MCP

List hygiene and data quality

Last verified: 2026-08-05

TL;DR

  • B2B contact data decays by roughly 22-28% a year as people change jobs and companies restructure (2026) [1]. A list is a perishable asset, not a permanent one.
  • Deduplicate by person and by company. Two colleagues receiving the same campaign in the same week is a complaint waiting to happen.
  • Suppression is part of hygiene and the part with legal weight: it must survive every rebuild.

The decay rate is the headline

At 22-28% a year, roughly a quarter of a list is wrong within twelve months, and about half within two years [1]. Nothing about the list changes visibly in the meantime — the rows still look like rows.

This has a direct operational consequence: re-verify before every campaign rather than relying on a verification result from when the list was assembled (Email finding and verification). It also means a list bought or built more than a year ago should be treated as a research starting point rather than as a sending list.

Decay is not uniform. Fast-growing companies and junior roles churn fastest; owner-operators of established small businesses barely move. If you segment by company size you have already segmented, roughly, by decay rate.

Deduplication, both ways

By person. The same human appears twice with two addresses, or under a maiden name, or with the same address in different cases. Normalize, then match.

By company. The subtler and more damaging case: three people at a twelve-person company all receiving the same sequence in the same week. They talk to each other, and what they conclude is that they have been spammed. Cap contacts per company by company size — one for the smallest, a small number for larger ones, and stagger them rather than sending together.

A related rule: if one person at a company replies, pause the others. Whatever they said, someone there is now aware of you, and a parallel sequence to their colleague reads as not paying attention.

Normalization

Merge fields are only as good as the data behind them, and unnormalized data is where personalized emails embarrass themselves — "Hi JOHN", "I saw that Acme Corporation, Inc." The fixes are mechanical: case-normalize names, strip legal suffixes from company names, standardize country and industry values to a fixed vocabulary, and trim whitespace everywhere.

Then check the extremes rather than the average, which is the rendering half of Pre-launch QA.

Suppression survives everything

The hygiene rule with legal force: opt-outs and hard bounces must persist across list rebuilds, tool migrations and new data purchases. A suppression list that lives inside one campaign is not suppression — the person who opted out in March must be excluded from the list you buy in September. See Negative replies and unsubscribes and Staying legal in a nutshell.

A maintenance rhythm

  • Before every send: re-verify, apply suppression, dedupe against recent sends.
  • Quarterly: re-check the segment definitions against what actually replied, and drop rows that have been through a full sequence with no engagement.
  • Annually: assume a quarter of the list is wrong and rebuild the parts that matter.

Data you no longer have a reason to hold should be deleted rather than archived — a smaller, current list outperforms a large stale one on every metric that matters, and holding personal data you are not using has its own compliance cost.

References

  1. Instantly — 2026 email verification benchmark: accuracy scores for 8 top tools (vendor benchmark, Jan 2026)