How to Pick a B2B Email Database in 2026

Your email database is the single biggest variable in whether a cold campaign works. Get the data right and a 1% bounce rate is realistic. Get it wrong and you damage your sending domain in a week.

This guide covers four things. Real bounce rate thresholds. How fast B2B data decays. How catch-all servers wreck campaigns. What to ask any vendor before paying.

What bounce rate Google actually allows

Under 2% is Google's published danger threshold, the practical safe target is under 1.5%, and the ideal is under 1%.

Industry average is around 7.5%, which is already damaging most teams' deliverability silently. Above 5% sustained for 5 to 7 days, Gmail starts suppressing inbox placement across every send from that domain. Not just the bounced ones.

Recovery time depends on how fast you stop. Fix the source inside 48 hours and stop sending from the damaged domain, reputation typically recovers in 2 to 4 weeks. Keep sending on a damaged domain and recovery takes 2 to 3 months. At that point most teams just spin up a fresh domain.

How fast B2B data actually decays

B2B email addresses go invalid at 2 to 3 percent per month, which means a list verified 90 days ago is already 6 to 9 percent dead before you open it.

Job changes are the main cause. Companies rebrand. People leave. Domains migrate. Annualised, that is 22 to 30 percent decay. A list compiled six months ago and stored in a spreadsheet is significantly degraded by the time you use it.

How fast B2B email lists go bad

−26%

About a quarter of emails on a fresh B2B list stop working within a year

Line chart showing how fast a B2B email list goes bad. A fresh list starts at 100 percent of emails working and drops to about 74 percent twelve months later.
Category	Percent still valid
0 mo	100%
3 mo	92.7%
6 mo	85.9%
9 mo	79.6%
12 mo	73.8%

This is why static list buys keep failing. A verified database with continuous refresh is what stops decay from showing up in your bounce rate.

What catch-all servers do to your campaigns

Catch-all servers accept any address sent to a domain, then silently drop or bounce later. They look valid to every verification tool, then they ruin your bounce rate.

A catch-all configured at the SMTP level returns a green valid check on gibberish123@company.com and john@company.com equally. Most large enterprises run catch-all setups, which is the exact buyer cold email targets.

The result is a verifier that tells you the email is good, you send to it, and it hard-bounces. The worse outcome is when it silently disappears with no bounce notification, which still damages engagement signals to Gmail.

The safe approach is to flag catch-all addresses risky, send at half normal volume, watch reply rates closely, and pull anything with zero engagement after two touches. Better databases exclude unverifiable catch-alls from the verified pool entirely instead of charging you for them.

Role-based addresses are usually a trap

Filter info@, sales@, support@, contact@, and admin@ out of every cold list. They bounce more, generate more spam complaints, and produce no engagement signals.

Role-based addresses are either unmaintained or shared between multiple people. Shared inboxes tend to mark cold outreach as spam more aggressively than individual mailboxes. They also train Gmail's algorithm that your sends are low quality, because no one ever opens them.

What the new Gmail and Microsoft rules mean

Google began actively rejecting non-compliant senders in November 2025, and Microsoft followed in May 2025. This is rejection at the protocol level, not the spam folder.

Three requirements decide whether you get accepted at all. SPF, DKIM, and DMARC have to be set up. Spam complaint rate has to stay under 0.3 percent, which is three complaints per thousand sends. Bounce rate has to stay in the range above.

Fail any of them and sends come back as 550 rejections. Your domain reputation in Google Postmaster Tools degrades fast, and every email from that domain gets filtered, not just the rejected ones.

Apollo and ZoomInfo accuracy in real use

Real user data puts Apollo at 65 to 80 percent accuracy and ZoomInfo higher at the cost of an annual enterprise contract. Most teams pick on budget.

Apollo claims 91 percent email accuracy through a 7-step verification process. ZoomInfo claims 95 percent with under 5 percent bounce and uses 300+ human researchers. The accuracy gap is real. The cost gap is bigger. ZoomInfo requires annual enterprise contracts. Apollo is dramatically cheaper.

Both struggle outside the US. Apollo phone accuracy drops to 30 to 40 percent in some regions. One operator on Reddit framed the choice as "I had found Zoom to be inaccurate the same as Apollo is now, so save a few dollars."

What a modern B2B email database needs

Continuous verification, native waterfall enrichment, and direct CRM sync. Those three things separate a usable database from a list you regret buying.

Continuous verification means the database re-checks email validity weekly. Static lists and quarterly refreshes lose to 2 to 3 percent monthly decay.

Native waterfall enrichment means the vendor queries multiple sources sequentially without you wiring API keys to Apollo, Hunter, ZoomInfo, and Dropcontact yourself. Match rates compound. Single-vendor coverage gaps disappear.

Direct CRM sync means enriched contacts land in HubSpot with owner assignment and custom field mapping. CSV exports lose data and create version drift between your prospecting tool and your pipeline of record.

Tech-stack filtering matters if your buyer is defined by their stack. Most legacy databases handle industry and headcount, but very few filter by what software companies actually run. Wappalyzer-powered databases can target "Shopify Plus stores running Klaviyo," which is impossible in Apollo or ZoomInfo.

How to evaluate any database before paying

Five questions answer most of what matters.

What is the actual hard bounce rate the vendor guarantees? That number is different from claimed verification accuracy, and both exist for a reason.

How do they handle catch-all servers? "We mark them risky and charge you" is a different answer than "we exclude them from the verified pool."

How often do they re-verify existing contacts? Anything less frequent than weekly is losing to decay.

What is the refund or credit policy on hard bounces? Vendors that stand behind their accuracy refund the bad data.

Where does the data come from? Vendors that source primarily through scraping have different accuracy profiles than vendors with first-party contributor networks.

How Tables.so handles this

Tables.so runs on a database of 300M+ professional profiles, 143M+ verified emails, and 125M+ verified mobile numbers. Email accuracy is 98 percent through a 5-step verification pipeline that handles syntax, SMTP, catch-all detection, spam-trap removal, and honeypot filtering. The whole database refreshes every 7 days, versus the 6-week industry average.

Mobile numbers come with a 30 percent pickup rate. The native waterfall runs across multiple sources internally with no API keys to manage. AI Search builds lists from natural-language ICP descriptions instead of complex filter UIs. AI Enrichments run research agents on each prospect for context like recent funding, hiring activity, or tech stack signals.

Wappalyzer integration filters companies by the technology they run, including Shopify, WooCommerce, specific SaaS, analytics, and payment processors. Most contact databases cannot match this. Native HubSpot integration with owner assignment, custom field mapping, and static fields means contacts land where your team works without CSV exports.

Next steps

Book a demo for a 15-minute walkthrough, or start with 100 free credits, no credit card required.