A founder asked me this on a call last week, almost word for word: "What ROI should I expect from AI sales automation?"
I told him the same thing I'm about to tell you. The question doesn't have a clean answer, because the ROI of AI sales automation depends almost entirely on what you're replacing — not on which platform you pick.
If you're replacing a sales rep, the math is one thing. If you're replacing an agency, it's a different thing. If you're stitching together five tools you bought separately, it's a third thing. The tool almost doesn't matter. Your baseline does.
Here's the actual math, by baseline.
Baseline 1: You'd otherwise hire a sales rep
The fully loaded cost of a US sales hire — salary, benefits, equipment, ramp, manager time — runs $60,000 to $120,000 per year for an entry-to-mid SDR. That's $5,000 to $10,000 per month before they ever close a deal. Then add 3 to 6 months of ramp before they're fully productive.
theKrew's Starter plan is $99 per month. Even compared to the cheapest possible new hire ($60K/yr loaded), the AI replaces ~50x the spend.
For the math to break even on a sales rep, that rep needs to generate at least $5K/month in net new revenue beyond what the AI would have produced. For most SMBs, the AI is generating activity — research, personalized outreach, follow-ups — at a cadence the human can't match. So the threshold for the human to justify themselves is high.
ROI in this baseline: somewhere between 30x and 100x, measured purely as cost-to-replace.
Baseline 2: You'd otherwise hire an agency
Most marketing agencies that work with SMBs charge $2,500 to $5,000 per month for a partial scope — cold email or social or SEO, rarely all three. A full-service agency runs $5,000 to $15,000 per month.
The AI is $99 per month. Even against the cheapest agency retainer, that's a 25x spread.
The catch: agencies bring strategic judgment and account management. AI doesn't. So the ROI question becomes — how much of the agency's hours were spent on strategy you actually used vs. execution you could have automated?
The honest answer for most SMBs I talk to: 80% of agency hours are execution work. Drafting emails. Posting social. Building landing pages. Setting up campaigns. The strategy part — the thing most agencies sell on — accounts for maybe 20% of the invoice. AI handles the 80% better than the agency does (no waiting days for revisions, no per-asset markup) and you handle the 20% yourself with help.
ROI in this baseline: 15x to 50x, measured as agency-spend-replaced.
Baseline 3: You'd otherwise stitch together a tool stack
The average small business marketing setup runs $300 to $800 per month — Jasper or Copy.ai for content ($50), Mailchimp or Instantly for email ($50-200), Buffer or Hootsuite for social ($30-100), HubSpot or Pipedrive for CRM ($50-200), an analytics dashboard ($50-150).
The AI replaces all of those for $99. That's a 3x to 8x cost spread.
But the real ROI in this baseline isn't the dollars saved — it's the time you stop spending as the integration layer between five tools that don't talk to each other. (I wrote about this last week; the stack itself is usually the bottleneck, not the individual tools.)
ROI in this baseline: 3x to 8x in cost, plus 5 to 15 hours per week back if you were actually doing the integration work yourself.
The revenue side most ROI conversations skip
All three baselines above measure cost-to-replace. The bigger ROI lever is usually on the revenue side, and it's the one founders forget to count.
If theKrew's outreach books one extra qualified meeting per month for your business, and your average closed deal is $5,000, and you close 1 out of every 5 meetings — that's $1,000 per month in net new revenue from one meeting. At $99 per month spend, that's a 10x return from a single meeting.
Most active clients I see are booking 3 to 8 meetings per month from the AI's outreach alone. Even at modest deal sizes, that's revenue an agency would charge $5,000+/month to deliver.
Research from Google Cloud puts AI sales automation ROI in a 1.7x to 10x range across early adopters, with the upper end held by teams that deploy with clear metrics. That tracks with what I see — the businesses getting 10x ROI aren't necessarily using better tools. They're the ones who measured what the tools were actually replacing and held them accountable to it.
How long until you see ROI signal
You should not expect to see clean ROI in the first 30 days.
The first month is the system learning your business — your ideal client, your voice, your geography, what works in your category. The first batch of outreach is the calibration round. The signal you should be reading isn't "did revenue go up." It's "are the right kind of replies coming in?"
Real ROI signal usually shows up at 60 to 90 days. Real confidence in the number takes 6 months. Anyone who promises faster than that is either lying or running a flat-fee outreach service that doesn't actually care about quality.
What ROI metrics actually matter
Stop tracking vanity metrics — open rate, impressions, follower growth. They're noise.
Track these three:
1. Cost per qualified meeting booked. AI spend ÷ meetings that turned into real conversations with the right person. 2. Revenue per dollar of AI spend. Closed-won revenue from AI-sourced opportunities ÷ what you paid the AI. 3. Time spent on marketing per week. If it's going down while pipeline is going up, the AI is doing what you bought it to do.
If you measure those three, you'll know your ROI in a month — not in vague "we feel like it's working" terms.
So what's your baseline?
Here's the question I'd actually start with before you sign up for any AI sales tool:
What were you going to spend on this otherwise? A sales hire? An agency retainer? A tool stack? Your own evenings?
Whatever your answer, that's the number AI sales automation ROI gets measured against. Not against zero, and not against an abstract "industry benchmark." Against your specific alternative — because that's the baseline that determines whether the math actually works for you.