Clean. Model. Alert. Act.
We are an operations business, not a dashboard business. We map how value moves through your company end to end, record the cost, revenue, conversion and time of every activity along it, and then show you where effort returns most.
Most firms are lost in spreadsheets. Not because they lack data.
They have plenty. It is spread across systems that were never meant to talk, so every question turns into an export, and every export turns into a spreadsheet somebody has to maintain. The business ends up busy with the reporting and short of the time to read it.
Someone logs into four systems, exports four spreadsheets, and spends two days making them agree. Next month, again.
By the time the pack is ready the month is over, the decision has been made, and the report is a post-mortem.
The target is in one file, actuals in another, and the gap between them is calculated by whoever asks loudest.
All the effort goes into assembling the numbers. Nobody is left with the hours to sit and read what they say.
Revenue and volume, because they're in the system. Margin, retention and cost to serve, because they're not.
Three people bring three numbers to the same meeting and the first twenty minutes go on whose is right.
We remove the manual reporting entirely.
No more downloads, no more reconciling, no more version. The data arrives on its own, the numbers agree because they come from one place, and the pack is already waiting when you open it. What you get back is the thing you actually wanted: time to look at the business instead of assembling it.
Every system in, deduplicated and reconciled.
Personal data obscured at ingest
Recorded once in BigQuery and never rewritten — every activity with its cost, revenue, conversion and time.
A conscious, real-time view of the business, focused on the next best action.
The one thing that needs you now
Agents work the exception off the back of it, rather than just noticing it.
We map your value stream. Then find where the value leaks.
Reporting tells you what happened. Operations is about where to apply effort next. So we model the business end to end — every activity, in order — and record four things against each one. Layer your business context over those four and you can see, rather than argue about, where the next pound of margin is.
Every step carries its cost — people, media, carriage, cost to serve — so margin is a fact rather than an allocation argument.
Attributed to the activity that produced it, not the last click or the loudest team.
Step-by-step throughput, so you can see which stage leaks and what a point of improvement is worth.
Duration between steps, because most operational value hides in the waiting rather than the doing.
Four units, every activity, end to end.
Cost, revenue, conversion and time are the atoms. Everything a business actually asks — which product to push, which channel to cut, which step to automate, where an hour of effort returns most — is a question about how those four behave across the stream.
It is also why the AI layer works. Give a model a wall of dashboards and it guesses. Give it an ordered history of activities with cost, revenue, conversion and time attached, and it can reason about cause — which is the difference between a chatbot and something that finds you money.
Three things. That's the whole stack.
BigQuery is where your data lives. shadcn is what you look at. Claude is who you ask. That is the whole thing — no vendor bake-off, no sprawl, nothing you have to re-platform later. We picked it because it works, it scales with you, and it lets us start in days rather than months. Connecting your own systems is the only part that varies, because that depends on what each one lets us read.
Google BigQuery
Where your data lives.
One place for everything, run by Google so it scales without you thinking about it. Everything is recorded once and never rewritten, so last month's numbers cannot quietly change. And where your systems only hold today's state, we rebuild the history you never had — so you can see a trend where before there was only a snapshot.
shadcn
What you look at.
Fast, clean dashboards and reports in your own brand — the same toolkit this website is built with. It is code you own, not a licence you rent, so nobody can put the price up or switch it off.
Claude
Who you ask.
Ask a question in plain English and get an answer from your own data — no analyst, no ticket, no wait. It is the COO and the FD using this directly, not just the data team. Runs in your own Claude account, so your data stays inside your boundary and your controls.
- Heavy data engineering
- Warehouse design from scratch
- A BI build project
- A five-person hiring round
In their place: BI delivery, tracking and decisioning, end to end, for less than half a senior data hire. The whole environment is Terraformed, so it hands over cleanly — or we keep running it and wire in new sources as they arrive. Your call, not a contract term.
The only thing we're waiting on is access to your data.
No discovery phase to sit through, no procurement marathon. Give us read access with personal data excluded, and we start — your first report inside a week.
Retries are expected and safe: a pipeline can re-run without double-counting, so it produces the same numbers every time.
Within a week to something real. Twelve months to not needing us.
No discovery phase that bills for six weeks and produces a document. Here is what actually happens, in order.
- Week one
Something useful, already live
We connect the first source, land the events in your BigQuery project and put a real number in front of you. Not a plan for a number — the number. It is deliberately the first thing we do, because it tells you whether you want to work with us.
- First month
The layer takes shape
Sources joined, definitions agreed and written down as code, history rebuilt as far back as the data allows. This is where the arguments about what a customer is get settled once, in the open, with you in the room.
- Months two and three
The Summary Page and the reporting behind it
The scorecard, the objectives and the alerting go live on top of the layer, and the manual reporting starts coming out of people's weeks. Regulatory and board reporting gets built here too, because it is the same layer.
- Months four to twelve
Self-service, then agents
We open the layer up to your team through Claude, with personal data walled off, so people can ask their own questions. Then we put agents onto the repetitive operational work — the chasing, the flagging, the routing — and keep your people on the decisions.
- After twelve months
You take the keys
A structured handover: everything Terraformed, the models documented, your team already running it. You can walk away, keep us on for £2k a month on a rolling contract so nothing falls over, or agree new datasets and training as projects. We expect you to be self-sufficient well before this point.
The transparency and the speed the Acta Data team work at is refreshing. Being able to query our own data and build our own reports is the biggest step forward we've had in five years — and it has only been three months.
Bring one decision your data should be helping you make.
30 minutes, no deck. We'll tell you on the call whether we're the right partner, and exactly what we'd ship in the first 30 days.