Services · Other industries

Your analytics team, without hiring one.

We run your analytics month to month, build data systems you own, build AI systems that have to be accurate, and fix reporting that has stopped being trusted.

In home services? See the Profit Leak Audit
$850K
a year, the client’s estimate

Saved by automating one report that took 60 hours to build by hand.

91%
about, from 61.5%

Accuracy on the key fields of an AI statement parser, measured against a hand checked set.

$173K
annual revenue lift, modeled

From a tiered pricing model built on a survey of the client’s own customers.

2022
two clients, since

The year two of our clients made us their analytics team. Both still are.

01 · Managed analytics

Run it for you

A managed analytics team that keeps your reporting running, answers new questions as they come up, and sits in the meetings where the numbers get used.

Tier 1

Maintain

We keep what you have working. Refreshes run, broken feeds get fixed, and small changes get made without a new project.

Tier 2

Manage

We own the reporting roadmap. New reports and analysis each month, planned with you and delivered on a schedule.

Tier 3

Embedded

We work as your analytics team. In your meetings, on your questions, with the context of someone who has been there for years.

Month to monthNo lock inChange tiers any month

If you don’t need us every month, we’ll tell you.

Proof

Two clients have kept us as their analytics team since 2022.

Book a call
02 · Builds

Build it

Builds you own. Everything goes into accounts in your name, and the code, dashboards, and documentation are yours to keep.

How pricing works: agreed before we start, one of three ways. A fixed price for the whole build, fixed price milestones, or a capped amount of work. Never an hourly meter.

01

Data warehouses and pipelines

One place where your CRM, ad accounts, accounting, and operations data land every day, cleaned and joined so the same question gets the same answer.

02

Dashboards and command centers

Views built around the decisions each role makes, from the owner to the call center, reconciled against the reports you already trust.

03

Reporting automation

The report someone builds by hand every week, rebuilt to refresh on its own and land where people already look.

Proof

About $850,000 a year saved, by the client’s estimate, by automating one report that took 60 hours to build by hand.

Book a call
03 · AI

AI systems and advisory

We build AI systems for work where a wrong answer costs something, and we give plain advice on where AI will help your business and where it is not ready yet.

Build

Systems that have to be accurate

Document processing that reads statements, forms, and invoices into clean data. Tools that answer questions from your own records, with the source shown next to every answer.

Advise

Where it helps, where it does not

A review of the tasks you are considering, which ones AI handles well today, which ones it gets wrong often enough to matter, and what a small test would look like first.

How we test before anyone relies on it:

  1. Build a hand checked set from your own documents or questions.
  2. Measure accuracy on that set before launch, and report the number.
  3. Keep a person reviewing output wherever an error is costly.
  4. Measure again whenever the inputs change.
Proof

An AI statement parser went from 61.5% to about 91% accuracy on its key fields, measured against a hand checked set.

Book a call
04 · Audits and cleanup

Fix what you have

Most companies do not need a new system. They need the one they have to be right. We audit your dashboards and reports, find where they go wrong, and clean them up.

  • Reports that disagreeMarketing, sales, and the books each show a different number for the same month. We trace every difference to a timing, definition, or mapping rule and write it down.
  • Dashboards that breakA refresh fails quietly and the screen shows old numbers. We find the cause, fix it, and add a check that flags it next time.
  • Reports that go unusedNobody opens them because nobody trusts them. We keep what answers a real question and retire the rest.
Our standard

No fix is finished until the numbers tie out line by line against a report you already trust, for a full reporting period.

Book a call
How we work

Three principles

01

Right sized

We build the smallest thing that answers the question. Sometimes that means telling you to keep the spreadsheet you have.

02

Figures you can defend

Every number ties back to a source you can check, and anything estimated is labeled as an estimate.

03

Analytics points, you decide

We show you where to look and what it is likely worth. You still make the call.

Questions

Frequently asked

01What is a good first project?+

Usually one question that costs money to leave unanswered: a report that takes a day to build by hand, two numbers that never agree, or a dashboard nobody opens. We scope it on a first call, agree the price before starting, and deliver it as one piece of work you can judge on its own.

02How does pricing work?+

Builds and fixes are priced one of three ways, agreed before any work starts: a fixed price for the whole project, fixed price milestones, or a capped amount of work. There is no hourly meter running. Managed analytics is a monthly fee set by tier, month to month. We quote after a first call, once we understand the scope.

03What systems do you work with?+

The ones you already run. That includes CRMs such as HubSpot, accounting in QuickBooks, Google and Meta ad accounts, databases and cloud data warehouses on AWS and Azure, and dashboards in Tableau, built in SQL, Python and R. If we have not worked with your system before, we will tell you on the first call.

04Who owns the work?+

You do. Accounts are set up in your name and billed to you directly. The code, the dashboards, and the documentation are yours to keep whether or not we keep working together.

05Will you sign an NDA or a BAA?+

Yes. We sign NDAs as a matter of course, and a Business Associate Agreement for any work that touches protected health information. Where we can, we ask for read only access, so we can see your numbers and change nothing.

06How is AI work tested before anyone relies on it?+

Against a hand checked set of your own documents or questions, before launch. We report accuracy on that set, keep a person reviewing output wherever an error is costly, and measure again whenever the inputs change. If it is not accurate enough to rely on, we tell you it is not ready.

Start here

Start with one question

Tell us the report that takes too long, the numbers that do not agree, or the thing you want AI to do. We will tell you whether we can help, and whether you need us at all.