Why B2B contact data goes stale, and what to do about it

Where decay comes from

Contact records go wrong in several distinct ways, and they need different responses. Lumping them together as "bad data" is why clean-up projects tend to fix the visible problems and miss the expensive ones.

• Job changes. The single largest source. The person is real, the company is real, but they no longer work together — and the record gives no hint that anything is wrong.
• Company changes. Acquisitions, rebrands and domain migrations invalidate addresses in bulk. One acquisition can break every contact at that company simultaneously.
• Role changes within a company. The person is still there but no longer owns what you care about, so your message goes to someone who cannot act on it.
• Entry errors. Typos, wrong fields, inconsistent formats. The most visible category and usually the least costly.
• Duplicates. Created by imports, forms and manual entry, and the reason your contact count is higher than your contact reality.

Why the first two are the dangerous ones

A typo announces itself — the address bounces, and you fix it. A job change does not. The record still looks perfectly well-formed, the address may still accept mail because the mailbox was left open and forwarded, and nothing in your system flags it.

So you keep sending to someone who left eighteen months ago, and the metric that would reveal it — reply rate — degrades slowly enough to be attributed to the message rather than the data.

This is why bounce rate is a poor proxy for data quality. It catches the cheap errors and misses the expensive ones.

Measuring your own decay rate

Published industry decay figures vary wildly and none of them describe your database. Measuring your own is straightforward and takes an afternoon.

1. 1Take a random sample of 100 contacts, stratified by how long ago each was added — not the most recent 100, which will flatter you.
2. 2Check each one by hand: does the person still hold that role at that company?
3. 3Record the failure rate against record age. Plotting those two gives you your actual decay curve.
4. 4Repeat quarterly on a fresh sample. The trend matters more than any single measurement.

What to do with the result

Once you know roughly how fast your data degrades, maintenance becomes a scheduling question rather than a judgement call.

• Set a re-verification interval shorter than the point at which your curve gets uncomfortable. If a quarter of records are wrong at eighteen months, do not wait eighteen months.
• Re-verify before campaigns rather than on a calendar. The cost of a stale record is only realised at send time, so that is where the check belongs.
• Prioritise by value. Re-verifying your entire database uniformly wastes effort on records nobody will contact. Start with active pipeline and target accounts.
• Automate it. Data hygiene that depends on someone remembering will not happen consistently, and a scheduled job with a spending ceiling removes both the remembering and the risk.

Prevention, where it is possible

You cannot stop people changing jobs, but several practices meaningfully slow the rate at which your records become unusable.

• Capture a stable identifier for every contact — a profile URL rather than only a name. It survives job changes and gives you something to re-resolve against later.
• Timestamp every record with when its details were last confirmed. One column, and it makes staleness visible instead of invisible.
• Standardise on entry: one format for names, one for company names, one for domains. Cheap at entry, expensive to retrofit.
• Store what you could not resolve rather than dropping it. A record marked "no verified address, no employer listed" is information; a silently deleted record is a gap you will rediscover.

The realistic goal

Perfect data is not achievable and pursuing it wastes effort that would be better spent elsewhere. The goal is a database where you know how good it is, where the confidence in each record is visible, and where maintenance happens without anyone championing it.

That is a lower bar than "clean data" and a much more useful one, because it is actually reachable.

ReachNow supports this pattern: re-looking up a record you have already enriched costs nothing, enrichment runs on a schedule with a credit ceiling, and unresolved records come back with a reason rather than being dropped — so your coverage figures reflect reality.