HubSpot Database Cleanup & Categorization — 71k Contacts

Overview

AviaryAI's HubSpot instance contains approximately 71,000 contacts accumulated over time, but the database is not usable for targeted outbound campaigns in its current state. A structured cleanup and categorization project is required before high-volume ABM email sends can begin. This work is a prerequisite for the ^[1] initiative.

The cleanup will be executed by Sebastian Gant using Clay and Apollo.

Problem Statement

The 71k-contact database has several compounding issues:

• ~20,000 irrelevant contacts — Blessing (AviaryAI's CEO) imported a bulk dump of VC firm contacts for a prior initiative; these are not target accounts for current outreach
• Broken/missing names — Some contacts show only "name" as a placeholder, the result of a failed automation attempt by Aaron
• Stale email addresses — The source data was ~6 months old at import; many contacts have emails tied to prior employers or roles that no longer exist
• Missing categorization — Insurance companies and other financial institution types are untagged; only Credit Unions and Banks have partial tagging
• Duplicates — Deduplication has not been run

Without cleanup, sending campaigns risks high bounce rates, domain reputation damage, and wasted spend.

Cleanup Scope

1. Deduplication

Remove duplicate contact records across the database.

2. Name Remediation

Identify and fill contacts with missing or placeholder names. Contacts with no resolvable name should be flagged or removed.

3. Categorization

Tag all contacts into one of the following categories based on their associated company type:

4. Email Hygiene / Validation

Run a hygiene check against all contacts to:
- Validate email addresses are deliverable
- Flag or remove addresses tied to former employers
- Reduce bounce rate before any sends go out

Tools: Clay, Apollo (AviaryAI has an existing Apollo account).

Why This Matters for Campaigns

The cleaned and categorized database directly enables:

• Targeted Tier 2/3 bulk sends via AWS SES — goal is 1,000+ emails/week at scale
• Segmented campaigns by institution type (Credit Unions vs. Banks vs. Insurance)
• A $1k/month learning budget to test messaging and measure open/click rates at statistically significant volume (Aaron noted that at 50–100 emails/day manually, sample sizes were too small to draw conclusions)
• Domain reputation protection — high bounce rates from stale data would damage deliverability for all sends

"We want to get them out ASAP. We don't care if they're perfect... our numbers weren't at a statistically significant [level] to see if the open rate and the message was right." — Aaron Grossman

Tooling

Action Items

• [ ] Sebastian Gant — Deduplicate contacts across the full 71k database
• [ ] Sebastian Gant — Fill missing/broken contact names; remove unresolvable records
• [ ] Sebastian Gant — Tag all contacts: Credit Unions, Community Banks, Insurance Companies, Other, VC/Non-Target
• [ ] Sebastian Gant — Run email hygiene/validation pass to reduce bounce risk
• [ ] Mark Hope — Align with Sebastian on $1k/month learning budget and KPIs before first sends

Related

• ^[2]
• ^[3]
• ^[4]
• ^[5]