The situation

Stratum Group's B2B division sold wholesale and partnership services to enterprise accounts. The sales team was organized across three regions, each with its own reporting cadence. The Sales Ops Director had inherited a Salesforce instance built by three previous admins, each of whom had approached the data model differently. The result was 80,000+ records, 200+ custom fields, and reports that frequently contradicted each other depending on how they were filtered.

Leadership had a weekly review where they looked at pipeline health and account risk. The problem: the data for that review was assembled manually from four different reports, reconciled in a spreadsheet, and usually three to five days old by the time it landed in front of decision-makers. During that window, deals moved, accounts churned, and the actions taken were based on a picture that no longer matched reality.

What we found

We extracted the full Salesforce dataset and ran it through a revenue intelligence analysis. Three findings stood out above the rest.

First, 31% of closed-won revenue was being attributed to the wrong source in the CRM. The original source field was being overwritten when contacts took secondary actions, so marketing was pulling channel performance data that systematically overstated paid advertising and understated organic and referral channels. Budget decisions were being made on incorrect data.

Second, three early-warning churn patterns were hiding in account activity data that nobody was tracking. Accounts that stopped responding to emails within 14 days of renewal had a 73% historical churn rate. Accounts where the primary contact changed in the 60 days before renewal had a 58% churn rate. Accounts with no logged product activity in the preceding 30 days had a 61% churn rate. All three signals were in the data. None of them were surfaced in any report.

Third, the reporting process itself was costing the business more than the insights it was producing. 15+ hours per week of a senior ops resource was being spent assembling reports rather than analyzing them. The manual reconciliation introduced error at every step, which was the root cause of the contradiction problem.

What changed

We rebuilt the attribution model from first-touch data, correcting the overwrite logic and recalculating channel performance for the previous 18 months. The marketing budget reallocation that followed moved investment away from the overstated channels and toward the ones the corrected data showed were actually performing.

The churn signal model was implemented as an automated dashboard that flagged accounts matching any of the three patterns more than 60 days before renewal. CS could see the list every Monday without building a report. The first quarter this ran, the team identified 14 at-risk accounts that would have been invisible under the previous system. Eight of those renewals were saved.

The reporting overhaul replaced the manual weekly process with a connected dashboard pulling live from Salesforce. The Ops Director went from 15 hours per week on reporting to roughly 3, using the difference for analysis rather than assembly. Decision lag dropped from 3-5 days to same-day.