I ran three cold email campaigns last quarter across two domains. Both domains had been on Lemwarm for 45 days before I touched them. SPF, DKIM, DMARC — all clean. I was following the playbook exactly.
Campaign one hit 22% bounce rate on day one. Campaign two hit 18%. By the end of week two, both domains were flagged in Google Postmaster Tools, reputation dropped to "Low," and I was rebuilding from scratch.
The warmup hadn't failed. The data had.
I spent two months after that pulling apart exactly what happened, testing verification workflows, and measuring what actually moved bounce rates below the 2% threshold. The conclusion is uncomfortable for the warmup industry: you can warm a domain for six months and a single dirty list will torch it in 72 hours. Data quality isn't a complementary concern to deliverability — it's the primary one.
What Warmup Actually Does (And the Problem It Can't Touch)
Lemwarm, Mailreach, Instantly's built-in warmup — these tools work by having your inbox send and receive emails from a network of other inboxes, building artificial engagement signals that tell ISPs your domain is legitimate. They fix one specific problem: a new or cold domain being flagged as suspicious due to low historical volume and no established reputation.
They do not fix:
- Email addresses that don't exist anymore
- Role accounts (
info@,sales@,admin@) that bounce or get filtered by spam rules - Catch-all domains that accept any email at the MX level but silently drop or bounce on delivery
- Contacts who switched jobs six months ago and whose old address now hard-bounces
When you send to bad addresses, ISPs see the bounces. Those bounces signal "this sender has no idea who they're sending to." Once you cross 2% sustained bounce rate in Gmail's system, you get penalized at the domain level — and no amount of warmup history reverses that within days.
Why 20%+ Bounce Rates Are a Data Problem, Not a Reputation Problem
B2B contact data decays at roughly 30% per year — about 2.5% per month. That number compounds. A list you pulled from Apollo eight months ago has likely seen 20%+ of its contacts change roles, get laid off, or switch companies.
When I audited the two campaigns that blew up my domains, here's what the breakdown looked like after a retroactive pass through ZeroBounce:
- Hard bounces (mailbox doesn't exist): 14.2% of the list
- Catch-all addresses I'd treated as valid: 5.8%
- Role addresses: 2.1%
That 14.2% hard bounce number was the killer. It came from a list that was 7 months old, sourced from a combination of Apollo and Hunter.io, and never re-verified before sending.
The warmup tools had done exactly what they promised. My reputation coming in was fine. But the data destroyed it faster than 45 days of warmup could protect.
The Three Address Types Destroying Your Sender Reputation
Most bounce rate content focuses on "invalid" addresses and stops there. That's the wrong frame. Here are the three categories that actually matter:
1. Hard-invalid addresses
These never existed or the mailbox has been deleted. Classic hard bounces. They damage you immediately. Every serious email verifier catches these.
2. Catch-all addresses
The domain accepts any email (the MX record is configured to catch everything), so verifiers can't confirm whether the specific address is real. These represent the gray zone. In my testing across roughly 500 catch-all addresses, about 45% actually delivered; the rest bounced, hit spam traps, or got silently dropped. Sending to unfiltered catch-alls without a confidence score is gambling with your domain reputation.
3. Stale-but-structurally-valid addresses
The email address format is correct, the domain is real, and it even responds to SMTP ping — but the person left that company. These are the hardest to catch. They become hard bounces on delivery. The only real defense is data freshness (enrichment sources with recent verification timestamps) or cross-referencing against job-change signals.
| Address Type | Do Verifiers Catch It? | How to Filter |
|---|---|---|
| Hard-invalid (deleted/nonexistent) | Yes — NeverBounce, ZeroBounce, Millionverifier, Hunter.io | Standard verification pass |
| Catch-all | Partial — flagged as "risky/unknown" | Remove or score; NeverBounce gives accept/reject confidence |
| Role address (info@, admin@, sales@) | Yes — flagged in all major tools | Filter by role flag in verification output |
| Stale (person left, address still pings valid) | No — SMTP shows valid | Re-verify within 30 days; use enrichment with timestamps |
A Data-First Deliverability Workflow
After the two blown domains, I rebuilt my entire process. This is what I ran against a 6,200-contact list that produced a 1.3% bounce rate on the first send.
Step 1: Verify at import, not just before send
Don't wait until you're ready to launch. Verify immediately when you pull contacts, and tag every address with its verified timestamp. NeverBounce and ZeroBounce both return results with verification dates. When you revisit a list more than 60 days later, re-verify. Decayed validity is the most common cause of creeping bounce rates on campaigns that were "clean last month."
Step 2: Segment and handle catch-alls separately
Don't delete all catch-all addresses — that throws away real contacts. Delete the ones you can't score. ZeroBounce's catch-all scoring assigns a send/no-send recommendation. NeverBounce flags them as "accept-all" and lets you decide. My rule: send to catch-alls only if the domain is a company I've confirmed exists and cap catch-all volume at 15% of any single campaign.
Step 3: Build a waterfall enrichment, not single-source dependency
Single-source lists — Apollo-only or Hunter.io-only — have verification blind spots. Accounts that pass one provider's internal check but have since gone stale don't get flagged. I use Clay to run a waterfall: Apollo → Hunter.io → Snov.io → NeverBounce final pass. Any address that comes back clean from three independent verification layers has dramatically lower bounce probability. My bounce rates on waterfall-verified lists averaged 0.9–1.4% vs. 2.5–4.2% on single-source lists.
Step 4: Scrub role addresses before touching volume
Millionverifier flags role accounts in their output. So does NeverBounce. Remove them. Even when info@, support@, admin@ addresses are technically deliverable, they either hit spam filters, land in shared inboxes where nobody replies, or generate spam complaints from people who never opted in.
The Real Cost Comparison: Warmup vs. Verification
This is the math the warmup vendor pricing pages skip.
| Tool / Approach | Monthly Cost | What It Fixes | What It Doesn't Fix |
|---|---|---|---|
| Lemwarm Essential | $29/inbox | New domain reputation | Bad data, high bounce rate |
| Mailreach | $25/inbox | Spam placement, reputation diagnostics | Stale lists, catch-alls |
| Instantly warmup (built-in) | Included in plans | Reputation for Instantly-hosted inboxes | Data quality |
| NeverBounce | ~$49 for 10k verifications | Hard bounces, role addresses, catch-all scoring | Stale-but-valid addresses |
| ZeroBounce | ~$50 for 10k verifications | Hard bounces + spam trap detection + catch-all scoring | Stale-but-valid addresses |
| Millionverifier | ~$27 for 10k verifications | Hard bounces, role addresses | Catch-all scoring is limited |
| Clay waterfall | $149–$800/month | Multi-source freshness + verification combined | Expensive for solo operators |
A 3-inbox outbound setup on Lemwarm runs $87/month. For that same $87, you can verify roughly 17,000 emails through NeverBounce — enough to clean a mid-size outbound list twice a month.
The warmup spend doesn't disappear from your stack entirely; you still want some warmup running on domains under 60 days old. But if you're spending $87/month on warmup and zero on verification, the priorities are inverted.
The warmup-tool playbook assumes the data is fine and the reputation is the variable. For most teams sending to scraped or enriched lists, that assumption is backwards.
What I Actually Use
NeverBounce is my first pass on most lists — it's fast, the API is clean, and the catch-all handling is the most actionable output on the market. For anything where I'm worried about spam traps (scraped or aged lists), ZeroBounce adds a detection layer NeverBounce doesn't cover as well.
For high-volume cleaning on a budget, Millionverifier is hard to beat at roughly $1 per thousand emails. The catch-all scoring is less sophisticated, but for fast bulk passes where I just need to eliminate obvious invalids, it gets the job done.
For enrichment and waterfall verification, Clay is the most capable option — but it's priced for teams. Hunter.io's verify endpoint handles real-time lookup well for smaller setups.
I still run Lemwarm on domains under 60 days old. The insurance is worth it on new infrastructure. But I stopped treating it as a deliverability fix for established domains — that's the wrong tool for the root cause.
For Twitter and Facebook profile-based research where I'm building contact lists from social data, Ziwa has been faster than hitting People Data Labs' direct API for those specific lookups — though a verification pass through NeverBounce or ZeroBounce is still mandatory regardless of where the emails came from.
The pattern that actually moved my bounce rates from 18% to under 1.5%: verify at import, re-verify after 60 days, waterfall enrich where budget allows, aggressively remove unscored catch-alls, and stop expecting warmup tools to solve problems they were never designed to touch.
Top comments (0)