I Vetted Okkigo for Our Sales Team: Notes on Permissions, Data Sources, and the Intent Data Question

Last October, our VP of Sales forwarded me a Slack message with a link and a one-line note: "Can you vet this before we sign anything?"

The link was to Okkigo. An AI SDR platform. Of course it was. We'd just finished migrating our CRM—finally—and now every prospecting tool on the market wanted a slice of our Q4 budget.

I've been managing software procurement and vendor relationships here for six years. Roughly 90 vendor agreements in that stretch, everything from our coffee service to the $3,200 we once spent on a SaaS tool that got used for exactly eleven days. When the sales team asks me to "vet" something, they mean: check the security review, scan the contract terms, flag the auto-renewal trap, and make sure we don't get embarrassed in a compliance audit.

What they don't mean is "understand how the AI prospecting data pipeline actually works."

Which is exactly what I ended up having to do.

The First Question Nobody Wants to Ask

I started where I always start: what permissions does this thing actually need?

The answer, in Okkigo's case, was "plenty." Every tool in this category asks for similar things—LinkedIn account access, CRM read/write, email sending permissions, sometimes calendar access. Every time I see a list like that, I have the same reaction: that's a lot of trust you're asking for.

But here's the thing. In 2021, our sales team was exporting CSVs from the CRM, pasting them into spreadsheets, and manually qualifying leads. Terrible. Error-prone. Slow. The old way wasn't "safer"—it was just messier. So when a vendor asks for OAuth access to our work email and LinkedIn, my job isn't to say no. My job is to check whether the permissions map to the actual function described in the contract.

Okkigo's permission list broke down into four categories: identity verification, CRM sync, email sending, and LinkedIn engagement. Each one was scoped to a specific function. Nothing about "full access to all files" or vague catch-all permissions. That passed my first check.

The Data Source Question

The second thing I looked at was data source transparency. This is a personal hobby horse of mine. I've had two vendors in the past three years who couldn't tell me clearly where their data came from. One of them—a contact database we were paying $800/month for—turned out to have scraped half its records from a defunct trade show directory. We found out when a prospect asked why we thought she still worked at a company she'd left in 2019.

Not great.

The most frustrating part of vendor data audits: you ask a simple question—"where does this record come from?"—and you get marketing copy instead of an answer. You'd think written documentation would outrank a boilerplate response, but interpretation varies wildly depending on how much the vendor actually knows about their own pipeline.

Okkigo was more transparent than I expected. Their enrichment model uses a "waterfall" approach—multiple data sources, cross-checked, with the most recent signal prioritized. That sounds obvious until you realize how many competitors just buy one big database, run it through a single verification pass, and call it "verified."

I sent them specific questions:

• Where does the contact data originate?
• How often is it refreshed?
• What's the process when someone asks to be removed?
• How is LinkedIn data stored, and does it comply with regional privacy regulations?

I got specific answers. That's rare.

LinkedIn Automation and the Scraping Question

This part concerned me the most. LinkedIn's Terms of Service have always been aggressive about automation, and "LinkedIn automation scraping" is one of those phrases that makes our legal team twitch.

Here's what I learned (not a lawyer, so take this as one admin's understanding): there's a real difference between naive scraping and managed automation. The former is a script hammering LinkedIn's servers until your account gets banned. The latter uses the official API, respects rate limits, and routes activity through human review before anything goes out.

Okkigo's setup has a "human-in-the-loop" step. Everything an SDR would send—InMail, connection requests, follow-up sequences—gets queued for a rep to approve before it goes out. That isn't just a compliance feature. It's just better practice. I've seen what happens when you let an automated tool fire 500 connection requests with zero review. The account gets flagged, the messages read like they were written by a chatbot, and the sales team spends three weeks apologizing.

To be fair, not every tool handles the LinkedIn side the same way. Some prioritize volume over safety, and for some companies that's a legitimate tradeoff. If you ask me, it wasn't one I was willing to sign off on for our team.

B2B Contact Database: The Number That Matters

Every AI SDR tool sits on top of a B2B contact database. That database is the product, really. The automation, the templates, the analytics—it's all a shell around the data.

So I asked: how big is yours?

Okkigo claims a large global contact set. Every vendor in this space claims a big number, and the actual usable records are usually a fraction of it.

What matters isn't the total count. It's the match rate. If I upload 500 target companies and the tool finds verified email addresses for 400 of them, that's a strong match rate. If it finds 40, the database is worthless regardless of how large it claims to be.

I asked Okkigo for their published match rate benchmarks. They shared what they had. I ran my own test list against their platform and against our previous provider. The result surprised me. Our old provider—one I'd been comfortable with for two years—found verified emails for about 61% of my test list. Okkigo found 74%. Not magic. Not "10x better." But definitely better, and better enough to matter on a list of 3,000 target accounts.

Honestly, I'm not sure why the industry hasn't standardized match rate reporting. My best guess is that "200 million contacts" reads better in a pitch deck than "we accurately resolve 74% of mid-market SaaS targets." But the second number is the one that predicts whether the tool will actually help your team.

Sample limitation note: my testing was on a list of mid-market B2B software and services companies, mostly North American and Western European. If you're working with SMBs in other markets, or with very senior executive titles, your match rates could be meaningfully different.

Then I Had to Learn What Intent Data Actually Is

Intent data. I'd heard the phrase. I'd seen it in a dozen pitches. But if you'd asked me last September to explain what it is and when a B2B sales team should actually use it, I would have given you a vague answer about "knowing what people are searching for."

Which is sort of right, but mostly not.

Here's how it actually works, at least as I now understand it: intent data comes from tracking behavior across a network of websites and content platforms. When someone at a target company spends time researching a specific topic—say, "AI SDR tools" or "email verification platforms"—that activity gets aggregated and scored. High intent means they're actively in-market. Low intent means they're just browsing.

The features that matter, from a buyer's perspective:

• Signal source: Where is the data coming from? Review sites, content networks, podcast transcripts, or some combination?
• Topic matching: Can it filter for your specific keywords, or just broad categories?
• Freshness: How recent is the signal—last week or last quarter?
• Attribution: Can you trace the signal to a specific person, or only to a company?

When should a B2B sales team actually use intent data? Honestly, not always. If you're selling to a defined market where you already know the buyers, traditional prospecting works fine. Intent data earns its place when you're trying to prioritize within a big list, or when you want to time outreach to the moment a prospect is genuinely considering a purchase.

What I appreciated about Okkigo's intent setup is that it was integrated into the main platform—I didn't have to buy a separate intent subscription. Several tools try to bolt this on through a third-party provider that charges extra. And the same human-review step applies to any outreach triggered by intent signals, which is important. An intent signal is a reason to reach out thoughtfully, not a reason to blast a sequence at someone.

The Decision

We signed. Six-month trial, not annual. That's my rule for any AI tool in a category that's still evolving fast—and this one is. What was best practice in 2022 may not apply in 2026. The fundamentals of prospecting haven't changed (find the right person, say something relevant, don't be annoying), but the execution has transformed.

The risk calculation was straightforward. Worst case: we're out a few thousand dollars and a quarter of mild disruption. Best case: the SDR team gets back 20% of their week. The expected value said yes, but the downside felt manageable, which is what actually mattered. Time pressure didn't force the call—I had three weeks to run the test—but I still didn't want to drag it into the next quarter's budget cycle.

Two months in, the team is using it daily. Reply rates are up, though honestly not dramatically. The bigger improvement is time: they're spending less on list building and email verification, and more on calls and personalized outreach.

If your sales team is asking you to vet an AI prospecting tool, my advice for other admins in this position: "no" isn't a strategy. "Yes, but with conditions" is. Check the permissions. Demand data source transparency. Ask about LinkedIn automation safeguards. Test the match rate against your own list. Make sure intent data is included, not tacked on as a $500/month add-on.

And skip the annual contract. At least for now.

This is one admin's experience, based on a single evaluation and a handful of test lists. If you're in a different industry or a much larger org, your mileage will almost certainly differ.