Revlution
Client Results / ROAS AI
Adtech · Channel Validation · Cold Email

323,014 cold emails. 296,237 prospects. Zero paying customers. Here's why we're publishing it anyway.

ROAS AI hired us to find out whether cold email could work as an acquisition channel for a narrow, hard-to-reach ICP. We ran a disciplined 90-day test, got a conclusive no about two months in, and released the founder early instead of running out the clock. This is what an honest pilot looks like when the channel doesn't work.

01 / At a Glance

The numbers

323,014Emails sent
296,237Prospects contacted
43Copy variants tested
0Paying customers
$5,000Credited back, unasked
~60 daysInto a 90-day pilot before we called it
"A lot of times when you don't get results with an agency, you feel like you got burned. This wasn't that. They did the work."
Ryan Tabb, Founder, ROAS AI
02 / Who Is ROAS AI

A working product, a narrow ICP

ROAS AI plugs into a company's ad accounts and lifts return on ad spend, roughly a 50 percent uplift, priced pay-on-performance with a free trial. Ryan already had traction through other channels. The acquisition problem was getting in front of a narrow ICP: companies running real, active ad spend.

IndustryAdtech / ad performance optimization
Target ICP100+ active LinkedIn or Google ad campaigns (50+ at lower priority), manager level and above, North America and Europe, qualified on live ad counts and spend via webhooks
Engagement90-day cold email pilot, October through late January
03 / The Challenge

Two agencies, nearly 10,000 emails, zero meetings

Ryan had tried cold email twice before he called us, with no results. The zero wasn't even the worst part. After nearly 10,000 sends across two agencies, he still couldn't answer the only question that mattered: is cold email a viable channel for this business? He didn't need another promise. On the first sales call we told him what we tell everyone: we can't guarantee results, but we can guarantee that after 90 days he would know whether cold email was the right channel for ROAS AI. He picked us because we'd already done this for marketing agencies before.

"Two agencies, nearly 10,000 emails, zero meetings, with no results."
Ryan Tabb, on a previous agency
04 / The Test

Six campaigns, 43 variants, 90 days

From October to late January we ran this as a structured experiment so the result would hold up either way, and recorded 15 Loom walkthroughs explaining what we tested and why. We set success criteria with every client before we start: the efficiency a winning variant has to hit, with clean deliverability and enough volume for the result to mean something.

Week 1

Onboarding and setup

50-minute onboarding with the Revlution team to lock the ICP: manager level and above, North America and Europe, qualified on active ad counts and spend via webhooks.

Month 1

First bright spot

ROAS AI's first-ever demo from cold email, plus early positive replies. "Ryan, one month in," on the early signal: it looked promising enough to keep testing.

Oct to Late Jan

The full test

323,014 emails sent to 296,237 prospects across 6 campaigns and 43 copy variants. We also ran dedicated Google Ads and LinkedIn Ads campaigns and tried new angles, and rewrote all the copy for a final push. No variant reached the success bar we set at the start.

~Two months in

We said it out loud

Revlution put it plainly: let the campaigns limp along another month, collect the final retainer, and blame the ICP, or say the obvious thing. We said it out loud: cold email wasn't a viable channel for ROAS AI's business right now.

Day of release

Released early

Released from the 90-day commitment instead of collecting the final month. We offered to keep campaigns running a couple of extra days as a courtesy, and handed over a full accounting of everything we ran.

5 days later

Unprompted follow-up

Ryan messaged us unprompted on Slack: "A lot of times when you don't get results with an agency, you feel like you got burned. This wasn't that. They did the work."

05 / The Results

What the engagement produced

323,014Emails sent
296,237Prospects contacted
0.82%Reply rate
807Replies logged
0.64%Bounce rate (1,892 bounces)
10Positive replies
1Demo booked, ROAS AI's first ever from cold email
0Paying customers

One meeting and a handful of positive replies out of 323,014 sends were not enough to call cold email a primary channel for this ICP. The data said no. So we said no. Ryan walked away with more than a null result:

06 / What We Did When Things Went Wrong

Radical transparency is easy to claim when things go well

Here's what it looked like when they didn't.

1. We owned deliverabilityDeliverability is our job, not the client's. We credited back $5,000, unasked, covering roughly three weeks of deliverability problems that were on us to fix.
2. We owned our mistakeA LinkedIn automation bot we were testing alongside email went sideways. It kept sending messages after being asked to stop, until Ryan finally flagged it. You should see the warts before you hire us.
3. We kept working after the no60,000+ enriched leads delivered free, a full post-mortem document so Ryan knew everything we did in detail, and six warm introductions to agency clients so he could pitch ROAS AI directly.
"Wow thanks guys, that goes a long way for us."
Ryan Tabb, on the deliverability credit

Our internal standard for every pilot: our goal is that worst case, a client knows cold email doesn't work for them, not that Revlution didn't work for them.

07 / Is This a Fit?

Who this is actually for

This story sits on our site next to the success stories where cold email became a core growth engine. We're a fit if:

We're probably not a fit if you want low-volume, low-commitment tests, or you want guaranteed customers regardless of market reality, or you don't actually want to hear that a channel doesn't work. On a one-month feedback call, Ryan said: if that's worth paying for even when the answer is no, we're probably a good fit.

08 / In Their Words

What Ryan says

"We have a difficult ICP to reach, and now I actually understand how to approach our go-to-market differently because of what we learned. The Revlution team tested a ton of copy and subject line variations, handled deliverability issues I didn't even know we had, and helped me think through how to manage replies. They were responsive in Slack and clearly cared about making it work."
RT
Ryan TabbFounder, ROAS AI
"A lot of times when you don't get results with an agency, you feel like you got burned. This wasn't that. They did the work."
RT
Ryan TabbUnprompted Slack message, five days after the pilot ended
09 / Common Questions

Common questions about channel validation

Why share a case study where cold email produced zero customers? Because we're not a lead gen agency. We help clients find their message-market fit, and going in, that fit is an unknown. Nobody can promise you upfront that cold email will print cash with ROI in month one. Anybody who guarantees that is straight-up lying to you. For ROAS AI we contacted 296,237 prospects and sent 323,014 emails across 6 campaigns and 43 variants, and the honest answer was 0 paying customers from cold email. Everybody publishes the wins. You should get to see what a disciplined "doesn't work" looks like too.

How do I know you won't just keep me on retainer if it's not working? With ROAS AI, we had a 90-day commitment. About two months in, we told Ryan what the data said: cold email wasn't a good channel for his business right now. So we let him out early instead of collecting the final month. That's the behavior you're buying.

Did ROAS AI get anything out of this if they got zero customers? Yes. A conclusive answer that cold email wasn't their channel, at least right now. Insight into which hooks and angles generated positive replies, even if they didn't convert. A fully documented post-mortem of the campaigns. A 60,000+ enriched lead file for their own outbound. Six warm introductions to aligned agencies in our network.

How do you decide when to call a channel dead? We set success criteria with every client before we start: the efficiency a winning variant has to hit, with clean deliverability and enough volume for the result to mean something. For ROAS AI, we ran 323,014 sends across 43 variants, kept bounces low, and no variant ever sustainably reached that bar. When the data is that clear, we call it.

Want a real answer on cold email?

We won't promise results. We promise serious sending infrastructure, real testing across offers, angles, and subject lines at real volume, and what the data actually says, not a guess. Bring your ICP, TAM estimate, current acquisition channels, and customer value. We'll tell you whether a 90-day outbound test makes sense.

Book a Channel Validation Call