CRM Baseline & Cleanup Report

Built over several years from five untracked import sources, and never measured. Nothing was known about whether any address worked.

1. Measured the database in full: volume, bounce rates, ownership, duplicates, orphaned records and import sources.
2. Ran syntax, DNS, MX and SMTP validation against all 7,896 addresses in a single pass.
3. Queried the public mail records of every domain validation could not resolve. This was added beyond the agreed scope and found 29 further contacts on domains that cannot receive mail at all.
4. Ran eleven public DNS checks on the sending domain to establish whether receiving servers trust its mail.
5. Built eight rule-based suppression segments that keep working on contacts added later.
6. Built working segments for duplicates, orphaned records and departed owners, record by record.
7. Switched 156 contacts to non-marketing.

Nothing was deleted and no history was touched.

Categories overlap, so the column does not total. All eight are rule-based and keep working on contacts added later.

The database now carries a deliverability status on every contact, eight segments that keep themselves current, and a live count against the 5,000 limit. Marketing contacts move to 2,660 at the next update date.

Every large send opens between 7.12% and 8.94%. One curated send of 52 people returned 33.87% from the same domain and the same sender. The content works when the list does. Every hard bounce carries a reason, and 512 of 546, 93.8%, are unknown user: the mailbox does not exist.

5,367 addresses sit behind corporate mail gateways that refuse verification as policy. One top-five device manufacturer returns no response for 204 of its 221 contacts and no invalid results at all. A large biotech's server, by contrast, answers the same check and returns 24 invalid. A domain producing no invalid results is a gateway declining to answer, not decayed data.

Four more of the sector's largest names behave the same way. Thirty-one domains do, covering 902 contacts. No further spend on validation tools changes this.

336 of the 487 contacts carrying a deal sit inside them. Suppressing on an unanswered check would remove the people the client has already sold to. See the split in annex A.

Eight of eleven domain checks pass. SPF is correct and both DKIM keys are published.

The policy reads "none", so a message failing authentication is reported to the client and delivered to the recipient anyway. Anyone can send mail claiming to be from the domain.

No contact in the database has a recorded website visit. The site runs six separate analytics tools and none of them is connected to the CRM. Attribution is impossible until the tracking code is installed.

2,272 of the 2,840 marketing contacts sit with a single owner. Any decision about the plan is a decision about that segment. See ownership in annex C.

78 contacts have been added since validation ran, 26 of them with no email address, against a database average of 14.8%. A cleanup delivered today decays from tomorrow.

Group A is a straight merge. Group B needs the client to decide whether it is one person who moved or two people who share a name. Nothing has been merged or changed.

The CRM's own tool flags 71 candidate pairs, and a pair is two records. After review those became 126 records to act on. The record-by-record listing follows.

The record to keep is marked. Open it, use Actions and Merge, and select the other record. The CRM preserves activity and associations from both.

The CRM flagged these because the names match. Each one is either a single person who changed employer, or two different people who happen to share a name.

Nothing in the data separates the two cases, so none of these carry a recommendation. The client confirms which case applies, and the action follows from that. Do not merge either way: merging a job change destroys a real second relationship, and merging two people corrupts both records.

Keep the record marked, it carries the domain or the correct spelling. Open it, use Actions and Merge, and select the others.

Checking these individually found a single cause behind four of them, and it is not duplication.

Four records were created on the same day in a single import and each one inherited the website, country, city, industry, revenue, and employee count of a completely different company. A pharmaceutical business carries a clothing retailer's domain and is filed under apparel.

A food producer carries a chemical group's data. Merging any of these would destroy a real company record. The correct fix is to clear the inherited domain and firmographic data on the newer record and enter the company's own.

In all four cases the older record holds every attached contact and the newer record holds none, which is consistent with the newer one being an empty shell created by the import.