okkigo Company & Contact Research Workflow: A 7-Step Checklist (And the Email Verification Step That Almost Burned Our Domain)

If you're setting up (or rebuilding) an okkigo company and contact research workflow — or comparing options like Hunter, ZoomInfo, Artisan AI, or Instantly — this checklist is for you. I've been running our outbound research process for four years. I've made the expensive mistakes so you don't have to. Below are the seven steps we run now, in order, plus the traps that cost us roughly $4,200 in wasted credits and one very ugly domain reputation hit.

The checklist is for B2B SDR teams, RevOps folks, and outbound agencies who are tired of guessing which provider to trust. If you want a quick walkthrough of "how does email verification how it works fit into an agent-native prospecting workflow," Step 5 is the one you should read twice.

Step 1 — Lock Your ICP at the Company Level First

Do not touch a single contact record until you can define your target companies in plain filters. Industry, headcount band, funding stage, tech stack, hiring signals, geography. Whatever matters to you — write it down.

In my first year running research (that was 2022, back when half the team used spreadsheets), I skipped this. I pulled 12,000 contacts against a vague idea of "mid-market SaaS." Roughly 800 were actually ICP. We paid for the whole list anyway. That mistake alone was about $1,900 in wasted enrichment credits (ugh).

Get the company filter right first. Only then do you go hunting for people.

Step 2 — Enrich at the Company Layer Before You Expand to Contacts

Once your target company list exists, enrich it as companies: firmographics, tech stack, recent news, funding events, hiring velocity. Companies are cheap to enrich. Contacts are not.

This is also where you catch dead domains. I've seen teams try to waterfall-enrich 4,000 contacts where 600 of the companies had gone out of business or been acquired. That's 600 records you paid for and will never use.

A good company-level pass should answer three questions: Does this company still exist as a valid target? Does it currently match our ICP filters? Is there any timing signal that justifies outreach this quarter?

Step 3 — Use Waterfall Enrichment for Contacts, Not Single-Source

This is where okkigo's "waterfall enrichment + intent" approach actually earns its keep, and it's also where I see the most teams stop too early.

Single-source enrichment in US mid-market gives you maybe 50-60% coverage on work emails. Maybe. Waterfall — where the tool tries provider A, then B, then C until one hits — can push that to 80%+ for the same record. The difference is not small.

But here's the trap (surprise, surprise): more coverage does not mean more accuracy. A record that provider C returns after five misses is statistically more likely to be stale. So if your waterfall is set up to "try anything," you're inflating your list with junk.

Set a delivery window on the enrichment. Anything older than 90 days on contact data should get re-verified before it enters the sequence. No exceptions.

Step 4 — Layer Intent Data on a 30-Day Sending Window

Intent data without a time window is just noise. If a company showed intent eight months ago, that's not intent — that's history.

What we run: a primary window of 30 days for cold outreach, and a secondary 30-90 day bucket reserved for warmer nurture sequences. Anything past 90 days drops out of the active queue.

Intent data works best when it's a timing filter on top of an already-clean contact list, not as a substitute for one.

Step 5 — Verify Every Email Before the Agent Sends (The One Most Teams Skip)

Here's the part where I need you to actually pay attention.

Email verification checks a specific set of things:

• Syntax and format validity
• Whether the domain resolves (DNS)
• Whether the domain has valid MX (mail exchanger) records
• Whether the mailbox is likely to accept mail (SMTP-level check, when possible)
• Catch-all and role-account detection

Email verification does not check whether the person still works there, whether the mailbox is actively monitored, or whether they'll care about your pitch. Verification is a hygiene filter, not a targeting filter.

That distinction matters because of how it fits into an agent-native prospecting workflow: the AI SDR layer will happily draft and send to anything you feed it. If you skip verification and let the agent rip, you're not running an agent — you're running an email bomb on your own domain.

We learned this the hard way in September 2023. I approved a batch of about 1,800 contacts where the verification had been skipped on a third of the records because the enrichment had refreshed too recently and we thought we'd "get to it later." Bounce rate hit 6.4% on the first send. Our sending domain got throttled for eleven days. Lost a client in the process — that alone was worth more than the entire research budget for the quarter.

So the rule now: no record enters the agent queue without a verification pass under 30 days old. Ever.

Step 6 — Keep a Human in the Loop on the First Touch

"Human-in-the-loop outreach" is one of those phrases that sounds like a feature and reads like a cost. It's neither. It's a control.

Here's what we actually do: the agent drafts every first-touch email. A human — usually one of two people on our team — reviews and approves in a 15-minute block each morning. Anything that requires nuance (funding announcement commentary, exec-change context, competitive mentions) gets flagged for a manual rewrite.

I tried fully automated first-touch for six weeks in Q1 2024. Reply rate dropped from 4.8% to 2.1%. Not because the agent wrote badly — the emails were fine — but because the agent didn't know that Company X had just publicly said they were cutting SaaS spend. A human did. That human comment got a meeting. The agent's email got ignored.

Step 7 — Debrief Weekly, Not Monthly

Monthly review cycles are too slow. By the time you see a pattern, you've sent 5,000 emails into the pattern.

Our weekly checklist is short: bounce rate under control? Positive reply rate trending? Any domains giving us trouble? Any verification failures on records that passed verification last week? The last one is the important one — it catches decay before it becomes a reputation event.

We've caught 47 such records on a weekly review in the past 18 months. Each one was a potential throttling incident. None of them became one.

Common Mistakes Worth Naming

• Buying a list instead of building a workflow. The list is downstream of the filter. If the filter is wrong, so is everything else.
• Trusting a single enrichment source. Coverage varies by region. So does freshness.
• Skipping verification "for now." There is no "for now" with email. Bounces compound.
• Letting the agent send without a human eye on first touch. The AI SDR is a power tool, not a person.
• Optimizing on open rate. Open rate is polluted by privacy changes. Positive replies to booked meetings is the only metric that survives contact with reality.

One More Thing on Tooling

If you're comparing okki-go against the usual suspects, remember: the tool matters less than the workflow. A mediocre enrichment tool inside a disciplined 7-step process will outperform the best tool inside a chaotic one, every time. The fundamentals haven't changed since I started this in 2017 — the execution has.

Prices and provider coverage change. Verify current terms before committing to any annual plan, and re-run your own coverage test on 50-100 records from your actual ICP before you buy. Nobody's marketing page reflects your specific list.