// real-time profitability
One source of truth, then numbers that warn you while there is still time to act. Not a wall of charts.
// the gap
Not a struggling one - a good one, the kind other businesses trust with their books. If the people whose job is the numbers can't see their own profit until tax day, what chance does everyone else have? Here's the standard we hold ourselves to instead, in our own company, every single day:
// every single day
// the standard
I want to know what our revenue and net profit is every day, not six weeks from now.
James Shaw COO
Month-end reporting isn't wrong, it's late. And late means unactionable. By the time the report lands, the month it describes is over. The bad pricing decision has run for six more weeks. The engagement that lost money has quietly lost more. You can't manage what you can only see in the rearview mirror. You don't need a data team for real-time profitability tracking. It's a build, and this page walks through how we do it.
// why you can't see it
Accounting lives in one system. Time tracking lives in another. Project status lives in a task tool, and the connective tissue lives in spreadsheets. Each tool is fine on its own. None of them agree with each other, and none of them was ever asked to.
Every growing company develops what engineers call magic strings - labels, codes, and abbreviations that carry real meaning, but only inside the head of the person who typed them. "ACME-2b-rush" means something precise to your operations lead. It means nothing to a report. The knowledge exists; it just isn't in a form a system can use.
When the numbers disagree - and across disconnected systems they always disagree - someone has to decide which version is real. Usually that decision happens once a year, when everything gets rolled together for the tax return. Which is how a capable, well-run firm ends up finding out in April.
This is normal. It's how nearly every company we've sat down with actually operates, and it's fixable - not with a heroic reorganization, but with a specific piece of work.
We'll draft it and you correct it.
// the work: instrumentation
Pull the data out of every system it lives in - accounting, time tracking, project tools, the spreadsheets - into one place. Nothing is judged yet. The goal is simply that everything exists together for the first time.
This is the unglamorous part, and we'd rather tell you about it plainly than pretend it away. On one current engagement, pattern matching got us to 98.5% of records matched across all the data. The remaining fraction - about 1,300 timesheet entries that didn't match their engagements - is being worked by hand, entry by entry, because those entries are typos and tribal shorthand that only a human can resolve. There is no shortcut through this, and anyone who tells you their data integration is effortless is describing data they haven't looked at yet.
Canonical shape means every record ends up in one agreed structure, so "revenue," "cost," and "engagement" mean exactly one thing everywhere. And here's the part that makes the cleaning permanent: every mismatch we fix becomes a rule that watches for that mistake forever. Fix the data and keep the detector. The next time a typo or an unmatched entry sneaks in, the system flags it the day it happens instead of letting it rot quietly for a year. The data stays clean after that, because every mistake we've already seen is being watched for. At the end of this work you have the thing your business has never had: one source of truth.
// the product on top
The common pitch is "all your numbers in one place." We think that pitch is backwards. A wall of numbers is just the fragmentation problem again, relocated to a single screen. What actually changes behavior is fewer numbers:
"You don't actually need to know all of those numbers. You just need to know the ones that are wrong."
So the system leads with warnings, not reports. It checks everything - transactions, timesheets, projects, recurring charges - every day, and surfaces only what's off - things like a subscription that doubled, or an engagement that has drifted below margin. A timesheet that doesn't add up gets flagged too. Everything that's fine stays silent.
The goal state is an empty warnings page. An empty page means everything is being watched, and today nothing is wrong. The executive view is the same idea rolled up: not a million numbers, but the ten things worth looking at today. When you want to dig into a warning, the dashboards are there underneath, built for investigation rather than daily staring.
// what changes
You see profitability at the level you actually manage. Not one blended company-wide margin, but per person, per engagement, per package - in real time. You learn which clients and services make money, and which team setups quietly lose it, while there's still time to act.
Mistakes get caught the day they happen. A mispriced engagement, a billing gap, a charge that shouldn't recur - flagged the same day, not discovered at month-end after it has compounded for weeks. The cost of an error becomes one day of it, not thirty.
Decisions move to mid-week, not mid-April. When the numbers are current, "should we change this?" becomes a Tuesday conversation instead of an annual postmortem. That shift in cadence is the entire return on the instrumentation work: you steer the month while it's happening instead of reviewing it after it's over.
// your turn
If any of this sounded familiar - the systems that don't talk, the numbers that only reconcile at tax time - the first step is smaller than you might expect.
First pass is a map of where your numbers live - you correct it. No commitment, no pitch deck, just a drawing of your own data flow that's probably wrong in ways only you can see. If the map is useful, we talk about what to build. If it isn't, you keep the map.