How to reactivate dormant customer databases safely: bounce thresholds and toxicity scoring
Example provider: Bouncer, a Cllimber client
What a safe database reactivation is
A safe reactivation verifies a dormant list before anything is sent, cuts it to a segment whose expected bounce rate sits inside the sending platform's limits, sets aside addresses that accept mail but carry known risk, and then mails what is left in stages.
Dormant databases decay in ways the CRM never shows. People leave jobs and their work mailboxes close; domains lapse; free-mail accounts are abandoned, and some abandoned addresses are later turned into spam traps by mailbox providers. None of this changes the record in the client's ESP. A list that was healthy when it last received a campaign can come back two years later with enough hard bounces to put the sending account under review on the first send.
For an agency, the reputation at stake is the client's: the domain a blind reactivation damages is the one the client's order confirmations, password resets and revenue campaigns rely on. So the decision about which addresses to mail is made on evidence: what the provider expects to bounce, and what it expects to cause harm even when it doesn't.
What goes wrong when a dormant list is mailed unverified
- Hard bounces: addresses that no longer exist return permanent failures, and sending platforms act on the rate.
- Spam traps: recycled trap addresses accept mail and never engage; hitting them tells mailbox providers the list is poorly kept.
- Complaints: people who have forgotten the brand mark the email as spam, and platforms often police complaints more tightly than bounces.
- Account review: one high-bounce send can trigger a review or pause on the sending account, holding up every campaign on it.
Verification deals with the first two directly and the third only indirectly. A technically valid address is not necessarily an engaged or consenting recipient, and verification is one component of deliverability alongside authentication, content, engagement and sender reputation.
The bounce thresholds a reactivation has to clear
Sending platforms publish the bounce rates they act on. Amazon SES, for example, advises staying below 2%, places an account under review at 5% and may pause sending at 10%; a reactivation segment has to fit under whichever limit governs the client's account.
Thresholds differ by platform and many ESPs publish none, so the first job is to find out which platform sends the client's mail and what it says. The SES figures are a useful reference because they are explicit, turning "keep bounces low" into a number a segment can be built against.
The exact math of list cleaning: 957 addresses, six thresholds
In Cllimber's test, the first 42 addresses removed did almost all the work: they cut Bouncer's bounce estimate from 4% to 1.6%. Every further step bought less, and the move towards maximum safety cost 259 addresses for half a percentage point.
On 8 September 2026 Cllimber ran a 957-address consumer list through Bouncer. The run finished in 20 seconds and returned 874 deliverable, 54 risky, 20 undeliverable and 9 unknown. Bouncer's export screen then offers a slider of preset positions from all results to maximum safety, each showing how many addresses it keeps and a live bounce estimate. The table records every position.
| Export position | Addresses kept | Given up at this step | Bounce estimate |
|---|---|---|---|
| All results | 957 | — | 4% |
| Maximum audience | 915 | 42 | 1.6% |
| Position 3 | 910 | 5 | 1.3% |
| Position 4 | 864 | 46 | 1.1% |
| Position 5 | 605 | 259 | 0.6% |
| Maximum safety | 601 | 4 | 0.6% |
Cllimber hands-on test of Bouncer, 957-address consumer list, 8 September 2026. Bounce estimates as displayed on Bouncer's export screen.
The marginal cost is the number to show a client. The first cut removes addresses expected to bounce: the undeliverable results, the unknowns and the riskiest of the risky. The deep cut removes addresses Bouncer expects to deliver, on toxicity grounds. Those are different decisions. One protects the bounce rate; the other is a judgement about whether working addresses with a bad history should be mailed at all, and the client pays for it in reach.
Two more figures from the same run matter for budgeting. Before any credits were spent, the upload screen ran a free pre-check that estimated 306 toxic addresses and a 4.3% bounce rate, close to the 4% the full run produced. And 948 credits were charged for 957 addresses: the nine unknown results were not billed, in line with Bouncer's stated policy of not charging for unknowns or duplicates (client-reported, Bouncer pricing page, read 2 Oct 2026).
Deliverable status against the 0–5 Toxicity Check
A deliverability status answers whether an address will accept the email; a toxicity score answers whether it should be mailed at all. In Cllimber's test, 273 addresses Bouncer classed deliverable were still excluded at the strictest export setting on toxicity grounds.
Bouncer returns both for every address. The status is one of four lowercase values: deliverable (the provider confirmed the address exists), risky (an accept-all domain, a full mailbox or a disposable address), undeliverable (invalid syntax, no valid domain, or rejected by the server) and unknown (no usable response from the provider). The toxicity score, a whole number from 0 to 5 in Bouncer's API, rates something else. Bouncer says its Toxicity Check flags widely circulated addresses, breached and hacked addresses, and addresses belonging to complainers, litigators and potential spam traps, and that the higher the score, the higher the probability of hitting one (client-reported).
| Signal | Deliverability status | Toxicity score (0–5) |
|---|---|---|
| Question it answers | Will the receiving server accept mail for this address? | Is this address likely to harm the sender if it is mailed? |
| Values | deliverable, risky, undeliverable, unknown, each with a reason code | 0 to 5; a higher score means a higher probability of a harmful address |
| What it catches | Closed mailboxes, dead domains, invalid syntax, full mailboxes, disposable and accept-all addresses | Potential spam traps, complainers, litigators, breached and widely circulated addresses |
| What it cannot tell you | Whether a working address has a harmful history | Whether the mailbox exists today |
| In Cllimber's test | 874 of 957 addresses deliverable | 273 deliverable addresses excluded at the strictest setting; every address scoring 4 or 5 came back risky, not undeliverable |
| Role in a reactivation | Sets the bounce floor | Sets the risk appetite: keep, suppress or quarantine |
Statuses and the 0–5 range: Bouncer API documentation, read 2 Oct 2026. Toxicity categories: Bouncer Toxicity Check page, client-reported, read 2 Oct 2026. Test figures: Cllimber, 8 Sep 2026.
The fifth row is the practical one. A rule written only on status, such as "remove undeliverable", would have kept every address that scored 4 or 5, because none of them came back undeliverable. In single checks Cllimber ran on 11 September 2026, a mistyped address at gnail.com came back risky with toxicity 4, a disposable mailinator address came back risky with toxicity 3, and a real Gmail address came back deliverable with toxicity 0.
A reactivation playbook built on both signals
Sample before paying, verify the whole list, set the bounce cut first, decide toxicity separately with the client, then send in stages and write each decision back to the ESP.
- Sample before spending: Bouncer's upload screen gives a free bounce estimate and toxic-address count before verification starts, and its free list-sampling page does the same on a random sample. In Cllimber's test, a 500-address sample of the list estimated 3.7%, against 4% for the full run.
- Verify the whole list: one run returns a status, reason code, score and toxicity score for every address. Cllimber's 957 addresses took 20 seconds.
- Set the bounce cut: choose the export position whose estimate fits under the sending platform's limit. In the test, maximum audience (915 addresses, 1.6%) was already under 2%.
- Decide toxicity with the client: the deep cut removes working addresses, so it is a business decision. Bouncer Recommends sorts addresses into keep, suppress and quarantine on rules the user can edit (client-reported).
- Send in stages: mail the most recently active segment first and watch bounce and complaint rates before releasing the next; a quarantine group can wait until the first sends show how the list behaves.
- Keep it clean afterwards: Bouncer's AutoClean verifies new contacts every hour and re-verifies existing ones every set number of days on HubSpot, Klaviyo, Brevo, Mailchimp, User.com and Inboxroad, so a reactivated list does not quietly decay again (client-reported, AutoClean page, read 2 Oct 2026).
What a reactivation looks like at scale: YOCTO
YOCTO, a lifecycle marketing agency working with high-volume DTC ecommerce brands, mainly in the United States, reports re-engaging 90,960 dormant customers at $4.56 per reactivated customer after adding Bouncer to its reactivation workflow.
The case study is Bouncer's own, and the figures are client-reported. It describes the workflow in outline: legacy databases validated before use, dormant profiles segmented, and inactive contacts reintroduced to campaigns with keep-or-suppress guidance, rather than the whole historical database mailed at once. It does not publish list sizes, bounce rates or timescales, so the per-customer figure cannot be broken down further.
A tested example: Bouncer
Bouncer is an EU-based email verification and list hygiene platform, founded in 2017 and based in Wrocław, Poland. Cllimber tested it on 8 and 11 September 2026 with a 957-address consumer list, four single checks and a free sample.
The run behind the figures on this page returned a status, reason code, score and toxicity score per address, with flags for disposable, role, free-provider, full-mailbox and catch-all addresses; the export presets can be edited and saved as custom rules. Of the 199 addresses on catch-all domains, 164 were classed deliverable. Pay-as-you-go credits start at $8 for 1,000 and do not expire (Bouncer pricing page, read 2 Oct 2026), which suits agencies whose volumes follow client campaign calendars.
The full agency test, including Organizations, team access and pricing, is in Bouncer: email verification software for marketing agencies.
What to check before reactivating a client database
Check the sending limit, the contract, the charges and how each decision gets back into the client's ESP before a single address is uploaded.
- Which limit applies: the client's ESP or sending platform, and the bounce and complaint rates it acts on
- Who is the processor: the client is the controller; an agency verifying its list is a processor and the verification provider a sub-processor, which needs the client's authorisation and a data processing agreement
- What is charged: whether unknown results and duplicates use credits, and whether credits expire between campaigns
- Whether toxicity is included: a deliverability status alone cannot separate working addresses from harmful ones
- How results get back: an export file, or recommendations written to the ESP so flows can branch on keep, suppress and quarantine
- Whether consent still holds: verification shows that an address can receive mail, not that its owner agreed to it
Reactivating dormant databases, answered.
How do you reactivate a dormant email database safely?
Verify every address, choose an export threshold whose bounce estimate fits under the sending platform's limit, decide separately whether to mail deliverable addresses with high toxicity scores, then send in stages and watch bounce and complaint rates. In Cllimber's test of Bouncer on 8 September 2026, removing 42 of 957 addresses cut the bounce estimate from 4% to 1.6%.
What bounce rate is safe for a reactivation campaign?
It depends on the sending platform. Amazon SES advises keeping bounces below 2%, places accounts under review at 5% and may pause sending at 10% (AWS documentation, read 2 Oct 2026). Build the reactivation segment to the limit the client's own platform sets.
What is the difference between a deliverable status and a toxicity score?
The status says whether the mail server will accept mail for the address. The toxicity score, 0 to 5 in Bouncer's API, says how likely the address is to be harmful to mail, from potential spam traps to known complainers. In Cllimber's test, 273 deliverable addresses were excluded at the strictest setting on toxicity grounds.
How many addresses should an agency remove before a reactivation?
Enough to bring the segment under the platform's bounce limit, then as many more as the client's appetite for risk justifies. In Cllimber's test the first 42 addresses cut the estimate by 2.4 points; the step towards maximum safety cost another 259 for 0.5 points.
Does email verification guarantee a reactivation reaches the inbox?
No. Verification removes addresses expected to bounce and flags risky and toxic ones; inbox placement also depends on authentication, content, engagement and sender reputation. Bouncer's own guarantee is narrower: a refund or credit where an address it classed deliverable bounces within 72 hours for a qualifying reason (client-reported).
How much does it cost to verify a dormant database with Bouncer?
Bouncer sells pay-as-you-go credits from $8 for 1,000 to $2,000 for 1,000,000, with a 1,000-credit minimum and 100 free credits to start. Purchased credits do not expire, and unknown results and duplicates are not charged (Bouncer pricing page, read 2 Oct 2026). Cllimber's 957-address run used 948 credits.