Skip to main content
Marketing Agencies

Audience Segmentation Data Quality: Why Your Segments Are Leaking

Poorly defined or executed audience segments send the wrong message to the wrong people. Here's how to audit your segmentation logic and plug the leaks.

Key Takeaways
  • Segments leak when date filters are static, exclusions are missing, or AND/OR logic is wrong
  • Spot-check 10 random contacts from any segment before the first send to that segment
  • Always use rolling date filters ('last 90 days') not static date ranges that become stale
  • Document every segment's intended definition — undocumented segments get rebuilt incorrectly
  • Review all segments quarterly — conditions accurate at creation may be wrong months later

What a 'Leaking' Segment Looks Like

Your "High-Value Customers" email segment has 2,400 contacts. You designed it to include customers who spent $500+ in the last 12 months. But your analyst built it 8 months ago and never updated the date filter. It now includes customers who spent $500+ 8–20 months ago — many of whom haven't bought since.

You're sending "loyal customer" messaging to people who may have already churned. The conversion rate on this "high-value" segment is mysteriously low. The segment is leaking.

The Three Types of Segment Leakage

1. Stale filters Date-based filters that aren't updated automatically. "Purchased in the last 90 days" written as a specific date range rather than a rolling window. Segments built on static lists rather than dynamic queries.

2. Missing exclusions Segments that don't exclude who they should. Your "new customer" segment includes people who made their second purchase last week. Your "lapsed" segment includes people who unsubscribed. Your "B2B" segment includes consumers who gave their work email.

3. Logic errors AND vs. OR conditions used incorrectly. "Customers who purchased Product A AND Product B" when you meant "Product A OR Product B." Nested conditions that don't behave as expected due to operator precedence.

Stop sending campaigns to bad data — Sohovi validates your list before you hit send — try Sohovi free.

The Segment Audit Process

For each segment in your ESP or marketing platform:

Step 1: Document the intended definition What is this segment supposed to contain? Who should be in it? Who should definitely not be in it?

Step 2: Review the actual filter logic Open the segment builder and read every condition. Are date filters rolling or static? Are the AND/OR operators doing what you intended?

Step 3: Check the count How many contacts are in the segment? Is that count plausible given what the segment should contain? If your "active email subscribers" segment has more contacts than your total opt-in list, something's wrong.

Step 4: Spot-check records Pull 10 random contacts from the segment. Manually verify they meet the intended criteria. Pull 10 contacts who should be in the segment and verify they are.

Building Better Segments

  • Always use dynamic/rolling date filters, not static date ranges
  • Document every segment's intended definition in a comment or separate log
  • Test new segments with a spot-check before sending to them
  • Review all segments quarterly — conditions that were accurate at creation may not be accurate today

Frequently Asked Questions

How do I set up rolling date filters in Klaviyo/Mailchimp/HubSpot?

All major ESPs support relative date conditions: 'made a purchase in the last X days' rather than 'made a purchase after [specific date].' Use the relative date option whenever time is part of your segment definition.

What's the most common segment logic error?

Using OR when AND is intended in multi-condition segments. 'Customer who purchased Product A AND spent over $100' (both must be true) is very different from 'Customer who purchased Product A OR spent over $100' (either is sufficient). Test with records you know the answer for.

How do I find contacts who should be in a segment but aren't?

Export contacts who meet individual conditions separately and compare. If a contact meets condition 1 and condition 2 individually but doesn't appear in the combined segment, the AND/OR logic is wrong.

Selva Santosh

Data quality, for people who ship

Selva writes practical guides on data quality, profiling, and governance to help teams ship better data.

Start for free

Stop guessing. Start knowing your data quality.

Sohovi profiles your datasets in minutes — surfacing completeness gaps, type mismatches, and duplicate patterns before they reach production.

No credit card required · Free forever plan