-
The Question Nobody Agrees On
- Scenario A: You're a Solo Founder or Two-Person Team
- Scenario B: You're Running an SDR Team With Quotas
- Scenario C: You're an Outbound Agency Running 5+ Clients
-
The Scenario Most People Don't Expect—Where Scraping Shouldn't Be At the Front
-
How to Figure Out Which Scenario You're In
-
What I'd Actually Do Tomorrow Morning
The Question Nobody Agrees On
Every few months I get the same Slack message from a RevOps friend: "Hey, do you scrape LinkedIn for your agent workflow or not?"
And every time, I want to give a clean answer. I can't. Because the honest answer is: it depends on which of three workflows you're actually running.
I've made this mistake in both directions. In 2021 I scraped aggressively for a 4-person outbound team and burned a domain. In 2023 I refused to scrape anything for a 20-rep org and watched them fall behind on pipeline for two quarters. Both were wrong calls, for opposite reasons.
What I've landed on after roughly five years of running agent-native prospecting setups is that the scraping question isn't really "is it good or bad." It's "does it belong at the front of your workflow, the middle, or nowhere near it?"
Three scenarios below. Find yours.
Scenario A: You're a Solo Founder or Two-Person Team
Volume: under 500 enriched contacts per month. No dedicated SDR. You're doing this between product calls.
What fits here
A LinkedIn Sales Navigator subscription plus light scraping (a browser extension that dumps your saved-search results into a CSV) is genuinely fine at this stage. You're manually reviewing 200–400 rows. The signal-to-noise ratio is high because you actually know your ICP.
Where the okki-go prospecting agent fits in here is downstream. You don't need it crawling LinkedIn—you need it taking that CSV, verifying emails (okki go email verification, or any real verifier), and letting you send a first touch that doesn't feel like a template. Human-in-the-loop outreach is basically the whole point at this size.
In the first year I ran outbound this way, I made the classic volume-hungry mistake: set up automated scrapers for 12 hours a day. Burned the LinkedIn account in 6 weeks. Cost me roughly $800 in Sales Navigator restarts and a week of no pipeline.
What breaks
When you try to automate the scraping and the sending. The whole workflow collapses into "spray and pray," and it shows. LinkedIn's rate limits are not a suggestion.
Keep one side manual. Usually the scraping side.
Scenario B: You're Running an SDR Team With Quotas
Volume: 3,000–15,000 contacts/month across 4–15 reps. Manager in the loop. Real pipeline attribution.
This is where the answer flips.
What fits here
You don't want reps scraping. You want a structured ingestion layer—usually Sales Navigator saved searches piped through a controlled scraper, then enriched through a waterfall (LinkedIn data → email finder → verification → intent signal). The okki-go agent workflow is built around this: contact list in, verified rows out, rep touches drafts in between.
Why does this work at 10 reps but not at 2? Because you can afford a compliance layer. One person owns the scraping rules. Rate limits are enforced centrally. Everyone else just sees the cleaned contact list.
The uncomfortable truth: most SDR teams that get banned don't get banned because scraping is bad. They get banned because each rep was running their own scraper.
What breaks
Waterfall enrichment without intent data. You get 8,000 verified emails and no reason to email any of them. Reply rates crater.
I watched this happen in Q1 2024 with a client—9,500 rows, okki-go verified emails, 0.3% reply rate. Nothing wrong with the enrichment. Everything wrong with the targeting layer feeding it.
Scenario C: You're an Outbound Agency Running 5+ Clients
Volume: 20,000+ contacts/month, multi-tenant, different ICPs, contractual deliverability guarantees.
What fits here
Scraping stops being "a step." It becomes an owned service. You need:
- A dedicated scraping tier (rotating infrastructure, not shared vendor pools)
- Per-client enrichment pipelines so one bad source doesn't poison everyone
- okki-go (or equivalent) as the agent layer orchestrating enrichment + verification + agent-native send sequencing
- Real-time deliverability monitoring per domain, not just per client
This is where "agent-native" stops being marketing language and becomes a survival requirement. At this volume, a human can't babysit enrichment. The agent has to catch a bad source, quarantine it, and re-route.
What breaks
When the agency treats scraping as a fixed pipeline instead of a modular one. Two years ago a vendor I know lost a $60k/year client because one DataProvider X node went stale and nobody noticed for 11 days. 40,000 rows, most of them unverifiable. Client fired them the week after.
The Scenario Most People Don't Expect—Where Scraping Shouldn't Be At the Front
Here's the one that surprised me: well-funded Series A and Series B teams with strong product-led motion.
They assume more prospecting volume equals more pipeline. It usually doesn't. Their product already generates inbound intent signals—trial signups, docs page visits, seat-limited usage. Scraping LinkedIn as the primary source is often worse than plumbing internal intent data into the same okki-go agent workflow.
The "scraping first" mental model comes from a 2016–2019 era when LinkedIn was the only reliable way to find ICP contacts at scale.
That's changed. Intent data (Bombora, G2 buyer intent, first-party website signals) now gives you names and a reason. The contact list becomes a byproduct, not the starting point.
What most people don't realize: a 1,000-row list from intent signals typically outperforms a 10,000-row scraped list in the first two weeks of an outreach sequence. It's not close.
How to Figure Out Which Scenario You're In
Three quick checks:
- Who owns the scraping rules right now? If the answer is "each rep" or "nobody specifically," you're either Scenario A (fine, keep it small) or you have a Scenario B/C problem disguised as a Scenario A habit.
- Do you have a hard monthly contact volume number? Under 500 → A. 3k–15k → B. 20k+ with SLAs → C.
- Do you have internal intent signals? If yes, ask yourself why scraping is your starting point. It might need to be your verification fallback instead.
I'd rather give you this checklist than a one-size-fits-all "scraping is fine" stamp. Because the teams that get burned are rarely the ones who chose the wrong tool. They're the ones who copied someone else's workflow without realizing it was built for a different scale.
What I'd Actually Do Tomorrow Morning
If you're under 500/month: keep the scraper manual, keep okki-go doing verification + drafting, ship every day.
If you're running SDRs: centralize scraping, layer intent data before enrichment, let the agent handle sequencing but not sourcing.
If you're an agency: modularize every enrichment node, monitor per domain, and treat okki-go's agent layer as the traffic controller—not the pipeline itself.
And if you're PLG-funded with real intent signals—try the boring thing first. Pull names from the data you already have. Scrape only what you can't get elsewhere.
None of that is universal advice. It's just what I'd do in each of those three rooms. You're in one of them.
