We Burned Our Sending Domain in 9 Days. Here's the Checklist I Wish I'd Run First.

It Started With a Number on a Dashboard

February 14, 2024. I'm sitting across from our CEO, and he's pointing at the Q2 pipeline dashboard. We're at 39% of target with six weeks left. "You've got 28 days to fix this," he says. "I'm not hiring anyone. Use tools."

I run RevOps at a mid-market B2B SaaS company. Seven years in, mostly cleaning up other people's messes — list hygiene, deliverability rescues, CRM migrations that went sideways. I've rebuilt sending infrastructure for three different companies after they got blacklisted.

I knew better. And still, I walked us straight into a wall.

How We Actually Got Here

We spent four days on evaluation. Landed on okkigo mainly because of the positioning — agent-native prospecting, waterfall enrichment, intent data baked in. Pricing was competitive. Docs looked solid.

The tool wasn't the problem. I was.

Deployment took about a week. Wednesday night we imported a 38,000-contact file. Picked our target industries. Wrote what I genuinely thought was a decent first-touch sequence. First send went out Friday at 9am.

Monday morning, numbers looked fantastic. A 56% open rate, 4.1% reply rate, fourteen meetings on the calendar.

Then Tuesday happened.

The Week Everything Fell Apart

Tuesday open rate: 22%. Wednesday: 11%. Wednesday afternoon, one of my SDRs DMs me asking why his Microsoft 365 account got flagged.

By Thursday, our primary sending domain was on two major blacklists.

Final numbers from that week: a 16.8% hard bounce rate (anything over 2% is a screaming alarm), 412 honeypot hits, and a complaint rate that crested at 0.9%. Per Google's bulk sender guidelines (effective February 1, 2024), complaint rates above 0.3% in Postmaster Tools trigger delivery problems. We were at three times that.

We lost roughly three years of sending history on that domain. Every lifecycle sequence, every demo-confirmation email, every onboarding drip — throttled or blocked.

My CEO didn't fire me. The conversation we had in the conference room instead? That one changed how I think about outbound infrastructure permanently.

What We Actually Did Wrong

I spent about forty hours afterward trying to figure out which step broke. It wasn't the copy. Wasn't send timing. Wasn't sequence design.

It was the data layer, and how much faith I'd placed in a checkbox called "verified."

Here's the embarrassing part. We did run our list through email validation. Syntax check, MX record check, the basics. I called it done.

What we didn't do:

• Real-time API email validation at send time — which would have caught roughly 4,100 addresses that had gone stale between our file's export date and our first send
• Proper lead enrichment to distinguish a decision-maker's inbox from a role account or an abandoned shared mailbox
• Domain warm-up. We went from zero to 4,200 sends per day inside 72 hours.

And the mistake that stings most: to run "segment tests," I configured okkigo to push eight overlapping campaigns through the same domain. To almost any receiving server, that reads like exactly what it was — spam-shaped behavior.

But the real root cause wasn't a config toggle. It was me assuming that because the tool could automate a step, I could skip the discipline I'd built from years of running this stuff manually.

Tools automate steps. They don't hand you judgment. Skip the judgment, and it comes back to you — usually as a blacklist notification at 6am.

The Rebuild (And the Checklist That Came Out of It)

Getting a new sending domain back to healthy deliverability took about five weeks. Getting pipeline back to normal took about three months.

During the rebuild, I started writing everything down. That document became the checklist my team still runs before any outbound campaign.

1. Re-validate anything older than 24 hours using an API-based check. Static CSVs rot fast. An address that validated in November can be a trap by February.
2. Enrich before you segment. If you don't know whether an address belongs to someone with budget authority or a defunct shared inbox, you're marketing to noise.
3. Warm new domains for at least three weeks. Start at 50/day. Double. Watch. Repeat.
4. One campaign per domain. Split on segments, not on domains. Overlapping sends from a single domain is the fastest way to tank reputation.
5. Three touches per contact, max. Silence is silence. Adding a fourth and fifth sequence doesn't earn replies — it earns complaints.

What Changed When We Reconfigured okkigo

When we rebuilt — first on a side domain, then on the main one — I realized how much of okkigo's configuration surface I'd ignored the first time around. The platform was fine. My understanding of how outbound infrastructure should be configured was the missing piece.

Three settings made the biggest difference:

Validation cadence. Instead of validating once at import, we now re-validate at campaign-push and expire any address older than 30 days. That one change eliminated over 90% of our subsequent hard bounces.

Intent plus standard enrichment layering. okkigo supports combining intent signals with standard firmographic enrichment. We'd only been using the basic import. Once we layered intent in, reply rates moved from 1.2% to 4.8% — because people were replying to topics that were actually relevant to them.

Human-in-the-loop approval. We stopped letting sequences go out fully automated. Now every send batch gets a fifteen-minute review from an SDR before it queues. Slower, yes. But the quality lift was real.

I have mixed feelings about that last one. Part of me wishes fully-automated SDR workflows lived up to the marketing claims. Another part — the part that lived through February 2024 — knows that a human pause is cheaper than a reputation rebuild.

What I'd Tell Anyone Evaluating Email Automation Right Now

If you're looking at an AI SDR or outbound automation platform — okkigo, or anyone else — here's the order I'd insist on:

Validation layer first. Enrichment second. Sending infrastructure third. Copy last. Most teams do this in reverse, which is why most outbound programs die before they get a chance to prove themselves.

And on the specific question of what RevOps teams should evaluate in email automation, these are my non-negotiables now:

• Does it validate at send time, not just at import? A platform that re-checks before every push saves you from most of the worst outcomes.
• Are data sources transparent, or a black box? You should at least know where your contacts' emails, phones, and firmographics came from.
• Does the API expose validation and quality signals? If your CRM can't filter on freshness and quality before data hits the system, you're flying blind.
• Is there a human approval surface? Every AI SDR vendor now claims full automation is the goal. My read is the opposite — the best configs blend agent speed with human review.

I can only speak from a B2B SaaS context — 20 to 50 seat sales orgs, mostly North America and EMEA. If you're running a 500-seat global team or you're in a heavily regulated vertical, the calculus probably shifts in ways I don't have first-hand experience with.

The Fundamentals Haven't Changed — The Execution Has

Outbound was a very different game five years ago. Volume mattered more than precision. Copy beat data. Lists were treated as inputs, not assets.

That's flipped. Data quality beats copy. Enrichment beats cadence. And the AI SDR layer on top of it all — including okkigo, including every competitor in the space — only works if the foundation underneath it is clean.

There's something satisfying about finally having this systematized. After the March blacklist, the April rebuild, and the May retro that took two full days, we came out the other side with a process that actually holds up under pressure. No more 6am worry sessions about whether the domain's going to get flagged overnight.

The tooling evolved. The fundamentals didn't. Validate. Enrich. Keep the human in. And don't let anyone — including yourself — talk you into skipping the order.