This is the checklist I work through at the start of every Platform Health Check. It is not a teaser version — it is the actual list. You can run it yourself in an afternoon with admin access and no tooling.
Most orgs fail four or five. That is normal and it is not a reflection on your team; Account Engagement is easy to configure badly and gives you almost no warning when you have.
How to use this
- Score each item pass, fail or don’t know. “Don’t know” counts as fail — if nobody can answer, nobody is watching it.
- Work top to bottom. The order roughly matches business impact.
- You need admin access to Account Engagement and read access to Salesforce.
- Anything you fail is a finding you can act on without hiring anyone.
Data and sync — the foundation
1. The sync error queue is at zero
Check: Account Engagement → the connector’s error view. Count the records waiting.
Pass: zero, or a handful being actively worked.
Why it matters: every failed sync is a record where marketing and CRM disagree. Reporting, automation and every handover to sales inherit that disagreement silently.
If you fail: group the errors by message, not by record. Thousands of failures almost always collapse into two or three distinct causes.
2. Somebody is alerted when the queue grows
Check: ask who receives a notification when sync errors appear. Name a person.
Pass: a named human, alerted at a low threshold, who looked at it this month.
Why it matters: sync problems get big because nobody is watching. The failure is the monitoring, not the sync.
3. The connector user can see every field it needs to write
Check: the connector user’s profile and field-level security against your mapped fields.
Pass: read and edit on every mapped field.
Why it matters: a field the connector cannot see silently never updates. No error appears. This is the most under-diagnosed cause of “the data is wrong but nothing is broken”.
4. You know which system wins for each mapped field
Check: open the field mappings and read the sync behaviour column.
Pass: the setting is deliberate for the fields that matter, and someone can explain why.
Why it matters: fields set to “use Salesforce value” overwrite marketing data on every sync. Correct for some fields, an unpleasant surprise for others.
Lead scoring — whether sales trusts you
5. Score decay is configured
Check: is there any automation that reduces scores over time?
Pass: yes, with a half-life that matches your sales cycle.
Why it matters: without decay, scores only rise. Your highest-scoring prospects become the people who were most interested eighteen months ago. This is the single most common scoring fault.
6. Grading is built from real closed-won data
Check: does a grading profile exist, and was it derived from your actual customers?
Pass: yes, built from the last 50–100 closed-won deals rather than an aspirational ICP slide.
Why it matters: score without grade sends sales highly-engaged students and competitors. Grade is what makes the number mean something.
7. Disqualification is a suppression, not negative points
Check: how are competitors, students, employees and existing customers excluded?
Pass: hard suppression rules.
Why it matters: negative points can always be out-voted by enough positive activity. A determined competitor browsing your site will eventually cross the threshold anyway.
8. Every MQL ends in a recorded disposition
Check: can you produce a breakdown of why MQLs were rejected last month?
Pass: yes, from a short controlled list — not free text.
Why it matters: without the return path this is a conveyor belt, not a process. Disposition data is the only thing that improves the model.
Deliverability — whether anyone receives you
9. SPF, DKIM and DMARC all pass
Check: send a test to a mailbox you control and inspect the headers, or use any public authentication checker.
Pass: all three pass, and DMARC is at least at p=quarantine.
Why it matters: major mailbox providers enforce authentication for bulk senders. This is the entry ticket, not the whole game.
10. You send from a dedicated subdomain
Check: the From address on your last campaign.
Pass: something like mail.yourcompany.com, not your root corporate domain.
Why it matters: reputation isolation. A bad campaign should not damage the domain your sales team and finance team send from.
11. Inactive contacts are suppressed
Check: is there a rule that stops mailing people who have not engaged in a year?
Pass: yes, with a win-back sequence before suppression.
Why it matters: mailing dormant contacts drags your engagement rate down, which lowers inbox placement for the people who do want your mail. Sending less to the wrong people is how you reach more of the right ones.
Hygiene — whether it stays fixed
12. Every Engagement Studio program has a success exit
Check: pick your three busiest programs. For each, find the rule that removes someone who converts.
Pass: all three have one, and you can name a prospect it fired for.
Why it matters: this is the most damaging common defect. A prospect requests a demo in week two and keeps receiving “have you considered our solution?” for another month — in front of the one person who was actually buying.
Scoring yourself
10–12 pass: you are in better shape than most. Focus on the specific gaps.
7–9 pass: typical. The failures are probably costing you real pipeline, and all of them are weeks of work rather than months.
Below 7: your platform is not the problem — the configuration is. This is worth fixing before you consider any migration, because every one of these failures follows you to a new platform.
What to do with the results
- Fix the monitoring first (items 2 and 9). Cheap, fast, and they stop the next problem becoming invisible.
- Then the sales-facing items (5, 6, 7, 12). These are what sales actually feels.
- Then the slow-burn items (3, 4, 10, 11). Less urgent, more structural.
- Write down what you found, with dates. In six months it is the only way to tell whether things improved.
If you would rather not run it yourself
The Platform Health Check is this list plus the things that need tooling and experience to spot — automation conflicts, deliverability history, reporting integrity, and the root causes behind each failure. You get a scored report and a fix list ranked by impact, yours to keep whoever implements it. From €2,900, one to two weeks, fixed scope.
A note on honesty
I have published the full list rather than a partial one because a checklist you can actually use is worth more to both of us than a teaser. If you run it and fix everything yourself, that is a good outcome — you will have a healthier platform and I will have spent nothing.
If you run it and find you do not have the time, you will at least know exactly what you are buying.