B2B SaaS
Your revenue numberdepends on who you ask.
Billing reports one figure, the CRM another, and the board deck a third. Each was built by a different team at a different time, so all three are defensible and none of them settles the question.
Every system has an answer. None of them agree.
A subscription changes shape constantly. Customers upgrade mid-term, downgrade at renewal, pause for a quarter, add seats in March and give them back in July. Every one of those is a judgment call about when revenue starts, when it stops and what it counts as, and billing, the CRM and the finance model each make that call differently.
So the disagreement is not a reporting problem you clean up once. It is structural, and it compounds. By the time somebody has reconciled it for the board, the number is a month old and the person who assembled it is the only one who can explain how it was built.
One definition of a customer, and of churn
A parent company with six subsidiaries is one relationship, not six logos. A downgrade is contraction, not a cancellation. We write those definitions once and apply them everywhere, so the same question stops returning a different answer depending on who ran it.
A revenue bridge that ties to the ledger
New, expansion, contraction and churn, each reconciled to what finance actually booked. Once the bridge foots to the ledger by construction, a variance becomes something to investigate rather than something to argue about.
Renewal risk you can see coming
Usage, support history and payment signals in one account view, delivered as a list somebody works this week. The save conversation happens while the renewal is still in front of you, not after the invoice fails.
The data function without the first hire
A data leader is a slow, expensive hire, and the first one usually spends a year building the team underneath them. We engage as your fractional Chief Data Officer instead: the ownership of an executive, available now, at a fraction of the role.
Who we work with
Nine departments. One set of numbers.
A data function that only serves finance produces a finance number and a company that argues with it. Every desk below owns part of the answer, and each one needs its own view of the same numbers rather than a different version of them.
Engineering
Product events that mean the same thing in the warehouse as they do in the app, so a question does not start with reading the schema.
Product
Adoption by cohort and by plan, so what a release did to behavior is measured rather than argued about.
Operations
The work in flight, where it is stuck, and what it costs to clear it, on a board that updates without anyone maintaining it.
Customer Experience
Support history sitting next to account value, so the loudest ticket and the most valuable account are told apart.
Finance
Numbers that foot to the ledger, so close is a reconciliation instead of a reconstruction.
People
Headcount, capacity and cost by function, on the same calendar as the plan they are supposed to deliver.
Sales
Pipeline against a quota that ties to what finance booked, and accounts that show their whole corporate family.
Marketing
Spend credited to the channel that earned it, judged on payback rather than on leads.
Executive
One number per question, and the same answer whoever in the building asks it.
What changed
What that looked like elsewhere.
546 accounts, one family tree
Sales could not tell which parent company an account rolled up to, so one global relationship looked like a dozen small ones. Every account shows its family, its health and its owner now, and account planning starts from what the whole relationship is worth.
23 systems, one place to look
A global business ran twenty-three systems that had never been connected, so every question about the company as a whole started with someone assembling exports by hand. Leadership stopped commissioning that work.
1.46 million rows, to the penny
The CFO's operating model had outgrown Excel, and only one person could open it safely. It runs in the warehouse now and foots to the consolidated P&L exactly, so a scenario question gets answered the day it is asked.
Nobody learns on your business
We have stood this up four times over, on BigQuery and on Redshift, for businesses from thirty locations to global enterprise. None of it was a first attempt.
Digital Nervous System
A nervous system for your data.It does not stop at a dashboard.
Most engagements end at a dashboard somebody has to remember to open. The last stage sends the numbers back out to the booking system, the CRM and the payroll file, so the work happens where your team already is.
Your systems
sources
Brought in
extract and load
Kept as it arrived
raw storage
Defined once
modeled, tested
One set of numbers
the warehouse
Views by role
reporting
Back in your tools
activation
Access you control · Auditable · Watched for silent failure
A few of the tools we build with
The whole practice
What a data function actually covers.
Most companies buy one of these and discover they needed all four. Reporting built on a foundation nobody tested becomes a number nobody trusts, and a model built on it inherits the same problem.
Data Engineering
Building the foundation the other three stand on.
Data pipelines and ETL
Production systems push into the warehouse on a schedule, rather than somebody exporting by hand.
Unit testing
Data is checked on the way in, so a broken feed is caught before a dashboard quietly shows it.
Documentation
The custom logic behind a number is written down and readable by the next person, not just its author.
Dimensional modeling
Data is shaped so a question can be answered without rebuilding the dataset to ask it.
Data Analytics
Four questions, worked in order.
Descriptive
What happened.
Diagnostic
Why it happened the way it did.
Predictive
What the future looks like.
Prescriptive
What we should do about it next.
Data Science and AI
Advanced modeling, for a decision or for the product itself.
Decision science
Statistical and probabilistic modeling behind decisions that carry real money.
Data products
Models running inside the product your own customers use.
Data Governance
Running underneath all three: a principled approach to data across its whole life, from the moment it arrives to the moment it is disposed of.
How we engage
Start with the number nobody trusts.
Most companies start us on the one figure that causes the most argument, usually revenue or churn. Settling it produces the definitions and the warehouse underneath them, and everything after it is faster because the pattern already exists rather than being invented for you.
Pricing is published: a $2,500 diagnostic sprint, project work at $350/hr, or a full fractional data function from $10,000 a month. We are glad to put you in touch with a client who bought the same thing.
Show us where the numbers disagree.
Thirty minutes with both partners. We will tell you what it would take.