions, and systems the next person can run.
• Convert useful one-off analysis into a dashboard, alert, review cadence or tool. Support adoption through short documentation and iteration on real use.
What this role is not
• Not a dashboard factory. Volume of reports is not a success metric.
• Not a research or platform-engineering role. You build working intelligence systems; Technology owns the core platform.
• Not the commercial owner. You diagnose, recommend and measure. Operations and partner teams decide.
What we need
Around 3–5 years in BI, analytics or commercial analysis is a useful signal. Capability matters more than tenure.
Essential
• Strong numerical investigation and a habit of asking why a number moved.
• Solid SQL and Power BI, used on live commercial problems rather than only to publish reports.
• Evidence of turning messy business questions into analysis the business can reuse.
• Practical use of LLMs, Copilot or similar — and the judgement to know when a simpler query is better.
• Comfort with high-volume transactional data, where small rate changes compound into material revenue or margin.
• Clear communication with non-technical colleagues, and independence suited to remote work.
Useful, not required
• Python or scripting; hands-on LLM APIs; Microsoft Fabric.
• Applied forecasting or anomaly detection — enough to use, not enough to research.
• Background in lead generation, affiliate marketing, financial services, ad tech or another high-volume transactional environment.
How success is measured
In the first six months:
• A working grasp of how SGNL makes money and which metrics actually drive performance.
• Trusted definitions and exception visibility across volume, acceptance, price, route, revenue and margin.
• A clear drop in recurring manual analysis because the same questions now have a system behind them.
• At least one AI-enabled capability colleagues use without you in the room.
• A living record of optimisation decisions, with restrictions and changes being reviewed rather than left to decay.
• Documented cases where investigation supported a better revenue, margin or operational outcome — and the result was measured afterwards.
Over time: issues seen earlier, investigated faster, decided with better evidence, and kept in the way the business operates. The infrastructure should still be valuable after the original question is forgotten.
Working relationships
You report to the Production Director, who sets priorities and provides commercial context. You own the investigation and the supporting intelligence layer. You work closely with Affiliate Management, Lender Management, Technology, Data and senior management.
