What the Confidence Score Represents
The confidence score is a probability estimate. A score of 80 means the infrastructure analysis and address pattern assessment point toward an 80 percent likelihood that the address is deliverable. A score of 30 means the signals point toward a 30 percent likelihood.
It is not a pass or fail judgment. It is a calibrated estimate based on what EmailAddress.ai can determine through non-intrusive SMTP probing, MTA fingerprinting, and domain-level infrastructure analysis. The score reflects the quality and consistency of those signals for that specific address at that specific domain.
Score Ranges and What They Mean
| Score Range | Classification | Recommended Action |
|---|---|---|
| 80 to 100 | High confidence | Send. Infrastructure signals strongly support deliverability. |
| 50 to 79 | Medium confidence | Send with caution. Some signals are ambiguous. Set realistic bounce expectations. |
| 25 to 49 | Low confidence | Suppress for cold outreach. May be usable for re-engagement if the contact has prior history. |
| 0 to 24 | Very low confidence | Suppress. Signals point strongly against deliverability. |
How the Score Is Calculated
The confidence score combines several weighted inputs from the resolution process:
- MTA behavior profile. The identified mail server type and its characteristic response patterns contribute a base probability range for addresses at that domain.
- Address format match. Addresses that match common naming conventions at the domain (for example, firstname.lastname where others on the domain follow the same pattern) score higher than addresses that do not fit the domain's observed patterns.
- Domain infrastructure quality. MX record configuration, IP reputation, and domain age all factor into the domain-level component of the score.
- Historical delivery signals. Where aggregate delivery history data is available for addresses at the same domain, this informs the score for new addresses being evaluated there.
No single factor determines the score. The model weighs all available signals together. A strong MTA profile can partially offset a weak address format match, and vice versa.
Setting Your Suppression Threshold
The right threshold depends on your use case and your tolerance for bounces:
- Cold outreach campaigns. Keep a higher threshold (send only above 75 or 80). Cold campaigns have no prior engagement baseline, and bounce rates directly affect sender reputation.
- Warm or opt-in lists. A lower threshold (send above 50) is reasonable when the contact has shown prior engagement. The combination of intent signal and moderate confidence score makes the risk acceptable.
- HCP and pharmaceutical outreach. Hospital and clinic domains frequently run catch-all configurations. Given the small, targeted nature of HCP lists and the high value of each contact, many teams in this segment send to scores above 60 and accept slightly higher bounce rates in exchange for list coverage.
- Deliverability-sensitive senders. If your domain's sending reputation is already under pressure, hold the threshold at 80 or above until reputation recovers.
How Confidence Scores Compare to Standard Verification Statuses
For non-catch-all addresses, EmailAddress.ai returns a standard status: deliverable, undeliverable, or risky. The confidence score is specific to catch-all addresses. It replaces the blunt catch-all or unknown label that most verifiers return with a number your team can use to make a segmented decision.
Think of the score as converting a binary unknown into a probability range. Instead of one pile labeled "catch-all: do not send," you get four tiers with different recommended actions. This is the practical difference between discarding 35 percent of your list and discarding only the genuinely low-confidence addresses within that group.
Scores Change Over Time
Catch-all domain configurations and infrastructure change. A domain that scored 40 six months ago may score 70 today if the organization updated its mail server software or changed its MX configuration. It is worth re-verifying catch-all segments before each major campaign rather than treating scores as permanent. EmailAddress.ai's monthly data refresh applies to HCP and B2B data records. For your own lists, re-verification before each significant campaign is a sound practice.
Frequently Asked Questions
What confidence score threshold should I use for cold outreach?
+
For cold outreach campaigns with no prior engagement history, most teams set their threshold at 75 or above. This range reflects infrastructure signals that strongly support deliverability while excluding addresses where signals are ambiguous. If you are targeting a new domain or sector, start at 80 and adjust down only if list coverage is too low to run the campaign.
Can a high-confidence catch-all address still bounce?
+
Yes. The confidence score is a probability estimate, not a delivery guarantee. A score of 90 means the available signals strongly support deliverability, but the actual existence of the mailbox cannot be confirmed without sending. In practice, high-confidence addresses bounce at rates significantly lower than the medium or low confidence tiers, but a small bounce rate should be expected even at the top of the range.
Do confidence scores expire or become outdated?
+
Scores can become outdated as domain infrastructure changes. Mail server software updates, MX record changes, and organizational email policy changes all affect the underlying signals the score is based on. For active campaigns, re-verifying catch-all segments within 30 to 60 days of the campaign send date is a reasonable practice.
Is the confidence score different from an email quality score?
+
Yes. Email quality scores from other platforms typically rate an address based on engagement history, domain reputation, and format validity. The EmailAddress.ai confidence score specifically addresses deliverability probability for catch-all domain addresses, based on infrastructure analysis and MTA fingerprinting rather than engagement data.
Ready to verify catch-all addresses?
EmailAddress.ai scores individual addresses within catch-all domains so you can send with confidence instead of suppressing your entire B2B or HCP list.