CRM Data Hygiene: Keeping Contacts Clean in HubSpot, Salesforce and Pipedrive

Bad email data in a marketing list costs you a bounce. Bad email data in a CRM costs you selling time, which is considerably more expensive.

A rep opens a record, sees a contact and an email address, and adds them to a sequence. The address died fourteen months ago when that person changed jobs. Over the next three weeks the sequence sends five messages, all of which bounce. The rep logs follow-ups, sets a reminder, and eventually marks the contact as unresponsive rather than unreachable. The record stays in the pipeline. The forecast includes it.

Nobody in that chain did anything wrong, and the whole sequence was wasted from the first send.

01Why CRM data rots faster than marketing lists

B2B contact data decays faster than consumer data, for one dominant reason: job changes. [email protected] stops working the day firstname leaves, and ordinary job-market turnover means a substantial share of any professional contact list becomes invalid every year.

CRMs make this worse in three ways specific to how they get used.

Records are never deleted. A CRM is an archive by design. A contact from 2021 is still there, still searchable, still addable to a sequence. Marketing lists get pruned because someone pays per contact. CRM records mostly do not.

Multiple people enter data by hand. Reps type addresses off business cards, LinkedIn and email signatures. Typos enter and nothing catches them.

Imports pile up without reconciliation. A conference list here, a purchased list there, a spreadsheet from a departing rep. Each import brings duplicates and its own error rate, and they merge into a single contact object that may hold three variants of one person's address.

02The cleanup order

Work in this sequence. Skipping to verification first means verifying duplicates, which wastes credits.

1. Deduplicate before you verify

Every CRM has duplicates and they multiply through imports. Verifying the same person three times costs three credits and gives you three records that will each bounce independently.

Dedupe on email address first, since that is the reliable key. Then look for the harder cases: same person at two addresses (j.smith@ and john.smith@), or the same company entered twice with different formatting.

HubSpot and Salesforce both have native duplicate management. Pipedrive has merge suggestions. None catch everything, but running the built-in tool first meaningfully reduces the volume you then verify.

2. Verify what remains, and record the result on the contact

Run the deduplicated address list and write the outcome back to a custom field on each record. This is the step that changes behaviour, and it only works if the result is visible where reps work.

Create a field, something like Email Status, with the four values: deliverable, risky, undeliverable, unknown. Populate it from the verification run.

The reason this matters more than a clean export is that a rep looking at a record needs to see the status before they act. A verification report sitting in a folder changes nothing. A red field on the contact record stops a sequence from starting.

3. Suppress rather than delete, mostly

There is a real tension here. Data-minimisation principles say you should not keep personal data you have no use for, and an undeliverable address serves no purpose. Sales teams, meanwhile, have legitimate reasons to keep a company record after a contact leaves.

The resolution that works: clear or flag the address while keeping the company and role context. You retain the commercially useful information, which is that this company had a decision-maker in this role, without holding a personal contact detail that is dead. Then find the current person.

Where a record has no value at all, delete it. Our post on verification and GDPR covers the minimisation obligation and where it actually applies.

4. Block sequences on bad status

The highest-return automation in this whole post. Wire the Email Status field into your sequencing rules so that a contact marked undeliverable cannot be enrolled.

Every CRM with sequencing supports enrollment criteria. Add the condition. This is a fifteen-minute configuration change that stops the failure mode described at the top of this post from ever recurring.

For risky and unknown, allow enrollment but consider flagging it, since those are judgment calls rather than certainties.

5. Fix the entry points

Lead capture forms feeding the CRM should validate before submission: syntax, a typo suggestion with a correction offer, a disposable domain block, and a check that the domain has a mail server at all. Our post on stopping bad addresses at the signup form covers the implementation and why offering a correction converts better than rejecting.

Manual entry is harder to protect. The realistic answer is periodic verification rather than validation at keystroke, since you cannot easily gate a rep typing into a field.

03The pipeline accuracy argument

This is the part that gets executive attention, and it is not about email at all.

If a meaningful share of your contact records hold dead addresses, your pipeline includes opportunities that cannot progress. Not because the deal is bad, but because nobody at your company can reach the person the deal is attached to.

Those opportunities sit in forecast categories. They get discussed in pipeline reviews. Someone assigns them a probability. And the honest probability is close to zero, because the only contact channel on the record does not work.

You cannot fix a forecast by adjusting probabilities if the underlying reachability is unknown. Verifying the contact addresses attached to open opportunities is a small job with a direct effect on forecast quality, and it is usually the version of this argument that gets budget approved.

Start there if you need to justify the work: verify the contacts on open opportunities only. It is a small credit spend and the finding tends to be uncomfortable enough to fund the rest.

04Cadence

Set this on a schedule rather than doing it once.

Quarterly for the active contact base. B2B decay runs high enough that annual is too slow.

Before any outbound campaign, for the segment being mailed.

On import, always. Every list that enters the CRM gets verified before it merges. This is the single rule that prevents the slow accumulation, because imports are the main vector.

Monthly for contacts on open opportunities. Small volume, high value, direct forecast impact.

Measure your own decay rate once rather than assuming the average. Verify the same 500-contact sample ninety days apart and divide the difference by three. The method is here, and B2B rates commonly land well above the frequently quoted 2% a month.

05What this does not fix

Verification tells you an address can receive mail. It does not tell you the person is still in that role, still at that company, or still the right buyer.

Someone can keep a working address while moving from a role where they made purchasing decisions to one where they do not. The address verifies as deliverable and the contact is commercially worthless to you. That is a data-enrichment problem, not a verification problem, and they are frequently sold together in a way that blurs the distinction.

Verification also cannot tell you whether you have a lawful basis to contact someone, which for B2B outbound in regulated jurisdictions is a live question and not one a tool answers.

06The short version

Dedupe, then verify, then write the status onto the record where reps can see it, then block sequences on a bad status. That last step is the one that converts a cleanup into a permanent improvement, because it stops the failure recurring.

Verify on every import without exception, quarterly for the active base, and monthly for contacts attached to open opportunities. And if you need to justify the work internally, lead with the forecast rather than the bounce rate.

Verify a CRM export. Free plan is 500 credits a month, Max is $49 for 75,000 with API access if you want to wire it into the import flow.