Every CRM has a graveyard.
Contacts who filled out a form two years ago. Leads who downloaded something once and disappeared. Former prospects who went cold before anyone followed up properly. They're all sitting in your HubSpot portal, aging quietly, doing nothing except inflating your contact count and skewing your engagement rates.
The question isn't whether they exist. It's which ones are actually worth one more attempt, and which ones you should archive and move on from.
This hack answers that in five minutes.
Go to CRM > Contacts in HubSpot and click into the Advanced filters panel.
You need two conditions working together:
Filter 1: Last marketing email open date - is more than - 90 days ago
Filter 2: Last marketing email send date - is known
That second condition is important. Without it, the filter will also pull in contacts who have never been sent a marketing email at all - people who have nothing to re-engage with because they were never engaged in the first place. Adding "send date is known" scopes the list to contacts who received emails but stopped opening them. That's the actual dead contact problem you're trying to solve.
A quick note on properties: Last marketing email open date tracks opens from HubSpot marketing emails specifically. It's different from Last engagement date, which captures a broader set of activities including meetings, form submissions, and one-to-one sales emails. For this use case, the marketing email open date is the right filter. You want to find contacts who have gone quiet on your outbound communications, not contacts whose sales rep hasn't logged a call recently.
Once both filters are set, check the contact count. If you have thousands, narrow further by adding a lifecycle stage filter - focus on MQLs, SQLs, or Opportunities first. Those carry the most re-engagement value and the clearest context for scoring.
This is the part most hacks skip. We're not going to.
When you export contact data from HubSpot and upload it to an external AI tool like Claude or ChatGPT, you're sharing personal data with a third party. That's a data processing activity, and depending on where your contacts are located, it has legal implications you need to be across before you click export.
Check consent and lawful basis first.
Under GDPR - which applies to any contact based in the EU or UK regardless of where your business operates - you need a lawful basis to process personal data. That basis might be consent, legitimate interest, or contract performance, depending on how the contact came into your CRM. If a contact withdrew consent or unsubscribed, their data shouldn't be shared externally for this kind of analysis.
Australian businesses are subject to the Privacy Act 1988 and the Australian Privacy Principles, which also require a legitimate purpose for sharing personal data externally.
HubSpot stores consent and legal basis information on contact records - specifically the GDPR fields including Legal basis for processing, Consent to process, and Unsubscribe status. Before exporting, filter out contacts who have withdrawn consent or opted out of data collection. If your portal isn't capturing this data at all, that's a separate problem worth fixing before running any export-based exercise.
Minimise what you export.
The AI doesn't need names and email addresses to score re-engagement potential. It needs job title, company, lifecycle stage, deal history, and last email activity. Consider removing first and last name from your export if you don't need them for the output - the scoring works on professional and behavioural attributes, not personal identifiers. Less personal data shared externally means less exposure.
Check your AI tool's data handling settings.
Both Claude and ChatGPT have enterprise or team versions with stronger privacy protections - specifically, data submitted in those tiers isn't used to train the AI models. If you're using a personal or free account, check the privacy settings and opt out of model training where available. For anything involving client or customer data, an enterprise account is the right call.
This is general guidance, not legal advice. Privacy law is jurisdiction-specific and the right approach for your business depends on your contact base, your data collection practices, and how your privacy policy is written. If you're unsure, run this past your legal or privacy team before proceeding.
Do it right once and you can run this exercise every quarter without the concern.
Before exporting, customise your column view to include the properties the AI needs to score these contacts meaningfully.
At minimum, include:
First and last name are optional here. As covered in the privacy section above, the AI scores on professional and behavioural attributes, not personal identifiers. Include names only if you need them to identify records in the output for follow-up action. If you do include them, make sure the contacts in your export have a valid lawful basis for external data processing.
In the top right of the contacts view, click Export, select CSV, and choose to export the columns in your current view. HubSpot will email you a download link, it usually arrives within a couple of minutes.
Upload the CSV and use this prompt:
"Here's a list of contacts who haven't engaged in 90+ days. For each, look at job title, company, lifecycle stage, and deal history. Score them 1 to 5 on re-engagement potential where 5 = worth a personal email, 1 = archive. Give me the top 20 worth one more shot, with a reason for each."
What comes back is a ranked table: contact name, score, and a one-line reason for the score. A 5 might read "Senior decision-maker at a mid-market company with a previously associated deal - strong candidate for a personal check-in." A 1 might read "Generic job title, no deal history, original source was a content download with no follow-up activity."
The scoring is only as sharp as the data you export. Contacts with rich CRM data - a deal, a job title, a known company - get scored with confidence. Contacts with minimal data get low scores by default, which is itself useful information. It tells you the record isn't worth the effort of a personal outreach.
The top 20 list is your re-engagement shortlist. These are the contacts worth a personalised, one-to-one email, not a bulk campaign, not a sequence, a real email from a real person that references something specific to them.
The contacts that score 1 or 2 are your answer to the other question this hack is asking: how much dead weight is your CRM carrying? If you have 3,000 contacts on your export and the AI can only find 20 worth a personal touch, the rest are candidates for suppression, archiving, or a sunset campaign before you stop emailing them entirely.
HubSpot does have a built-in unengaged contact management feature -called graymail suppression - that can automatically suppress contacts who consistently don't open marketing emails. If your unengaged list is large, that's worth turning on alongside this exercise. But the AI scoring gives you something graymail suppression can't: a prioritised shortlist of who to try to win back before you let them go
Dead contacts accumulate faster than most teams realise. Someone who was active six months ago can slide into the unengaged category without anyone noticing until your email deliverability starts taking a hit. Because high volumes of unengaged contacts on your list affect your sender reputation over time.
Running this exercise quarterly keeps the list clean, keeps re-engagement outreach focused on the contacts most likely to respond, and gives you a clear picture of how your database health is trending.
If this surfaces a bigger problem - a contacts database that has grown without a clear suppression or sunset strategy, engagement rates that don't reflect the size of your list, or a CRM that has more noise than signal - that's the kind of thing we help with.
Neighbourhood is a Diamond HubSpot Partner. Database health is one of the most common things we fix in portals that have been running for a few years. Get in touch and we'll take a look.
Talk to us about your HubSpot setup.
Ops Intel is our newsletter for HubSpot admins and RevOps teams -practical, no-filler hacks delivered straight to your inbox every week. Subscribe to Ops Intel.
We're also breaking down hacks like this one on Facebook and YouTube if you'd rather watch than read. Come find us there.
Happy HubSpotting!