What I thought the problem was
I run outbound operations for a B2B SaaS team. In April 2023, I bought 200,000 "verified" B2B contacts from a data vendor. The pitch was clean: 98% deliverability, real-time email verification, intent signals included. Total cost: $7,200.
I thought the problem was volume. We needed more contacts. More sequences. More at-bats. That's what every GTM automation pitch tells you, right?
Turns out the problem wasn't volume at all. And it cost us roughly $47,000 before I figured that out.
Quick context on who I am: I've been running outbound for B2B teams for about 8 years. I've personally made (and documented) 11 expensive mistakes, totaling somewhere around $48,000 in wasted budget. Now I maintain our team's pre-purchase checklist so nobody else has to learn this the hard way.
This is the story of mistake #7.
The real problem: we were evaluating vendors on the wrong axis
Here's what I didn't understand until I watched our bounce rate climb from 3% to 38% over six weeks.
Almost every RevOps team evaluates data providers on coverage — how many contacts, how many fields, how many filters. That's the wrong axis. Coverage is cheap. Accuracy is expensive. And "accuracy" means something different from what most vendors are selling you.
An email verification service tells you whether an address is technically valid. It doesn't tell you whether the person still works there, whether their inbox is actually monitored, or whether they're anywhere near a buying decision. That's a huge gap, and it's the gap most teams don't see until after they've signed the contract.
We bought 200,000 "verified" emails. About 40% of them bounced or hit catch-all domains. Another 22% were role-based addresses (info@, sales@, admin@) that our sequences couldn't personalize properly. So of the 200K we paid for, maybe 75,000 were actually usable for outbound. That's not 98% deliverability. That's 37%.
"The vendor wasn't lying exactly. Their verification check was real. It just checked a much narrower thing than 'will this email reach a real human who might buy.'"
And this is where the second layer of the problem shows up: we didn't catch it because our dashboard looked fine. Our sending tool reported emails as "delivered." Our CRM logged replies. It took a month of manual spot-checking to realize the domain reputation damage was already done.
The part that actually hurt: what bad data costs you
Everyone knows bad data is expensive in the abstract. Here's what it actually cost us.
Direct spend: $7,200 for contacts we couldn't use. Plus $2,800 on a deliverability consultant to help us recover our sending domain. Total: $10,000, gone.
SDR time: We tracked it. Three SDRs spent about 140 combined hours over six weeks writing personalized intros to contacts that never existed in a usable way. At roughly $35/hour fully loaded, that's around $4,900 in wasted labor. That's before counting the opportunity cost of what they could have been doing instead.
Pipeline: This is the one that stings. Our Q2 outbound pipeline came in about $32,000 short of forecast. Not all of that is traceable to the bad data — but the SDR team's calendar was jammed with dead-end follow-ups for most of April and May. I'd conservatively attribute 60% of the gap to it.
So: $10,000 + $4,900 + roughly $19,000 in lost pipeline = $33,900 in obvious damage. Add the softer stuff — my own time rebuilding trust with sales leadership, the general dip in SDR morale — and $47,000 is not a stretch.
And here's the part nobody warned me about: the deliverability damage outlived the campaign. Google and Microsoft don't forget. For three months after we stopped using that data, our legitimate transactional emails were landing in spam. That's a brand problem, not just a marketing problem.
What I wish we'd been told (and what I now tell other RevOps leads)
If you're evaluating data enrichment, email verification, or a GTM automation platform, the vendor's slide deck is not going to volunteer any of this. So here's the checklist I built afterward. It's not exhaustive, but it would have caught our specific failure.
- Ask for the actual definition of "verified." Real-time SMTP check? Meaningful only if you know what happens to catch-all and accept-all domains. Ask specifically: what's your catch-all policy, and what percentage of your database falls into that bucket?
- Test the data on your own list first. Send them 500 contacts you already know are good, mixed with 500 you know are dead. See what comes back.
- Ask about sourcing provenance. Where did this data actually come from? Scraped, purchased, contributed, or consent-based? This matters for GDPR-adjacent exposure, and it also correlates strongly with data quality.
- Check whether enrichment is waterfall or single-source. A single-source enrichment tool will always have coverage blind spots. Waterfall enrichment (pulling from multiple providers, filling gaps) performs better — but you have to ask.
- Place a test order before you sign an annual contract. A $200 test will tell you more than a two-hour demo.
The other thing I'd add: demand human-in-the-loop review as a feature, not an afterthought. Our best campaigns have always been the ones where a human read the contact list before it went into the sequence. AI SDR tools that skip that step just accelerate the damage.
Tools like okki-go exist in this space, and okki go's agent-native prospecting approach is interesting specifically because it tries to preserve that human checkpoint rather than replace it. But I'll be honest — I haven't run enough volume through okki-go yet to give you a real verdict. If you're buying anything in this category, run the five-point test above before you trust the pitch.
When this approach doesn't fit
I'll be straight with you: this checklist is built for B2B teams doing outbound at some scale. If you're a two-person team sending 50 emails a week, most of this is overkill. Buy the cheapest list you can find, manually verify the ones that matter, and move on. The overhead doesn't pay off at that scale.
And if you're running a massive enterprise ABM motion where every contact is hand-sourced, you probably don't need to buy data at all. Your problem is elsewhere.
But if you're somewhere in the middle — buying contacts by the tens of thousands and wondering why your reply rates keep dropping — the problem is almost never the copy. It's the list.
Trust me. I found out the expensive way.
For what it's worth, the FTC's guidance on advertising substantiation (ftc.gov/business-guidance/advertising-marketing) is a good reference point when a vendor's claims about accuracy start to feel slippery. Vendors can't legally guarantee outcomes they can't substantiate. That's worth remembering at the negotiation table.