How we work
We find where your data can earn the most and prove the first system against a control group. Then we stay on to keep it earning, which is where most of the value lands.
From first call to first result
The first call
Thirty minutes on where you want to grow. You tell us about the business and the number you most want to move, and we ask what data you keep and what you have tried so far.
You leave knowing where we would look first.
The opportunity audit
We work through your data and systems with read-only access, inside your own environment. Every opportunity we find gets a revenue figure worked out from your numbers, along with the data and effort it needs.
You get a plan ranked by what each step is worth.
The first build
We build the top item on the plan in your cloud and on your stack, and your engineers review the code as it lands. It ships in weeks and goes live next to a control group that carries on as before.
The uplift is a number you can take to the board.
The retainer is where most of the value lands.
A model earns most in the months after it ships, as long as someone keeps it fit. Customer behaviour shifts, and a model trained on last quarter slowly loses its edge. The retainer keeps every system earning and adds new ones from the plan, so each month builds on the last.
- Run
- Pipelines and models run on schedule, with alerts on data freshness and model drift.
- Retrain
- Each model retrains on fresh data as your customers and your market change.
- Improve
- We tune what already runs and build the next item on the ranked plan.
- Report
- Every month you see what each system earned against its control group.
Working together
Everything we build runs in your environment and belongs to you.
- Your cloud, your stack
- Models and pipelines run in your own cloud account, on the warehouse and tools your team already uses. Your engineers can see and run every part of it.
- You own the code and models
- Code, trained models and documentation live in your repositories from the first commit. They stay yours whether or not we keep working together.
- Access and security
- We sign an NDA before we see any data. Your team grants least-privilege access, read-only wherever that is enough, and can revoke it at any time. Your data stays in your environment.
- Results against a control
- Every system runs next to a control group. Once a month you get a report of the uplift each one earned, with the method laid out so your finance team can check the figures.
Common questions
Do you work with our stack?
In most cases, yes. We build on the tools you already run, from warehouses such as BigQuery, Snowflake, ClickHouse or Postgres to dbt and orchestrators such as Airflow or Dagster. Where a piece is missing, we propose a standard tool and set it up in your cloud account.
Who owns the work?
You do, from the first commit. The code, trained models and documentation live in your repositories and your cloud account, and your team can run and change them without us.
How soon do we see results?
The audit gives you a ranked plan with a revenue figure on each opportunity. The first system ships in weeks and is measured against its control group from the day it goes live. How fast the uplift firms up into a number you can rely on depends on your volumes and how long your customers take to buy.
Can you work alongside our team?
Yes, and it works best that way. We join your chat and work in your repositories, with your engineers reviewing each change, so the knowledge of how every system works stays in-house. If you have no data team yet, the retainer covers running the systems and the documentation lets a future hire take over.
What data do you need to start?
Nothing for the first call. For the audit we need read access to the data behind the opportunity, usually event or transaction history and whatever already feeds your reports. The audit also tells you what is missing and how to start collecting it.
What happens if a model stops earning?
Monitoring catches it early, because every model is checked against live outcomes and its control group. Most of the time a retrain on fresh data brings it back. If the gain has gone for good, the monthly report says so and we move the retainer on to the next item on the plan.
Do you sign an NDA?
Yes. One is in place before the audit starts, or before the first call if you want to go into specifics on it.
Do you only work with games and iGaming?
Those are the industries we know best, along with esports and consumer brands. The same methods work for any business with customer or transaction data and a revenue number it wants to move, from retail and travel to media and marketplaces.
Start with a thirty-minute call.
We'll talk about where you want to grow and where your data can do the most. Bring your questions. No prep is needed.