Most "we drink our own champagne" posts show you the wins and hide the leak. This one shows you the leak.

For 30 days, dolv ran dolv's own go-to-market. Not a demo account. Our actual funnel, our actual North Star, our actual approvals inbox. Here's what we got: 24 GA4 sign-ups, a TOFU-to-MOFU funnel of 60% / 44% / 100%, and one obvious, embarrassing bottleneck sitting right in the middle. This is the real number, the real diagnosis, and the real fix.

why publish the leak

Because the leak is the point. Anyone can screenshot a green dashboard. What an operator actually wants to know is: when the numbers go sideways, does the tool tell you where and why — or does it just make more content?

dolv is an AI go-to-market operator. The pitch isn't "it writes copy." The pitch is "the loop closes": it drafts the work, you approve it, it ships, and then it measures what shipped against your funnel. Connect, execute, measure. If we can't show that loop working on our own funnel — including the ugly middle — the pitch is just words.

So: numbers first, spin never.

the 30-day scoreboard

Here's the funnel dolv reported on itself.

North Star: 24 sign-ups

Our North Star metric is GA4-verified sign-ups at app.dolv.work. Over the window, that number was 24. Not a vanity impression count, not "reach." Actual accounts created, pulled straight from analytics — the same source of truth we'd trust for a client.

Twenty-four is a small number. We're saying it out loud on purpose. When the base is small, every stage of the funnel is diagnostic. You can see exactly where people fall out.

The funnel: 60% TOFU / 44% MOFU / 100% BOFU

dolv scores every action against the classic three stages, so the funnel reads like this:

TOFU (top of funnel): 60%. Attention and first-touch. People are finding us and clicking. Sixty percent conversion at the top is healthy — the messaging is landing and the traffic is qualified enough to move.
MOFU (middle of funnel): 44%. Consideration. This is where someone who noticed us decides whether we're worth their time. 44% is the leak. Nearly six in ten interested people stall here.
BOFU (bottom of funnel): 100%. Decision. Everyone who reaches the bottom converts. One hundred percent. The people who make it to the end are certain.

Read those three numbers together and the diagnosis writes itself. The top works. The bottom works perfectly. The problem is a middle that's dropping more than half the people who were already interested.

That's not a traffic problem. Buying more TOFU would just pour more people into a bucket with a hole in it.

the worked example: what 44% actually costs

Let's do the arithmetic an operator would do, because the percentages hide the pain.

Say 100 qualified people hit the top. At 60% TOFU, 60 move into consideration. At 44% MOFU, only about 26 make it through the middle. At 100% BOFU, all 26 convert. Final yield: 26 out of 100.

Now watch what fixing the middle does versus fixing the top.

Fix TOFU from 60% to 75%: 75 into consideration → 44% → ~33 through the middle → 33 convert. You spent effort at the top and netted 33.
Fix MOFU from 44% to 60%: 60 into consideration → 60% → 36 through the middle → 36 convert. You fixed the leak and netted 36 — with the same top-of-funnel traffic.

The middle fix wins, and it's cheaper, because you're not paying to acquire new attention. You're keeping the attention you already earned. This is the entire reason dolv scores every action against a stage instead of just counting outputs. "We published 12 posts" tells you nothing. "MOFU is at 44% and it's your ceiling" tells you exactly what to do Monday morning.

how dolv found it (the loop, on itself)

None of this required a human staring at a spreadsheet. The loop ran it.

Connect. dolv reads from GA4 and the CRM, so the funnel is grounded in real events, not a guess about how things are going.

Execute. The AI Director drafts the actual work — posts, follow-ups, lead capture, CRM updates — grounded in our own knowledge base so it sounds like us, not like a generic content mill. Nothing goes out on autopilot. Every external send lands in an Approvals inbox first, and nothing ships until a human clicks approve. That guardrail is the whole reason we're comfortable letting agents do real work instead of just suggesting it.

Measure. Every approved action gets tagged to a funnel stage and checked against the North Star. When MOFU sagged, the OKR ladder — which auto-syncs from funnel data — flagged the middle as the constraint on its own. We didn't go hunting for the problem. The problem raised its hand.

That's the difference between a tool that makes things and an operator that closes the loop. One hands you more output. The other hands you a diagnosis.

the fix we're running

The MOFU leak told us the consideration stage wasn't doing its job: people were interested, then went quiet. So the plan for the next 30 days is deliberately narrow — no shotgun.

More middle, less top. We're shifting effort away from net-new attention and toward consideration content: proof, comparisons, and "here's exactly how it works" pieces (this post is one of them). We already have the traffic; we need to earn the second click.
Faster, grounded follow-up. The Director drafts consideration-stage touches tied to what each lead actually did, and those queue in the Approvals inbox for a human to send. The goal is to close the gap between "interested" and "gone cold."
One metric, watched. MOFU is the number. The OKR ladder is already tracking it, so we'll know inside the next cycle whether the middle moves — same way we'd report it to a client.

Our target is boring on purpose: move MOFU off 44% and hold TOFU and BOFU. That's it. One leak, one fix, measured.

where this does not fit

Honesty cuts both ways, so here's who should skip dolv.

If you have zero traffic, dolv can't fix a funnel that doesn't exist yet. It measures and executes against a funnel — it isn't a paid-ads agency that manufactures demand from nothing. Get some top-of-funnel first; then the loop has something to close.
If you want to fully automate sending with no human in the path, we're the wrong tool. Human-in-the-loop approval is a default, not a setting you can switch off. Some teams find that annoying. We think it's the only responsible way to let agents touch your audience.
If you're a 200-person org with a dedicated RevOps team and a HubSpot admin, you'll want more knobs than dolv exposes today. We're built for lean teams of 2 to 20 who need a GTM team they can't afford to hire — not an enterprise stack.

If you're in that lean middle — a founder, a small agency, a solo operator wearing the marketing hat you don't have time for — this is built for you.

the honest takeaway

Twenty-four sign-ups. A middle that leaks. A fix that's measured. That's not a triumphant case study; it's a working one. And that's exactly what you'd get: a GTM operator that shows you the leak, names the constraint, drafts the fix, waits for your approval, and then tells you whether it worked.

Don't just plan it. Dolv it.

If you want the same read on your own funnel, grab the free Funnel-Health Teardown: paste your TOFU/MOFU/BOFU numbers and get a graded scorecard back — the exact diagnosis we ran on ourselves, run on you. Find your leak before you spend another dollar on traffic. And when you're ready to close the loop, start free at app.dolv.work.

Watch The Loop Close: 30 Days of Dolv Marketing Dolv (the real numbers)