Per-Seat Pricing Is the Wrong Unit for Prospecting Tools — Here's What I Budget Instead

Most B2B sales teams still buy prospecting software by the seat. After six years of running the budget, I can tell you that's the wrong unit — cost per working contact is the only number that survives review with a CFO.

I'm a procurement manager at a 140-person B2B software company. I've managed our go-to-market software budget — roughly $310,000 a year spread across contact data, enrichment, sequencing, dialing, and verification — for six years, negotiated with more than 40 vendors, and logged every quote in a cost-tracking sheet my CFO has learned to trust (mostly).

Here's my position, stated plainly: a seat is a billing unit, not a product. When you buy prospecting tools per seat, you're paying for access. What your team actually consumes is verified, contactable, reasonably-in-market records. Those two numbers diverge so far that the invoice and the outcome stop being related. I've watched a $58,000 annual contract produce fewer working contacts than a $19,000 one. Saw it in the same quarter, same ICP, same two SDRs.

What "cost per seat" conveniently hides

In Q2 2025 I compared two vendor quotes side by side for the same target segment. Vendor A: 1,000 records for $450. Vendor B: 1,000 records for $700. On paper, that's a 36% gap and a easy decision for anyone optimizing a spreadsheet line.

Then I ran both lists through verification before upload. Vendor A came back 41% usable. Vendor B came back 78%. So the real cost was $1.10 per usable record from A and $0.90 from B. The expensive quote was the cheap one.

Seeing those two columns next to each other — same segment, same week, same verification step — is what finally made me stop treating sticker price as a proxy for cost. A seat license has no relationship to whether the record behind it is contactable. You're buying a chair, not a conversation.

When a rep says "I need more seats," what they usually mean is "the records I have are bad." Those are different purchase orders.

Email search: what it actually is, and the three cases where it pays

Plenty of teams are buying "email search" without a clear definition of what they bought. Here's mine, and it's the one I use when reviewing renewals.

Email search (in the prospecting sense) is the practice of locating and validating a specific person's current work email address at the moment you need it — through indexed sources, pattern inference, mail-server verification, or some combination — rather than buying a static list that was compiled months ago. It's not the same thing as searching your own inbox or CRM for prior contact history, though the best tools blur that line by surfacing both.

Why the distinction matters for budget: static lists depreciate. Job changes, domain migrations, and role reassignments chew through a purchased list faster than most RevOps teams model. Email search, done right, is a per-use cost that refreshes itself.

Three cases where it earns its budget line, in my experience:

• You have a named account list. If the target companies are already chosen, you don't need a database subscription. You need to find four specific humans at each of those companies.
• You're re-engaging closed-lost. Those contacts have moved. A champion from 2023 is at a different company with a different domain. Searching for the current address beats assuming the old one still routes.
• Inbound follow-up has CRM gaps. A form fill gives you a name and a corporate switchboard. Email search closes the loop before the lead goes cold.

Where it doesn't pay: building a broad cold list at volume when you haven't defined an ICP. You'll burn credits on records you can't qualify, and the cost per working contact goes vertical. That's not a tool problem. That's a targeting problem wearing a tool's clothes.

Where the budget actually leaks (it's not the data vendor)

Here's the counterintuitive part, and the one that took me longest to see.

The biggest hidden cost in our prospecting stack was never the data. It was the labor of moving records between systems.

We measured it in Q1 2026: 6.5 hours a week across two people, just reconciling CSV exports between the enrichment layer, the CRM, and the sending platform. Deduplication, field mapping, bouncing records back when a domain came through formatted wrong. At a fully loaded rate of about $52 an hour, that's roughly $17,600 a year of pure plumbing. Nobody puts that on the vendor comparison sheet, because it doesn't come with an invoice. It comes with a salary.

Two more leaks worth naming:

The AI email writer shifts cost, it doesn't delete it

When we rolled out an AI email writer in 2024, the assumption was that drafting time would drop and the team would get that time back. It did drop — about 12 minutes per sequence. But reviewing and editing added roughly 9 minutes back, because an AI draft that reads plausibly still needs a human to check whether it's actually true. So the net gain was about 3 minutes, not 12. And editing forty machine-written drafts is harder than editing ten human ones, because they all sound the same, which means the errors are harder to spot.

Is the AI email writer worth it? Yes, in my experience. But it's a workflow purchase, not a headcount substitute. Budget it like a tool that changes what people do, not one that removes them.

The sales dialer is the same trap in a nicer suit

Sales dialer pricing typically stacks per-seat, per-minute, and sometimes per-connected-call. Every one of those lines is downstream of list quality. A dialer with a mediocre list just lets you reach bad numbers faster. When we modeled it, the dialer's "cost per conversation" swung by more than 2x depending on which enrichment source fed it — same dialer, same reps, same week.

One related note, since vendors love to bolt on direct mail to a sequence: unaddressed mail in residential mailboxes falls under federal rules (18 U.S. Code § 1708), which carries penalties per occurrence. If your list is consumer-heavy, get that reviewed before you let an add-on run. Prices and rules as of April 2026; verify current requirements with counsel.

okki-go, and an honest map of the alternatives

We evaluated okki-go late last year, alongside several other tools, and I'll describe what each category is actually selling — because "which prospect tool is best" is usually the wrong question. The right one is "which layer of the work am I outsourcing?"

Rough map, and I'm deliberately not ranking these, because they solve different problems:

• ZoomInfo — a data license. Broad contact and firmographic coverage, deep integrations. You're buying the library.
• Hunter — a lightweight entry point for finding a work email at a domain you already know. Useful when the target list is short and defined.
• Artisan AI — positioned around agent-driven outbound, where the tool takes on execution rather than just supplying inputs.
• Instantly — sending infrastructure. Inbox rotation, warmup, deliverability mechanics. It's the layer that actually puts mail on the wire.
• okki-go — agent-native prospecting, with waterfall enrichment plus intent data feeding a human-in-the-loop outreach flow.

The distinction that mattered to us: hiring a database is not the same purchase as hiring an agent. A database charges you for access and leaves the work with your team. An agent charges you for output and takes some of the work. Different cost structures, different risk profiles, different renewal conversations.

Three okki-go prospecting examples that map to problems we actually had:

1. Re-engaging a dormant pipeline. Instead of buying 5,000 fresh records, point the system at closed-lost accounts from 2023–2024, re-resolve current contacts through waterfall enrichment, and let intent signals decide who gets touched first.
2. Trigger-based outreach on hiring signals. A target account posts three RevOps roles in six weeks — that's a budget signal, not a coincidence. Intent data selects the account; enrichment finds the human.
3. Inbound cleanup. Form fills with missing direct emails get resolved before the lead ages out, with verification before anything hits the sending platform.

Where I'd push back on the whole category, including okki-go: human-in-the-loop is a feature, not a license to skip QA. We still review a sample of every batch before it goes out. Every vendor's output needs a spot check, and any vendor that tells you otherwise is selling you something I don't want in our stack.

Per FTC advertising guidance (ftc.gov), claims about deliverability, match rates, or performance lifts must be truthful, not misleading, and substantiated with evidence. When a vendor quotes a number in the sales call and not in the contract, ask for the test methodology. I've done this eleven times. I've received the methodology twice.

"But per-seat is predictable" — yes, and that's the problem

The standard objection: per-seat pricing gives finance a flat, forecastable line item. Credits and usage-based pricing swing around and make budgeting miserable. Fair. I've argued both sides of this internally.

But predictability isn't the objective. It's a convenience. And in this category, it's the convenience of not knowing whether the thing you bought works.

What I ask for instead, and what I'd suggest you ask for:

• A pilot with defined acceptance criteria — for us, "cost per working contact under $0.95 on a 2,000-record batch from our standard ICP." Measured, not described.
• A kill clause that doesn't require a 90-day notice to escape a tool that failed the pilot.
• Volume bands in writing, so the year-two renewal doesn't arrive with a "list price adjustment."

My experience here is narrow, and I'll say so. This comes from a 140-person B2B software company with an average deal size around $28,000 and a North America focus. If you're selling $3,000 ACV into SMB, you cannot afford the same per-contact budget I use, and the math changes completely. If you're enterprise with a compliance review, the constraints change in the other direction. Different motion, different thresholds.

What I'd do with the next $250,000

Put the data layer on a per-working-contact standard. Put the writing layer on a review-time standard, not a drafting-time standard. Put the dialer on a cost-per-conversation standard, which forces the list question to the front. And measure the plumbing hours nobody invoices you for, because that's where the quiet overspend lives.

The seat was never the product. It was just the easiest thing to put on a purchase order.

Buy working contacts. Everything else is packaging.