Article
How Accurate Is Email Verification Really? A Practical Answer
By Unlimited Verifier Team ·

Summary
Email verification is nearly exact on syntax, domain, and MX checks, but mailbox-level results are estimates, so there is no single accuracy number. Judge a verifier by its false-valid rate, false-invalid rate, and undecided share, measured against your own ESP bounce records.Email verification is nearly exact on syntax, domain and MX checks but only an estimate on the mailbox check, so there is no single accuracy figure. Judge it by how often a Valid result still bounces, and keep that under 2%.
Most people asking this want a yes or no. The honest answer is that verification is accurate enough to drive decisions, but only if you define accuracy the way a sender experiences it. For the full picture, see our complete email verifier accuracy guide. Below, we separate what verifiers can prove from what they estimate, show the numbers that matter, and give you a test you can run on your own list.
Quick answer
- No tool is 100% accurate, and no single number tells the whole story. Measure two error rates: the false-valid rate (how many Valid addresses bounce) and the false-invalid rate (how many Invalid addresses were actually delivering). Add the undecided share (how many land in Risky).
- Syntax, domain and MX checks are close to deterministic. Most errors and "can't tell" results come from the SMTP mailbox probe and from catch-all domains.
- Vendor figures rest on different ground truths. Compare tools only after testing them against your own ESP bounce records.
- Recommended thresholds: keep bounces to Valid addresses under 2%, and aim for under 1% on your first sends after cleaning.
- A practical self-test is cheap. A 2,000-address sample fits in one Pro list and uses 2,000 of the 25,000 deep verifications included each month.
What does "accuracy" mean for an email verifier?
Accuracy is how often a verifier's label matches what happens when you mail the address. A verifier never sends a message, so its labels are predictions. You judge them against outcomes you already have in your ESP logs.
Four measures matter:
| Measure | Definition | Why it matters | How to check it |
|---|---|---|---|
| False-valid rate | Share of Valid addresses that later hard-bounce | Drives your bounce rate and sender reputation | Cross-reference Valid results with ESP bounce records |
| False-invalid rate | Share of Invalid addresses that were actually delivering | Quietly deletes real customers | Check Invalid addresses that have a recorded successful delivery |
| Undecided share | Share of addresses returned as Risky or unknown | Shows how much manual judgement you still need | Count Risky results in the export |
| Stability | Whether the same address gets the same label on re-check | Reveals noisy or transient results | Re-run a sample after a few days and compare |
The two error rates pull against each other. A stricter tool lowers false-valid results but flags more real addresses as Risky or Invalid. We cover this trade-off in more depth in false positives vs false negatives in email verification.
For most senders, a false Valid is the expensive error, because it damages reputation. For acquisition teams, a false Invalid is the expensive one, because it deletes reachable prospects. Decide your tolerance for each before you compare tools.
Bottom line: Accuracy is two error rates plus an undecided share, and you can only measure it against outcomes you have already seen.
Which parts of email verification are reliable, and which are estimates?
The first three checks are rule-based and rarely wrong. The mailbox check is an inference from a server's reply, and that is where uncertainty lives.
| Check | What it does | How reliable it is |
|---|---|---|
| Syntax | Tests format rules, such as one @ sign, legal characters and a dotted domain | Near-deterministic. Some syntactically legal addresses are still rejected by particular servers |
| Domain and MX | Looks up whether the domain publishes mail-exchanger DNS records | Accurate at the moment of the check. Domains expire and change, so stale lists need re-checking |
| SMTP mailbox probe | Connects to the receiving server, names the sender and recipient, reads the reply, then disconnects without sending a message | Probabilistic. The reply is interpreted, not proven |
| Catch-all detection | Tests a made-up address on the same domain. If it's accepted, the server accepts everything | Shows the mailbox can't be confirmed, so these addresses belong in Risky |
The probe is the part to understand. RFC 5321 defines the SMTP dialogue, and the reply codes carry the meaning. A 250 reply to the recipient step suggests acceptance. A 550 reply with enhanced status code 5.1.1, which RFC 3463 defines as "bad destination mailbox address", suggests rejection. A 4xx reply is temporary, which is typical of greylisting or rate limiting. A 4xx should produce Risky or unknown, not Invalid. Treating every non-250 response as a dead address is one of the most common ways a list gets over-cleaned.
Two invisible factors also shape results. Many cloud and residential networks block outbound port 25, so probes must run from networks that allow it. And large consumer providers generally reveal little to outside probes, so verifiers often get ambiguous answers for those domains. That is why two tools can disagree about the same address at the same minute.
Bottom line: Trust the syntax, domain and MX results. Treat the mailbox result as a well-informed estimate, and expect a good verifier to show its uncertainty as Risky rather than force a Valid.
How accurate are results for different kinds of addresses?
Accuracy is not uniform across a list. It is highest on well-run corporate mail servers and lowest on catch-all domains and long-dormant addresses. Lists built for email verification for ecommerce and SaaS purposes, such as signups, trials and demo requests, usually mix all of these types.
| Address type | How confident a verifier can be | Why | Practical handling |
|---|---|---|---|
| Corporate domain with its own mail server | High | The server usually answers the probe definitively | Valid results can go in early sends |
| Large consumer providers (Gmail, Outlook, Yahoo) | Good on format and domain, less granular on mailbox | Providers limit probe detail | Use Valid results, then watch early bounces. See Gmail vs Outlook accuracy |
| Catch-all business domains | Low on mailbox existence | The server accepts every address, so acceptance proves nothing | Keep in Risky and send last, in small tests. See accuracy on catch-all domains |
| Role addresses (info@, sales@) | High that the mailbox exists, low that anyone reads it | Shared inboxes get low engagement and more complaints | Send only with clear consent, and segment them |
| Disposable domains | High where the domain appears on a maintained list | Detection depends on list upkeep, and new providers appear | Suppress where the signup should have required a real contact |
| Long-dormant addresses | The mailbox may exist, but nothing shows whether anyone uses it | Abandoned addresses can become spam traps | Segment by recent engagement, not verification alone |
Verification answers one question: can this mailbox receive mail right now? It cannot answer whether the person wants it. Our catch-all email verification page explains how catch-all results are flagged so you can hold them back. Does a valid result mean the email will be delivered? covers the gap between a Valid label and an inbox placement.
Bottom line: Accuracy depends on the address type. Split your results by type before you judge a tool.
Why do different verifiers disagree on the same address?
Verifiers disagree for five practical reasons:
- Probe origin. Each tool probes from its own network, and some networks get better answers from certain providers.
- Timing. A mailbox can be created, disabled or filled between two checks, and caching windows differ.
- Label mapping. One tool's "Risky" can be another's "Valid" or "Invalid". Mapping alone can make two tools look several points apart even when they agree on the facts.
- Catch-all policy. Some tools call every catch-all Risky, while others try to sub-classify them.
- Re-check behaviour. Some tools re-test failed probes before returning a result, and others don't.
When you compare the best email verification tools on a shortlist, reconcile the labels first, then examine the disagreements. Our article on why ZeroBounce and NeverBounce disagree on the same email walks through one such case.
Bottom line: Disagreement is useful information. Map the labels onto one scale before you judge which tool is more accurate.
What does a vendor's accuracy figure really tell you?
A vendor's figure is a claim about its own methodology, and methodologies are rarely comparable across vendors. Unlimited Verifier states 99.5% accuracy for deep verification. That is the company's own figure. We are not aware of a widely adopted independent benchmark that tests all vendors on the same data.
Vendors can differ on:
- Ground truth. Outcomes from real bounces are far more convincing than a verifier re-checking its own results, which is circular.
- Population. Accuracy on fresh signups is not accuracy on old imports.
- Label scoring. Some figures count Risky as a miss, and others exclude it.
- Time window and sample size. Mailboxes change, and small samples carry wide error bars.
The table below compares ways to judge a claim:
| Evidence | What it proves | Main weakness |
|---|---|---|
| Vendor's internal test | The vendor's own performance on its chosen data | Often undisclosed population and scoring, so hard to reproduce |
| Independent benchmark on shared data | Relative performance across vendors on identical inputs | Rare, and quickly out of date as mailboxes change |
| Your ESP bounce cross-tab | Performance on your own population, against real outcomes | Limited to addresses you have already mailed |
| Seed addresses you control | Behaviour at specific providers, under conditions you set | Costly to maintain and can't show inbox placement |
Even a very high figure leaves many errors at scale. At 99.5% accuracy, 0.5% of 1,000,000 addresses is 5,000 wrong labels. That is why bounce monitoring still matters after cleaning. Our analysis of whether vendor accuracy claims are trustworthy lists the questions to put to any vendor, and how email verifiers measure accuracy explains the methods behind headline numbers. For a shortlist-level view, which email verifier is the most accurate shows why a single ranking rarely holds.
Bottom line: Treat a vendor's accuracy figure as the start of your evaluation, not the conclusion. Your own cross-tab is the evidence that counts.
How do you measure a verifier's accuracy on your own list?
Test on addresses whose outcomes you have already seen. This framework takes about an afternoon. For a longer walkthrough, see how to test an email verifier's accuracy yourself.
- Build a ground-truth sample. Take 1,000 to 2,000 addresses from a campaign sent in the last 6 to 12 months, with each outcome (delivered or hard-bounced) recorded in your ESP.
- Match the sample to your list. If most of your list is old imports, sample old imports. Accuracy on last month's signups says little about a 2019 import.
- Keep the outcomes on your side. Upload only the addresses. Join the outcomes back afterwards.
- Verify the sample and export the results.
- Build a cross-tab of Valid, Risky and Invalid against delivered and bounced.
- Calculate the four measures from the first table.
- Set your thresholds before reading the results. For example: a false-valid rate at or below 2%, and an undecided share your team can review by hand. Our guide to what a good unknown rate looks like helps you pick the undecided limit.
Hands-on tutorial: run the test with Unlimited Verifier
- Export two files from your ESP. The first holds addresses only, for upload. The second holds each address with its outcome, and stays on your computer.
- Create an account. Sign up to start. The Free plan caps each list at 150 emails, so a 2,000-address sample needs 14 Free lists or the Pro plan ($29/month, or $25/month billed annually), which allows lists of up to 25,000 emails.
- Upload the addresses file. Upload and verify a list accepts CSV or Excel files up to 10MB and detects the email column automatically.
- Run deep verification. Unlimited Verifier states its 99.5% accuracy figure for deep verification, so that is the tier to test.
- Export the results as CSV, Excel or JSON before they expire. Results are kept for 24 hours on the Free plan and 30 days on Pro.
- Join the results to your outcomes file on the email address, and build a pivot table with labels as rows and outcomes as columns.
- Calculate the four measures and compare them with the thresholds you set in step 7 of the framework.
- Spot-check surprises with the free email checker, for example an address that bounced but came back Valid.
One caution: a "delivered" status means the receiving server accepted the message. It does not prove that a person read it.
Worked example: a 2,000-address sample
The numbers below are illustrative. They show the arithmetic, not a measurement of any vendor.
Suppose your ESP shows 1,790 delivered and 210 hard bounces for a 2,000-address sample. The verifier returns 1,700 Valid, 150 Risky and 150 Invalid. The cross-tab looks like this:
| Verifier label | Previously delivered | Previously hard-bounced | Total |
|---|---|---|---|
| Valid | 1,666 | 34 | 1,700 |
| Risky | 110 | 40 | 150 |
| Invalid | 14 | 136 | 150 |
| Total | 1,790 | 210 | 2,000 |
The four measures:
- False-valid rate: 34 ÷ 1,700 = 2.0%. This sits exactly at the recommended ceiling.
- Dead addresses that passed as Valid: 34 ÷ 210 = 16.2% of all dead addresses.
- False-invalid rate: 14 ÷ 150 = 9.3% of Invalid addresses were live. As a share of all live addresses, that is 14 ÷ 1,790 = 0.8%.
- Undecided share: 150 ÷ 2,000 = 7.5% of the sample was Risky.
What this tells you:
- Sending only to Valid addresses reaches 1,666 of 1,790 live addresses, or 93.1%. Deleting all Risky addresses would discard 110 live contacts, or 6.1% of the live total.
- Valid still needs care. A 2.0% false-valid rate sits at the ceiling, not comfortably under it, so send to the oldest Valid addresses last.
- Keep Risky, don't delete it. It holds 110 live addresses and 40 dead ones, so a short, monitored test is worth running.
- Confirm high-value Invalid contacts through another channel before you delete them, such as a login or a form resubmission.
Cost check: this test fits on Pro. On the Free plan, 2,000 addresses would need 14 lists of up to 150 (2,000 ÷ 150 = 13.3, rounded up), and its 750 monthly deep verifications would cover only part of the sample. Standard verification carries no per-email fee on any plan. Email verification pricing is easiest to compare as cost per 1,000 addresses on the list you actually clean.
Bottom line: A 2,000-address test takes an afternoon and produces the four numbers that decide whether a tool's labels are good enough for your list.
What bounce rate should a verified list achieve?
Sources disagree on the threshold, so choose the stricter number. Amazon SES's documentation describes account review at a 5% bounce rate and a possible pause at 10%. Other platforms act at lower levels, especially on a single campaign to an unproven list. Google's sender guidelines set a user-reported spam threshold of 0.3%. That measures complaints rather than bounces, so it is a separate limit.
Our recommendation:
- Ceiling: 2% bounces to Valid addresses. This leaves room before most platforms' automated triggers.
- Target: under 1% for first sends after cleaning, especially on a list you haven't mailed for a while.
Keeping bounces low is part of email verification compliance and hygiene. It protects delivery and shows ESPs and regulators that you maintain your list.
Bottom line: Plan for 2% as the ceiling and 1% as the target. The ESP's trigger is the last line of defence, not the goal.
What should you do with Valid, Risky and Invalid results?
Each bucket carries a different risk, so each needs a different action.
| Bucket | What it usually contains | Send? | Action |
|---|---|---|---|
| Valid | Addresses whose mailbox the probe confirmed | Yes, first and most engaged | Monitor bounces on every send |
| Risky | Catch-all domains, disposable and role addresses, and temporary or ambiguous replies | Only in small, late tests | Segment by engagement and re-verify later |
| Invalid | Syntax failures, dead domains, missing MX and rejected mailboxes | Never | Suppress, after confirming high-value contacts through another channel |
For forms, real-time email validation catches bad addresses before they enter the list. If you need checks inside your own systems, the email verification API supports single-email checks and bulk batches of up to 10,000 emails per request, and Pro includes 100,000 API emails a month. Our guide to email verification API and automation explains the workflow patterns. Email list maintenance best practices covers how often to re-run checks.
Bottom line: Valid is for the first sends, Risky is for small tests, and Invalid is suppressed.
What are the most common mistakes when judging accuracy?
- Comparing headline figures that use different ground truths and different scoring of Risky results.
- Testing on the wrong population. A sample of fresh signups says little about an old import.
- Treating every non-250 response as dead. Temporary 4xx replies are not Invalid.
- Counting Risky as wrong, or as right, by default. Decide how each bucket is scored before you start.
- Measuring on a list you already cleaned. Survivor bias makes the remaining addresses look healthier than the original list.
- Using a stale sample. Mailboxes change, so a sample from two years ago tests a different population.
- Assuming a Valid label means inbox placement. It means the mailbox answered, not that the message reached the inbox.
- Deleting Risky wholesale. In the worked example, that would have discarded 110 live addresses.
- Skipping bounce monitoring after cleaning. Verification reduces bounces but doesn't remove the need to track them.
Bottom line: Most bad accuracy judgements come from comparing unlike things, so define the population, the ground truth and the scoring before you run any test.
How we compared (and limitations)
This article is based on public documentation: RFC 5321 and RFC 3463, Amazon SES's published thresholds, Google's sender guidelines, and the verification methods vendors describe publicly. We have not run controlled tests against any vendor, and we have not independently verified any vendor's accuracy claim.
The 99.5% figure for deep verification is Unlimited Verifier's own stated figure. The worked example uses illustrative numbers to show the arithmetic. Plan limits are the ones listed for Unlimited Verifier. Sending thresholds vary by platform and can change without notice, so check the current documentation of the ESP you use.
Bottom line: Use the framework and thresholds here, and let your own bounce data, not any vendor's headline number, decide whether a verifier is accurate enough for your list.
Self-test: measure a verifier against your own bounces
- Export 2,000 addresses at random from your list.
- Verify them and record each label (Valid, Risky, Invalid).
- Send one campaign to the Valid group only.
- Pull hard bounces from your ESP after 48-72 hours.
- False-valid rate = Valid addresses that hard-bounced / total Valid.
- False-invalid check = Invalid addresses with a recorded delivery or opens.
- Compare results against your thresholds: under 2% false-valid, and a Risky share you can handle manually.
Frequently asked questions
Is email verification 100% accurate?
No. No email verifier is 100% accurate. Syntax, domain, and MX checks are highly reliable, but the mailbox probe is an inference from server responses, so some Valid results still bounce and some Invalid addresses may be deliverable.
What is a good false-valid rate?
Many senders aim to keep bounces on addresses marked Valid under 2%, and under 1% on first sends after cleaning. Your own ESP bounce records are the right benchmark for checking this threshold.
What is a false positive in email verification?
It depends on the framing. For senders, a false positive usually means an address marked Valid that later bounces. Teams should define the term explicitly, because some vendors use the reverse meaning.
Why do catch-all domains reduce accuracy?
Catch-all servers accept mail for any address, so the mailbox probe cannot confirm whether a specific person exists. Verifiers usually return Risky or an unknown status for these addresses.
How do I test an email verifier's accuracy?
Take a sample of about 2,000 addresses, verify them, send a campaign to the Valid group, and compare bounces with the verifier's labels. Also check Invalid addresses for recorded successful deliveries.
Why do vendors report different accuracy figures?
Vendors use different ground-truth datasets, sampling methods, and definitions of accuracy. Compare tools only after testing each one against the same set of outcomes from your own sending history.
Should I delete addresses marked Risky?
Not automatically. Risky addresses are undecided, so teams often suppress them for cold outreach, keep them with lower send frequency, or decide based on engagement history.
Do verification results change over time?
Yes. Mailbox status can change, and some results are transient, so re-checking a sample after a few days shows how stable a tool's labels are.
Cut Bounces and Protect Your Sender Reputation With Unlimited Email Verification
Every invalid address you send to hurts your sender reputation and your inbox placement. Verifying your list before each send means fewer bounces, healthier deliverability, and a predictable cost. With Unlimited Verifier, standard verification is unlimited on every plan, with no per-email fees.
Unlimited Verifier checks syntax, domain, MX records and the SMTP mailbox. It also detects catch-all domains and flags disposable and role-based addresses. Export a CSV from your email platform or CRM, upload and verify the list, then re-import the clean results. You can also automate checks with the email verification API.
| Free | Pro | |
|---|---|---|
| Price | $0 | $29/month ($25/month billed annually) |
| Lists | 25 a month, up to 150 emails each | Unlimited, up to 25,000 emails each |
| Deep verifications | 750 a month | 25,000 a month |
| API access | No | Yes (100,000 emails a month) |
| Result retention | 24 hours | 30 days |
Why teams choose Unlimited Verifier
- Unlimited standard verification on all plans, including Free, with no per-email fees.
- Catch-all detection so you know which domains accept every address.
- Clear Valid / Risky / Invalid results, exportable as CSV, Excel or JSON.
- An open-source verification engine you can inspect on GitHub.
Unlimited Verifier states 99.5% accuracy for deep verification.
Create a free account and verify your next list before you hit send.
Want to test a single address first? Try the free email checker.