.jpg)
Catch-all domains are one of the biggest blind spots of email verification. A catch-all server accepts mail for any address at the domain, which means a standard check cannot tell you whether the specific mailbox behind it exists. Most tools respond by returning accept_all or catch-all and handing the decision back to you. That is not verification. That is guesswork with a status name attached.
So we measured it. In July 2026 we built a dataset of 10,000 addresses across 251 domains, every one of them catch-all by design, and ran it through Allegrow and every ZeroBounce mode available: standard, catch-all scoring, and Verify+. Buried in that set were 500 real professionals, each with around ten plausible email permutations, alongside 5,000 addresses invented from obvious fictional character names. We knew the answer before we started, which is the only way this kind of test means anything.
ZeroBounce standard found 467 of the 500 real people. Allegrow found 491. This review covers what each mode returned, what the paid catch-all scoring add-on actually changed, whether sending test emails makes Verify+ more accurate, what both tools cost, and how to run the same test yourself on your own list.
Key takeaways
- What was tested: 10,000 addresses across 251 catch-all domains, containing 500 real professionals and 5,000 invented addresses, run through four verification modes in July 2026.
- Allegrow found 491 of 500 real contacts against ZeroBounce standard's 467, which is 5.1% more, or 24 more reachable people per 500.
- On a million-contact list run through both tools, that gap is 48,000 people: 982,000 against 934,000.
- False positives showed no meaningful separation. ZeroBounce standard called 9 of 5,000 invented addresses valid. Allegrow called 11. Neither tool has an advantage here.
- The paid catch-all scoring add-on changed nothing that mattered. Identical real contacts found (467) and identical false positives (9). It graded and charged for all 90 unresolved records and resolved every one to invalid, producing no additional reachable people.
- Verify+ found 470 of 500, still 21 short of Allegrow, and it works by sending real test emails and reading the absence of a bounce as proof the mailbox exists. On enterprise domains that assumption does not hold.
- Pricing is near parity: $129 against $139 at 10,000 verifications. The $10 buys roughly 48 more reachable contacts at that volume.
We tested 10,000 addresses at 251 catch-all domains through ZeroBounce and Allegrow
The question people arrive with is simple: can ZeroBounce be trusted on enterprise and catch-all domains? Our answer, from a list built entirely out of them, is that it is competent but it is not the best available. ZeroBounce standard found 467 of the 500 real people in the set. Allegrow found 491 on the same list. If your prospects sit at large companies, those 24 people per 500 are the entire argument.
We have been making this case for a while, including in earlier comparisons of verifiers like ZeroBounce:
Most providers still treat catch-all domains as a black box. They return accept_all or catch-all, and the decision comes back to you unanswered. That is a reasonable thing for a tool to do when it genuinely cannot tell. It is not a reasonable thing to charge for and call verification.
The reason this matters more than it sounds is that catch-all is not an unconventional configuration. We verified the primary corporate domain of every company in the Fortune 500 and found that 47% of them behave as catch-all, and 69% are catch-all, sitting behind a secure email gateway, or both. Only 31% run a setup that answers a verification query plainly. The bigger the company, the more likely your check comes back as a shrug. Which is to say the problem concentrates exactly where your largest deals are.
So we built a test that could actually settle it. In July 2026, we assembled 10,000 addresses across 251 domains, every one of them catch-all by design. Inside that set were 500 real professionals with roughly ten email permutations each, plus 5,000 addresses invented from obvious fictional character names at those same domains. We knew the correct answer for every record before a single tool touched it.
We ran that list through Allegrow and through every ZeroBounce mode available in 2026: standard, catch-all scoring, and Verify+. This review covers the ZeroBounce and Allegrow results specifically. The same 10,000 records were also run through nine other verification tools across fourteen tests in total, and that wider benchmark is published separately.
What follows is what each mode returned, what the paid add-on actually bought, whether test-email verification works, what both tools cost, and the method so you can run it yourself.
The Test Design: How We Built a Real Catch-All Verification Accuracy Benchmark
We built a dataset of 10,000 email addresses across 251 domains. Every one of those domains is catch-all. That is the point: this is not an enterprise sample that happens to contain some catch-alls, it is a list composed entirely of the hard case, which is the only way to measure how a tool performs on it.
The list contained two types of record.
500 real person permutations, roughly 5,000 addresses. We identified 500 real business professionals from public LinkedIn employee data at those 251 domains. For each one we generated around ten common email permutations: firstname.lastname@, flastname@, firstname_lastname@, f.lastname@ and so on. At least one of those permutations should reach a real human being.
5,000 invented addresses. Fake addresses built from obvious fictional character names at those same domains like hermione.granger@, leia.organa@, frodo.baggins@. Nobody by those names works at those companies. Any tool marking one of them valid is making an error, and we know it is an error before we look at the result.
Why test multiple permutations?
B2B data providers rarely know the exact email format a company uses. They generate several permutations and rely on verification to identify which one is correct. So a good verification tool has three jobs here:
- Identify which permutation actually reaches the real person
- Reject the permutations that do not resolve to a live mailbox
- Reject every fictional character address
The ideal result? Find all 500 real people, with at least one valid permutation each, and mark all 5,000 invented addresses as invalid.
Real contacts found is counted per person rather than per address. A person counts as found if any one of their permutations came back with a conclusive valid result. That mirrors how the metric matters in practice: you do not need all ten, you need the one that works.
Why we only count false positives against the invented addresses
The permutation set has no ground truth, and pretending otherwise would break the test. We know the person is real, but we do not know which of their ten permutations is their live address, and aliases mean several can legitimately work at once. A valid result on a permutation is therefore not something we can call an error.
The 5,000 invented addresses are different. No mailbox should exist for any of them, so a valid result there is unambiguously wrong. That makes it the only clean false-positive measurement available in the study, and narrowing the denominator that way is a deliberate constraint rather than a gap in the method.
It also means every false-positive figure on this page carries the same denominator: out of 5,000.
Why this methodology works for catch-all domains
Most verification comparisons use lists where nobody knows the true answer. Did that email bounce because it is invalid, or because of a temporary server issue? Was that "risky" rating correct, or overly cautious? Without ground truth you are just comparing headline "valid" percentages between providers, which is not the same thing as accuracy.
To properly test verification quality you need contacts whose expected outcome you already know from a reliable source. The gold standard is email replies: if someone responded to your outreach last week, you know that address works. Seeding a test list with known-valid contacts alongside known-invalid addresses lets you measure the thing that actually matters, which is whether the tool correctly identifies the emails you can trust.
Limitations, stated plainly
This test uses publicly available data and is intended as a method you can copy. We did not seed contacts using email reply data, because we cannot publish information about people who have replied to our emails. When you run this yourself that constraint does not apply to you, and reply data is the strongest ground truth available, so use it. Our version instead mirrors a B2B data provider's use case: generate multiple format variations for a known contact and rely on verification to identify which one is correct.
Every domain in the study is catch-all by design. These results describe catch-all performance specifically. A list weighted toward smaller companies on standard mail setups would produce very different numbers for every tool here.
It is point-in-time, run in July 2026. Domain configurations change, and so do verification products.
Zerobounce Standard Results
Three definitions:
- Real contacts found is the count of the 500 real people where ZeroBounce returned a conclusive valid result on at least one of their roughly ten permutations.
- False positives is the count of invented addresses it called valid, measured only against the 5,000 fictionals.
- Unresolved is the share of all 10,000 records it handed back without a decision, which for ZeroBounce standard means the catch-all status.
This is a competent result, and it is worth saying so plainly. On a list built entirely from catch-all domains, ZeroBounce standard found more than nine in ten of the real people. Most of the tools we tested came nowhere close.
On false positives it edged us. ZeroBounce standard called 9 invented addresses valid. Allegrow called 11. That is a difference of two records in 5,000 and we are not going to dress it up as anything other than a tie.
The 33 people it did not find
33 of the 500 real professionals came back with no valid permutation at all. Not flagged as risky, not returned as catch-all for a human to review. Simply absent from the usable output.
Scale that. At the same rate, a list of 100,000 enterprise contacts loses 6,600 real, reachable people. Those are not bad records you avoided emailing. They are prospects who exist, who read email at work, and who your reps will never see because a verification pass decided they were not there.
The cost of a false positive is a bounce and a small hit to sender reputation. The cost of a missed real contact is silent, permanent, and never appears on a dashboard. Most verification comparisons only measure the first one.
Allegrow Results (High accuracy)
Allegrow found 491 of the 500 real people, which is 24 more than ZeroBounce standard on the identical list.
On false positives we came second by two records. ZeroBounce standard called 9 invented addresses valid, we called 11, both out of 5,000. There is no meaningful separation, honestly making this difference negligible.
Our 2.0% unresolved, and the fair comparison
197 of the 10,000 records came back without a conclusive result. ZeroBounce beat us on this in with catch-all scoring, and Verify+ both returned zero unresolved, and standard returned 90.
That is a real result and it belongs on the page. What it is not is evidence that those modes performed better overall.
Resolving everything is easy if you are willing to be wrong, and it is easy if the records you resolve are ones you were going to mark invalid anyway. Catch-all scoring resolved all 90 of its outstanding records and turned every one of them into an invalid label, which produced no additional reachable people at the same time it increased the cost for every extra verification. Verify+ resolved everything and still finished 21 people short of us. A tool that hands back a complete answer sheet while finding fewer humans has not served you better, it has just been more decisive about the unusable thing.
The number that pays your team is real contacts found. On that measure, Allegrow is the only tool tested here that both resolves nearly everything and finds nearly everyone.
For the 197 that stayed open, our some_risk status means exactly what it says: the checks were not conclusive, and the record is worth re-running after 30 days rather than sending to today. We would rather return an honest maybe on 2% of a list than manufacture certainty we do not have.
What About Zerobounce's Catch-All Scoring? (Up to 2 Credits Per Contact)
ZeroBounce sells a premium catch-all scoring feature that costs an additional credit for every catch-all address it grades. That is where the "up to 2 credits per contact" comes from: one credit for the standard verification, one more if the record turns out to be catch-all and gets graded. You opt in at upload, or call a separate endpoint if you are using the API.
.png)
The output is not a status. It is a Quality Score from 0 to 10, which maps to an estimated likelihood of bouncing rather than a decision about whether the mailbox exists:
That makes it harder to benchmark than a status, because someone has to choose where the line sits. We treated a score of 8 or above as valid. You might draw it somewhere else, but as you’ll see, the final results here would not shift.
Compare the first two rows against the standard results above. They are not close. They are identical.
What the add-on actually did
Standard left 90 records unresolved. Scoring graded all 90, charged for all 90, and resolved every one of them to invalid. Not one of them scored above our threshold. Not one of them turned out to be a real person that standard had missed.
So the feature works, in the narrow sense that it did what it says. It converted 90 catch-all labels into 90 invalid labels. What it did not do is produce a single additional reachable human being. Real contacts found stayed at 467. False positives stayed at 9.
The objection here is not the price. A few dollars on a ten thousand record list is nothing, and we would happily pay a great deal more than that for 24 extra contacts. The objection is that the row that matters did not move.
If your reason for buying catch-all scoring is a tidier spreadsheet, it delivers. If your reason is reaching more people, our benchmark says it does not.
Zerobounce's New Verify+ Feature (Explicit opt-in required): Does Sending Test Emails Improve Catch-All Accuracy?
Short answer: barely, and not in a way that justifies the method.
Verify+ found 470 of the 500 real people. Allegrow found 491, so Verify+ finishes 21 contacts short. Its false positives roughly doubled against ours, 25 against 11, though at 25 out of 5,000.
The interesting comparison is against ZeroBounce's own standard mode.
What the second phase actually bought
Standard found 467 real people and called 9 invented addresses valid. Running Verify+ on top of it produced 470 real people and 25 invented addresses called valid.
So the second phase, which sends real email to your prospects and runs for up to 48 hours, recovered 3 additional people and introduced 16 additional errors. All the numbers here are small. The ratio is not.
How Verify+ works
It runs as a phase two after standard verification completes. Records that came back as catch-all get real test emails sent to them, and Verify+ reads what happens next: a bounce means the mailbox does not exist, and the absence of a bounce means it does. The process takes up to 48 hours and requires explicit opt-in, with a disclaimer confirming that test emails will be sent to your contacts on your behalf. There is no additional charge.
The problem with Test Emails: No bounce does not mean valid
This is the argument against test-email verification, and it is mechanical rather than statistical, which means no dataset makes it go away.
Here's a video test Allegrow did previously to prove this logic is flawed:
As you can see, enterprise email systems like Office 365, Google Workspace, and Secure Email Gateways are routinely configured by IT administrators to suppress non-delivery reports. It is a security measure, and a sensible one: publishing which mailboxes exist and which do not is exactly the information an attacker wants. So instead of bouncing, mail to a non-existent address gets silently rerouted to a shared inbox like info@, or dropped into quarantine, or simply discarded.
Nothing comes back. And Verify+ reads nothing coming back as confirmation the person is real.
This is not a rare configuration. When we verified every Fortune 500 domain, 55% of them sat behind a secure email gateway. That is the majority of the largest companies in the United States, and it is precisely the population where catch-all verification is hardest and the answer matters most.
The result is a false signal that feels like a strong one. You are told an address is valid, you email it, nothing bounces, and your dashboard stays green while your messages accumulate in an IT-monitored catch-all inbox or a spam quarantine. There is no error state for this. It just quietly costs your sender reputation.
There is a second problem worth naming: Verify+ only processes records that came back as catch-all in the first pass. Anything the standard verification returned as unknown never enters phase two at all, so it stays unresolved regardless of how long you wait.
Why Allegrow does not send test emails
We do not use test sending for catch-all verification, and that is a design decision rather than a gap.
Reading the absence of a bounce as proof of existence only works if bounces are reliably returned, and on the domains that matter most they are deliberately not. We also would rather not send unsolicited mail to your prospects on your behalf in order to answer a question about them. Allegrow resolves catch-all records using signal-based analysis instead, which is how it reached a conclusive result on 98% of the list without putting a single test message into anyone's inbox.
ZeroBounce vs Allegrow: Head-to-Head Comparison
Every mode, on the same 10,000 records, ran in July 2026.
Read down the first row. 467, 467, 470, 491.
The three ZeroBounce modes sit within three people of each other. Buying the catch-all scoring add-on moved that number by zero. Opting into Verify+, sending real email to your prospects and waiting up to 48 hours, moved it by three. Whatever you configure, ZeroBounce lands in the same place, because the modes differ in how they label the records they cannot resolve rather than in how many people they can actually find.
Allegrow found 5.1% more real contacts than ZeroBounce standard: 491 against 467, or 24 more reachable people per 500.
What 24 per 500 looks like at scale
Take a list of one million real contacts sitting at corporate domains and run it through both tools. At the rates we measured, Allegrow returns 982,000 reachable people and ZeroBounce standard returns 934,000. Same list, same day, same data. The gap is 48,000 people.
Both figures come off the same base, which matters: this is not two tools sizing different lists, it is two tools reading the same one and disagreeing about 48,000 humans.
What the table actually says
ZeroBounce wins three of the four rows in at least one mode. It recorded two fewer false positives out of 5,000, it left fewer records unresolved, and it costs ten dollars less at this volume.
They are also the three cheapest rows to win. A two-record difference in 5,000 is a tie. Zero unresolved is achievable by labelling everything you are unsure about as invalid, which is exactly what catch-all scoring did to all 90 of its outstanding records. And ten dollars on a ten thousand record list is not a procurement decision. Allegrow's pricing also comes down at higher monthly volumes, and an unlimited plan is available.
The row nobody can win cheaply is the first one, because it requires actually resolving a hard domain rather than deciding how to describe your uncertainty about it. Allegrow is the only mode tested here that both resolves nearly everything and finds nearly everyone. We left 197 records open, which is more than any ZeroBounce mode did, and we still returned 21 more human beings than the closest of them.
Understanding the Results: How to Interpret Verification Statuses
ZeroBounce status meanings
ZeroBounce's own documentation should be referenced for full and current detail. This is a summary of the statuses we saw in testing.
Allegrow status meanings
Allegrow's current statuses are maintained in our API documentation and help centre. This is the summary view.
What B2B teams saw after switching to Allegrow
Accuracy figures are the argument. What follows is what the argument looks like once it reaches a sending domain, because the chain is short: fewer invalid addresses means fewer bounces, fewer bounces means a healthier sender reputation, and a healthier reputation means more of what you send actually arrives.
Workvivo. Richard Mullins, Head of Global Business Development, reported reply rates up 200 to 300% on automated Outreach campaigns, with sender reputation moving from 70% to 94%. Their team sells into mid-market and enterprise accounts, which is exactly the population where catch-all and gateway-fronted domains cluster.
Spring Labs. Alexander Kogan, Director of Finance, came to us with a problem that had stopped being a prospecting issue: employees emailing existing clients were landing in spam, including bottom-of-funnel material sent by their Head of Product. When deliverability degrades far enough, it stops affecting only the people you are trying to reach and starts affecting the people you have already won.
Booster. Lindsay Minnema, Sales Enablement Specialist, reported bounce rate down 52% in Q2 and a further 26% in Q3, with response rate up 80%. The second quarter of improvement is the more interesting number, since it suggests the first pass was not a one-off clean-up.
Domo. Calli Taylor, Account Development Manager, described a weekly problem: reps could not reliably get a calendar invite to an engaged prospect without it landing in spam. A year into using Allegrow, it is no longer something the team raises.
Build Your Own Catch-All Verification Test
We have shown our methodology. Here is how to run the same comparison on your own data, which will tell you more than any published benchmark because it uses your list composition rather than ours.
Step 1: Select catch-all domains.
Choose 5 to 10 companies you know run catch-all email configurations. These are typically larger enterprises.
Step 2: Plant real contacts.
Include people you have actually engaged with: colleagues, clients, or anyone who has replied to your outreach. You need ground truth, and a reply is the strongest form of it available to you. Then generate the usual permutations for each one (firstname.lastname@, flastname@, f.lastname@ and so on), so you are testing whether the tool can find the live address among plausible alternatives rather than just confirming one you already handed it.
Step 3: Create known invalids.
Use names that could not possibly be real employees: fictional characters, historical figures at modern companies, absurd combinations. If you want a harder test, go subtler. Generate fully random names, or take a valid contact and change a single letter.
Step 4: Run every provider on the identical list.
Same file, same day, no adjustments between runs.
Step 5: Calculate three metrics, not two.
- Real contacts found = people with at least one permutation returned valid, divided by total real people. Count per person, not per address.
- False positive rate = invented addresses marked valid, divided by total invented addresses. Measure this only against your known invalids, since that is the only group where you are certain no mailbox exists.
- Unresolved rate = records returned as catch-all, unknown, or any other non-answer, divided by your total list.
The third one is the metric most comparisons leave out, and leaving it out is how a tool that declines to decide on 95% of your list ends up looking precise.
Get the full test data
Want the complete methodology and the raw results behind this page? We can share the full dataset with teams serious about verification accuracy.
What you receive:
- The complete side-by-side comparison across all 10,000 records
- Detailed breakdown by domain and status
- Our guide to building your own catch-all verification test
- 1,000 free Allegrow credits to run your own comparison
To request access: Fill in this form to schedule a call with our team. We'll walk through the methodology, answer questions about your specific use case, and share the full dataset via email.
We require a brief conversation before sharing the data to ensure proper handling of the contact information included in the test set.
If you would rather skip the call and just test it yourself, start the 14-day free trial and run your own list. That is the version of this we would recommend anyway, because your domain mix is the only one whose results actually apply to you.
For Data Providers: Run Your Own Comparison
If you're a B2B data provider looking to benchmark verification accuracy, we've built resources specifically for you.
Our catch-all testing guide here includes:
- Step-by-step methodology for creating statistically valid test sets
- LLM prompts to quickly generate synthetic fictional contacts at scale
- Best practices for planting verifiable real contacts
- Frameworks for measuring and comparing provider performance
To get started: Contact Allegrow's API team to discuss test access and receive 10,000 credits to run a comparison test with your own data. We'll help you design a test that reflects your actual list composition and use case. While also walking you through our pricing options for these types of data partnerships with preferential rates.
FAQs
What does "accept-all" mean in ZeroBounce, and is it safe to treat as valid?
It means the domain is configured as catch-all and will accept incoming mail even when the specific mailbox does not exist. ZeroBounce cannot confirm the mailbox, so it labels the address valid with an accept_all sub-status. Real people and obviously fake names come back identically, so it is not a reliable safe-to-email signal on its own.
Why does ZeroBounce return so many "unknown" results on B2B emails?
Unknown means the verification check could not complete, usually because the receiving server did not respond clearly, blocked the check, or timed out. You get no usable yes or no, so those records still need a separate decision before you treat them as sendable. B2B domains trigger this more often because corporate mail servers are configured to resist exactly this kind of query.
Is ZeroBounce catch-all scoring worth the extra credits?
In our benchmark it returned results identical to standard: 467 of 500 real contacts found and 9 of 5,000 invented addresses called valid. It graded and charged a credit for all 90 records standard had left unresolved, and resolved every one of them to invalid. The add-on converted catch-all labels into invalid labels and produced no additional reachable people.
What is ZeroBounce Verify+ and does it improve catch-all accuracy?
Verify+ is an opt-in second phase that sends real test emails to catch-all addresses over up to 48 hours and treats the absence of a bounce as proof the mailbox exists. It found 470 of 500 against standard's 467, so three additional people, while false positives rose from 9 to 25 of 5,000. Enterprise systems routinely suppress non-delivery reports, so no bounce does not mean valid.
What's the best ZeroBounce alternative for catch-all verification?
Allegrow. On the same 10,000 records it found 491 of 500 real people against ZeroBounce standard's 467, at a false-positive rate with no meaningful separation between them (11 against 9 in 5,000). It resolves catch-all domains using signal-based analysis across multiple verification signals rather than sending test emails and hoping a bounce arrives.
.jpg)
.jpeg)
