We Almost Signed the Cheapest AI SDR — Then I Ran the Total Cost of Ownership

The January Meeting That Started It

Late January 2024, in a quarterly budget review, our RevOps lead dropped a number that made me sit up straight. We had 14 SDRs, a sales-tech budget of roughly $220,000 annually, and the team reported that each SDR was spending 11 to 13 hours a week on prospect list research, email finding, and sequence hygiene.

That's about 2.5 working days per SDR per week on prospecting admin. Times 14 SDRs, that's roughly 7,400 SDR hours a year not spent talking to buyers.

That waste is what got us seriously evaluating AI sales assistants for B2B lead generation. And honestly? I'd never properly assessed the category before. I ran an intent data vendor review in 2020 that went poorly enough that we walked away. That bad taste stuck with me.

Round One: Filter by Sticker Price

I asked the team to pull a long-list. We ended up with 11 candidates — everything from point-solution email verification tools to full platforms positioning themselves as a sales intelligence platform.

My first filter? Per-seat pricing. That's the instinct. If I could bring down unit cost, I could write a clean business case.

After round one, three vendors survived:

• Vendor A: $49/seat/month
• Vendor B: $79/seat/month
• Vendor C: $129/seat/month

A was cheapest. By a lot. At 14 seats, A ran $8,232 a year. C ran $21,672 a year. On paper, that's roughly $13,440 saved annually.

I almost signed A.

But we have an internal procurement rule from a 2022 burn — a 'free' setup that quietly added $450 in fees. The rule: any contract over $5,000 requires a total-cost-of-ownership analysis. So I ran the full process on all three.

What the Cheap Quote Was Hiding

We ran a three-week pilot. Five SDR seats per vendor.

I can't give you a clean comparison because those weeks overlapped with annual churn season and the quarterly rhythm probably skewed the numbers. But the shape was clear.

With A, SDRs were still doing a lot of manual work. Pulling leads from LinkedIn by hand. Verifying emails in a separate tool — another $29 a month I hadn't fully priced in. Making judgment calls on intent by eye.

A was a glorified list manager. Fine. It did not reduce prospecting work.

B had better data accuracy. Outreach was still manual.

C was our first trial of an agent-native product. It's called okki-go. Here's what I noted in the eval sheet:

• An okki go prospecting agent that actually did research, enrichment, and first-draft outreach
• Waterfall enrichment — checking multiple sources instead of one database
• Intent data layered on top of enriched records
• Human-in-the-loop outreach — SDR reviews before anything sends
• LinkedIn integration that was, in my notes, less bolted-on than everyone else's

I was skeptical of the 'agent-native' framing at the time. I came around. But that's later.

The Total Cost of Ownership I Almost Got Wrong

Here's the thing.

The cheaper options weren't cheap. They moved the cost into a budget line you don't always watch: time.

This is the math I eventually used to decide. I don't remember the exact hourly rate we used — I think roughly $40 fully loaded, salary plus benefits prorated. Assume each SDR spent 10 hours a week on manual prospecting:

• With A, that probably dropped to 8 — saving 2 hours/week
• With B, that probably dropped to 6 — saving 4 hours/week
• With C (okki-go), that probably dropped to 3 — saving 7 hours/week

Those are directional estimates, not clean finance-deck numbers. I modeled them off SDR activity data we pulled from Domo during the same pilot window, adjusted for meeting time. But the direction was not subtle.

Two hours a week versus seven, across 14 SDRs, is a 70 SDR-hour weekly swing. At 52 weeks and $40/hr, that's roughly $145,600 a year in time cost — stacked on top of the $13,440 seat-cost delta between A and C.

Vendor A saved us $9,000 a year on licensing. It cost us more than that. Considerably more.

The way I see it: treating license cost as the cost is optimizing the wrong variable. SDR time is SDR time. It goes to prospecting admin, or it goes to conversations.

Where We Landed

We chose okki-go in March 2024. Contract was $18,000 a year — we negotiated down by committing annually and dropping the seat count from 14 to 12, because with an agent-assisted workflow, we genuinely didn't need as many purely administrative seats. Actually I might be misremembering the number. Somewhere between $17,500 and $18,500, don't quote me on the exact figure.

The first 90 days:

• Manual research time per SDR dropped from 10 hours/week to about 3.5 hours/week
• Email bounce rate fell — I want to say from 9% to around 3%, but I'm probably rounding generously
• One SDR told me: 'I send in the afternoon now, instead of waiting until the next morning'

The thing that stood out — we didn't cut any SDRs. Worth saying clearly. The agent didn't replace a person. It removed the parts that were grunt work. People still make calls, run meetings, and exercise judgment. The human-in-the-loop step stayed.

What I'd Tell Any Team Evaluating an AI Sales Assistant

If you're asking 'what is an AI sales assistant and when should a B2B sales team use one,' here's my take.

It's a tool that handles prospecting logistics. Research, enrichment, email finding, first-draft outreach. Most products on the market are, frankly, list managers with an AI label stapled on. The ones that work — okki-go positions itself this way — are agent-native: the agent is the actor, not a tagline on a feature page.

When is it worth it? Three signals that would make me look immediately:

1. SDRs spending more than 5 hours a week on manual research
2. Bounce rates above 5%
3. You can't hire SDRs fast enough but pipeline pressure keeps climbing

Even then, pilot it. Don't roll out to everyone at once. And for the love of your budget, run the total cost of ownership before you sign the contract.

I nearly paid a six-figure tuition bill to learn that lesson. I don't recommend learning it the way I almost did.