Skip to content

Worked sample · illustrative composite · no client data

What the audit actually hands you.

Every consultancy in this category asks you to buy a document you have never seen. This is ours, worked end to end, so you can judge the quality before you spend anything.

Read this first.

The findings below are a composite, built from the failure modes we see most often. The numbers are illustrative and no client's data appears here — not anonymised, not paraphrased, not at all. We do not publish client instances, which is the same reason your instance will never appear on this page either. What is real is the shape: how a finding is written, what evidence sits behind it, and how the work is ordered.

Section 1

The findings, ranked by what they cost.

Not by severity label, not by how easy they are to fix. Every finding carries the query that produced it and the record count it returned, so your own team can reproduce it without us — and disagree with us if we are wrong.

P1

The connector drops records that fail CRM validation, silently

Evidence
Integration error queue, 90 days. 2,148 create failures, none surfaced to a human. Reproduce: sync error log filtered to object=Contact, status=FAILED.
What it costs
Of the 2,148, 431 carried a form fill from a paid campaign. At the account average deal size that is roughly one quarter of target in leads that were bought and then discarded.
The fix
Route the error queue to an owned inbox, then fix the four validation rules causing 86% of failures.
P1

Lifecycle stage moves backwards for 8.4% of records

Evidence
4,902 of 58,300 contacts have a stage-change history containing a regression. Reproduce: field history export on lifecyclestage, flag any transition where rank decreases.
What it costs
Every funnel conversion rate in the executive dashboard is computed across these records. MQL→SQL is overstated by an estimated 11–14 points.
The fix
One workflow is re-stamping stage on re-engagement. Add a guard so stage only ever advances.
P2

Campaign influence is attached to 23% of closed-won opportunities

Evidence
312 of 1,357 closed-won opportunities in the last four quarters have any campaign attached. Reproduce: opportunity report, campaign influence related list count > 0.
What it costs
Three quarters of won revenue is invisible to marketing attribution, which is why the marketing-sourced number and the finance number have never agreed.
The fix
Enable auto-association, then backfill the trailing four quarters from contact-level activity.
P2

31% of keyed CRM accounts carry no domain

Evidence
13,761 of 43,728 accounts with an external key have a blank domain field. Reproduce: account export, key IS NOT NULL AND domain IS NULL.
What it costs
Domain is the match key for company auto-creation. Every keyed account without one is a standing generator of duplicate shells.
The fix
Backfill domain from the primary contact email, gated and in batches. This slows production of the duplicates rather than merging them afterwards.
P3

Two sending domains are unauthenticated

Evidence
DNS check on all four sending domains: two have no DKIM record and no DMARC policy.
What it costs
Both fall below the current bulk-sender requirements. Inbox placement is already degraded on one.
The fix
Publish DKIM and a DMARC record at p=none, then monitor before tightening.
P3

68% of automation assets have not run in twelve months

Evidence
1,204 of 1,770 workflows, programs and lists show zero enrolment in 12 months. Reproduce: asset export with last-enrolment date.
What it costs
No direct revenue cost. The cost is that nobody can reason about the instance, which is what makes every other finding here take three times as long to fix.
The fix
Archive in two passes with a 30-day recall window. Not urgent; do it while waiting on the P1 work.

Section 2

The roadmap, ordered by dependency.

Sequenced so that nothing is rebuilt twice. Merging duplicates before throttling what creates them is the most common wasted quarter in this work.

  1. Weeks 1–2Stop the bleeding

    Error queue routed and owned. The four validation rules fixed. Stage-regression guard deployed. Nothing here needs a project — it needs a decision and an afternoon.

  2. Weeks 3–6Restore the numbers

    Campaign auto-association on, four quarters backfilled, and the marketing number reconciled to finance with the variance written down rather than argued about.

  3. Weeks 6–12Slow the production

    Domain backfill on the keyed accounts, gated. Then the duplicate merge itself, which only makes sense once the thing creating them is throttled.

  4. OngoingHand it over

    Runbooks, the standing monthly checks, and a recorded walkthrough so the next hire runs this without us.

Section 3

What we did not touch.

The audit is read-only. This section exists in every deliverable because the things we deliberately left alone are as much a finding as the things we flagged.

  • Anything at Opportunity or Customer stage. Out of scope by policy, not by oversight. Changing a record a seller is working is a commercial decision, not an ops one.
  • The upstream system. Where the sync is one-way, duplication that originates upstream cannot be fixed from this side. We say so with a number rather than leaving it as a surprise.
  • Anything we could not score. Records with unreadable names or missing keys are held and listed, not guessed at. A guess that looks like a finding is worse than a gap.
  • Your team's judgement calls. Where a choice is defensible and simply differs from ours, we note it and move on rather than dressing a preference as a defect.

That is the artefact. Three weeks, from $1,500.

Credited in full against the work that follows. If the audit finds nothing material, you will be told that in writing and you will not be sold a remediation you do not need.

What the audit includesThe 60-point checklist