For Directors of Marketing Ops, MarTech and Marketing Automation
You already know what's broken. You need the evidence and the hands.
A dropped-lead handoff between the MAP and the CRM, a scoring model sales stopped trusting, a migration that never reconciled, a sync error queue you check on Fridays. We read the instance in your sandbox, put the export behind each finding so you can change it upward without an argument, then fix what the evidence says under gates, with before-states captured and a runbook your team operates from. Senior people who have inherited a portal themselves. From USD $1,500 for the read.
The moment
When we get the call
Four open reqs and a migration. A mandate to "fix attribution" with no headcount. A scoring model you know is wrong and cannot prove without stopping production. An agency that sent someone who had never seen a Marketo instance. The common thread is that you do not need strategy; you need someone who can read a sync log and write the guard set before touching a record.
Three questions we get asked
Answered the way we would answer them on the call.
Nine places, in our experience: sync filter rules nobody reviewed, required-field failures, duplicate-blocked creates, ownership rules that resolve to nobody, lifecycle stage writes that lose the race with the CRM, and four more. Each has a query. We run all nine and hand you the counts.
With the overlap number. Score ≥ threshold versus what sales actually accepted, over the last two quarters, as one table. When the overlap is under 30%, nobody argues about the model; they argue about the fix, which is the argument you wanted.
The runbook written as the work was done, the recordings of every working session, the before-state for every irreversible action, and access revoked in writing. The measure of a good handover is that the second incident is handled without a call to us.
A finding, in the format you would get it
Finding, evidence, exposure, fix. Always in that order.
Anonymised from a real engagement; numbers rounded. Every figure in a real readout carries the export that produced it.
Setup
B2B SaaS, ~1,100 people, HubSpot → Salesforce, two previous admins.
Evidence
Sync error queue: 4,306 records, oldest 14 months. 1,918 blocked on a picklist value removed in Salesforce in 2025. 640 net-new leads from three campaigns never created in Salesforce; the campaign reports counted them as MQLs.
Exposure
Three quarters of paid-search MQL reporting overstated by a population that never reached a rep.
Fix
Picklist reconciled, the 640 replayed under a gate with owner assignment verified per batch, and a weekly queue-age alert routed to ops rather than to a Friday spreadsheet.
The full sample is on the worked sample audit page. The audit is from USD $1,500 — up to 100k records, credited in full against the fix — and the pricing page has every band.
Questions
The four that usually come next.
Every inherited instance is. The method is the same across HubSpot, Marketo, Pardot, Eloqua, Braze and Klaviyo; the connector rules differ, and that is exactly what the audit reads.
Yes, and we prefer it. Nothing runs in production without your authorisation, and every batch in a rollout is gated so you can stop it.
The readout is written to be forwarded: finding, evidence, exposure, fix, cost. Directors send it to the VP or the CFO as-is.
Both. The audit is from USD $1,500 and is credited in full against the rescue sprint or custom build that follows within thirty days.
Prove it, then fix it, without stopping production.
Start with the read-only audit or bring your own evidence to a thirty-minute call and go straight to scoping the sprint.