Service 2 of 5 · Fixed price
Trusted Numbers. One clean customer list. Numbers leadership can stand behind.
Trusted Numbers gives you one customer list everyone trusts and one page of numbers leadership can stand behind. It is made of two fixed-price pieces: a cleanup of your CRM (the customer relationship management system where your team keeps customers and deals), and a rebuild of your reports from the raw records. You can buy one piece or both, each at its own fixed price, after a free audit.
Fixed quoteOwners and operations leaders whose sales, finance and operations teams each bring a different figure to the same meeting.
You might be saying
- “Sales says one number, finance says another, and I do not know which one to believe.”
- “We have the same customer in the system three times, spelled three ways.”
- “Finance wants CRM numbers that can go in a board deck, and I am not sure ours can.”
What is in it
Two sizes. Pick the one that fits.
Reporting and forecast rebuild
3 weeksone fixed quote, in writing
Pick this when the records are mostly sound but nobody agrees on what the reports mean or how they were worked out.
- Your pipeline, the list of open deals and what each is worth, rebuilt from the raw records
- Your win rate, the share of finished deals that you won, rebuilt from the raw records
- Your forecast, the estimate of what will close and when, rebuilt from the raw records
- Every definition written down in plain words, so "won", "open" and "lost" mean the same thing to everyone
- Every exclusion written down, so you can see what was left out of each number and why
CRM cleanup
3 to 6 weeksone fixed quote, in writing
Pick this when duplicates, blank fields and mismatches with billing are the reason nobody trusts the CRM.
- Dedupe: duplicate records for the same company or person merged into one
- Field standards: one agreed way to fill in each field, so the same thing is always recorded the same way
- Reconciliation against billing: the CRM checked line by line against your billing records, with the differences resolved
- The rules that keep it clean after we leave, set up in the CRM itself
A fixed price, in writing
You get one number in writing before any work starts. It does not move unless the scope does.
30% waits for your test
Before kickoff we agree what done looks like. That part of the fee is due only when your test passes.
Yours to keep
Everything we build is in your accounts, with documentation your own team can run.
How you buy it
Four steps, the same every time.
- 1Free auditWe look at your CRM, your reports and how information moves between your systems, and tell you which of the two pieces you need and in what order. Book it at /contact/.
- 2Fixed quote in writingYou get a written scope and a fixed price for each piece you choose, along with the acceptance test you will use to judge the work. Nothing starts until you have agreed to it.
- 3BuildWe do the work on the tools you already own. Merges and corrections are shown to you before they are applied, and the definitions are agreed with the people who use the numbers.
- 4HandoverYou run your own acceptance test, and part of the fee is held until it passes. Everything we built, including the written definitions, the rules and the log, is yours to keep.
The detail
Everything else, when you want it.
Open a line to read it. You do not need any of it to get a quote.
Clean customer datawho it is for, what is included, the steps
Duplicates, dead fields, deals counted twice, lead sources that went nowhere. We find them, fix them, and write the rule that keeps them from coming back.
This is for you if
- Your pipeline report, your finance report and your marketing report give three different numbers.
- There's a migration, a fundraise, an audit or a new VP coming, and the CRM has to hold up.
- Years of imports, integrations and well-meant manual edits have piled up.
$2.57M
of booked revenue counted twice in one CRM: 61 onboarding records duplicating sales deals. Booked 2023 to 2026 went from $13.78M to $11.21M once we corrected it. That correction was to our own headline number.
What is included
Duplicates
Companies, contacts and deals matched on domain, name, value and timing. Anything we're not sure about goes to a review queue, not an auto-merge.
Reconciliation
CRM against billing, contracts and support, row by row. Every mismatch gets a cause, not just a fix.
Field standards
For the fields you already have: a list to choose from instead of free text, and fields nobody uses retired. Which fields a deal stage asks for is set up under CRM Setup.
Lead source repair
Unmapped sources go to a named bucket instead of vanishing. We backfill history where the evidence supports it.
Bulk changes, checked
Large corrections are exported, fixed and loaded back, then verified with counts and spot checks on both ends. A move to a new CRM is a size of CRM Setup.
Keeping it clean
Validation workflows, import templates, and a monthly hygiene report.
What goes in, what comes out
What comes in, what we build, and what you have at the end. Every box is something you can point at.
What comes in
- Duplicates and dead fieldsYears of imports and integrations
- Billing and contractsThe system finance trusts
- Your definitionsWhat a customer, a deal and a win mean
What we build
- ProfileDupe rates, fill rates, orphans
- RulesField standards and merge logic
- ReconcileEvery revenue figure against billing
- GuardrailsValidation and workflows that keep it clean
What you get
- One record eachPer company, person and deal
- Revenue that matches financeNo more three numbers
- A dupe rate under your limitMeasured, not promised
The steps
In order. You sign off before anything changes in production.
- 1ProfileOne week. Duplicate rates, fill rates by field and by outcome, orphans, broken associations. Measured, not guessed.
- 2RulesMatch rules, survivorship rules and field standards written down and signed off before a single record changes.
- 3RunBatches. Dry run first, backup taken, rollback ready.
- 4HoldAutomation and a hygiene dashboard. You get the rules, not just the result.
The kind of rule we build
One decision, written the way your team will read it: what sets it off, the one question the system asks, and what happens on each answer. In the audit we map the decisions your team makes by hand today and pick the ones worth automating.
Two records look like the same company
The system asksDo the domain and the billing ID both match?
Yes, it is the same company
- Merge automaticallyThe record with billing history survives
- Move every contact and dealNothing is orphaned
- Log the merge on the recordWho, when, and what was kept
No, or not sure
- Send it to a review queueA person decides, with both records side by side
- Record the decision as a ruleThe next one like it is handled without asking
- Report the queue weeklySo it never becomes a backlog
Where it can go next
None of these is on the menu. They are things we can scope with you as a custom fixed quote, on the tools you already own, once the first build is running. Ideas, not promises: each one gets a written scope and a fixed number before any work starts.
Right after the build
- A data quality scorecardDuplicate rate, fill rate and stale records, per team, on a schedule.Custom quote
- Import guardrailsEvery list is checked and deduplicated before it touches the CRM.Custom quote
Once it is running
- Enrichment that refreshes itselfJob changes and company updates applied on a schedule, with a log of what changed.Custom quote
- Parent and child accountsSubsidiaries roll up, so whitespace and revenue by group make sense.Custom quote
When you are ready
- A warehouse copy of the CRMHistory kept, so you can ask what a field said last quarter.Custom quote
- A merger playbookTwo CRMs into one, with a mapped and tested cutover.Custom quote
Reports and forecastwho it is for, what is included, the steps
Pipeline, win rate, attribution, renewals, forecasting. Every number comes with its denominator and a list of what got excluded.
This is for you if
- Two teams gave you two different win rates, and both are confident.
- Finance needs CRM numbers that can go in a board deck.
- Partner and channel attribution runs on whoever argued hardest.
9.1%
the real new-business win rate at one company. The 24.7% in a deck that had been quoted for seven months had counted renewals and expansions as wins against new-business losses.
What is included
Pipeline and conversion
Stage-reach conversion computed properly, young cohorts shown separately, unmapped sources logged instead of absorbed.
Win and loss
A generated exclusion waterfall from every closed deal down to true new business, plus a gate that refuses to cut a rate by any field that's filled in differently for wins and losses.
Attribution
Campaigns, partners, channels. Confirmation logic instead of last-touch guessing. A ranked view, not a debate.
Renewal and churn risk
Renewal dates with a confidence rank on where each one came from. A risk model back-tested before anyone acts on it.
Forecasting and board reporting
Methods tested against prior quarters, an honest range, and a list of what we didn't verify.
Dashboards
The rebuilt figures shown in HubSpot or in the reporting software you already use, documented, with a dated log of every correction.
What goes in, what comes out
What comes in, what we build, and what you have at the end. Every box is something you can point at.
What comes in
- Raw recordsClosed deals, pipeline, renewals
- Finance's numbersThe ones that have to reconcile
- The deck being quotedWhat people believe today
What we build
- DefineEvery metric with its denominator
- RebuildFrom raw records, not from reports
- Back-test and reconcileAgainst outcomes and against finance
- PublishWhere decisions get made, with a change log
What you get
- Win rate by real cohortNew business, expansion, renewal apart
- A forecast you can defendTo the board and the CFO
- Dated correctionsNever silent ones
The steps
In order. You sign off before anything changes in production.
- 1QuestionThe one decision the analysis has to inform. Written down first.
- 2BasisCohort, exclusions and denominators agreed before we produce a number.
- 3AnalysisReproducible from the data. Thin cells flagged, not hidden. Caveats travel with the numbers.
- 4PublishWhere decisions get made, with the method, so you can re-run it next quarter without us.
The kind of rule we build
One decision, written the way your team will read it: what sets it off, the one question the system asks, and what happens on each answer. In the audit we map the decisions your team makes by hand today and pick the ones worth automating.
Someone asks for win rate by segment
The system asksIs that field filled in the same way for wins and losses?
Yes, it is safe to cut
- Run it from the raw recordsNot from a saved report
- Publish with the denominatorHow many deals, and which were left out
- Save the definitionThe next person gets the same number
No, the field is biased
- Refuse the cutA wrong rate is worse than no rate
- Show the fill ratesWins against losses, side by side
- Propose the fixUsually a required field at close
Where it can go next
None of these is on the menu. They are things we can scope with you as a custom fixed quote, on the tools you already own, once the first build is running. Ideas, not promises: each one gets a written scope and a fixed number before any work starts.
Right after the build
- A metric dictionaryEvery number leadership quotes, with its definition, denominator and owner.Custom quote
- The board pack, generatedBuilt from the raw records on a schedule, not assembled by hand.Custom quote
Once it is running
- Cohort and retention viewsNet and gross retention by start quarter, reconciled to billing.Custom quote
- Attribution you can defendSourced and influenced pipeline, with the rules written down.Custom quote
When you are ready
- Scenario planningWhat a change in hiring, pricing or win rate does to the plan.Custom quote
- Anomaly alertsA named owner is told when a number moves more than it should.Custom quote
What you have at the endand what this does not cover
What you can point at when it is done
- A written page of definitions and exclusions behind your pipeline, win rate and forecast, so any figure can be traced back to the records it came from.
- A customer list with the duplicates merged, the fields filled in one agreed way, and the totals checked against billing.
- The rules that keep the list clean after we leave, set up in the CRM itself.
What this does not cover
- We do not promise the rebuilt numbers will look better. In one engagement, a win-rate analysis that reported 24.7% turned out to be 9.1% once it was rebuilt from the raw records. In another, booked revenue of 13.78 million dollars was corrected to 11.21 million after deals counted twice were found. Those are two past results, not a prediction for your business.
- This is not accounting, tax or audit work. We check the CRM against your billing records, but your books stay with your finance team and your accountant.
- Connecting your CRM to another system. That is one system integration, a size of Automation and AI.
Paying monthly and the tools it works withwhat you can add, and what it runs on
Prefer a monthly fee? Either piece can run as a workstream on The Bench, for a flat monthly fee.
Works with
- The CRM's own reports
- Postgres
- Supabase
- Snowflake
- BigQuery
Proof and numbersthree case studies and three calculators
From the work
Names left out, numbers exact. All case studies.
CRM cleanup
$13.78M → $11.21Mcorrected booked revenue
The booked revenue that was counted twice
Read the caseRevOps analytics
24.7% → 9.1%real new-business win rate
The win-rate analysis nobody could reproduce
Read the caseRevOps analytics
763 company-yearsback-tested
The churn signal every CRM encodes, that had no signal
Read the caseRun your own numbers
Every formula is shown. No email needed. All calculators.
What does clean CRM data give back?
Hours your team gets back, and the contacts you stop paying for.
$71,040value recovered per year, at the defaults
Run your numbers CalculatorHow much more accurate could your forecast be?
The number you can stand behind once duplicates are out and the win rate is real.
$885,500surprise avoided at quarter end, at the defaults
Run your numbers CalculatorWhat is an early churn warning worth?
ARR kept when at-risk accounts are flagged 90 days ahead.
$96,000arr kept per year, at the defaults
Run your numbersQuestions about this service
Do I need both pieces?
No. Each piece is sold on its own, at its own fixed price. The free audit tells you which you need, and it is common to start with the piece that is causing the most arguments.
Is Trusted Numbers just a CRM cleanup?
The cleanup is one of its two pieces. The other rebuilds your pipeline, win rate and forecast from the raw records, with every definition and exclusion written down. It is about what a number counts, not how tidy the records are. A win rate that counts renewals as wins stays wrong after a cleanup. Setting up or moving the CRM itself is CRM Setup.
Which piece should come first?
It depends on where the distrust comes from. If the records themselves are wrong, cleanup usually comes before reporting, because a report built on duplicate records repeats their mistakes. If the records are sound and the dispute is over what the numbers mean, start with reporting.
Do we have to buy new software?
No. The work is done on the tools you already own. If a piece does need a software license or AI model spend, it is billed to you at cost, with nothing added.
What happens if the real numbers are worse than the ones we report today?
Then you find out now, with the working shown, instead of at the end of a quarter. Every definition and exclusion is written down, so you can see exactly why a number moved and explain it to your board, your bank or your team.
How long does a CRM cleanup take?
Profiling takes a week. A typical HubSpot portal with 20,000 to 100,000 records takes three to six weeks, including the dry run and verification.
Will we lose data?
No. Merges follow written survivorship rules, run in batches after a dry run, and every batch is backed up. Records we can't match with confidence go to a human queue rather than getting merged automatically.
Can you reconcile our CRM against billing?
Yes, and that's usually where the biggest errors are. Revenue counted twice, fourteen customers blended into one shared company record, that sort of thing.
Do you clean up Salesforce too?
Yes. Same method. Most of our work is in HubSpot. A full move from Salesforce to HubSpot is a size of CRM Setup.
Why do our win rates keep changing?
The cohort changes. Renewals, expansions and onboarding records get mixed into new business, or deals get cut by a field that's filled in for wins and empty for losses. Fix the definition and the number stops moving.
Can you build a churn or renewal risk model?
Yes, and we back-test it on your history first. At one company, engagement recency, worth 30 points in the existing model, predicted nothing across 763 company-years. Churning accounts had been contacted more recently, not less. We rebuilt it around the four signals that held up.
Do you do partner and channel attribution?
Yes. Multi-dimensional attribution across partner ecosystems with confirmation logic and coverage gaps, so you can see which alliances actually produced revenue.
What tools do you use?
SQL and Postgres, Python for back-testing, HubSpot reports, and your BI tool if you have one. Everything reproducible from a query.
Start here
Tell us what is going wrong.
One tap and an email address. We reply within one business day with which size fits and a fixed quote to follow.
Not this one? The other four services: CRM Setup · New Business System · Automation and AI · Sales Training. Or see all five on one page.