How I Built dolv’s First 100 Customers in Public: An AI GTM Operator’s Founding Launch Playbook (Including the FOUNDING50 Offer)

founder-led sales motion AI GTM operator

The first 100 customers didn't come from a funnel I optimized — they came from me, personally, in the DMs, for 112 straight days. But the reason I could sustain that while also shipping product was that an AI GTM operator did the parts of the founder-led sales motion that don't require a founder: the research, the enrichment, the drafting, the follow-up math. That distinction is the whole playbook. Founders should own the relationship; agents should own the retrieval and the remembering.

Concretely, my daily loop looked like this. Every morning at 07:00, dolv pulled new signups and site visitors into the CRM, scored them against my ICP (2–20 person teams, B2B SaaS, agencies), and enriched each one with company size, stack, and a signal — a hiring post, a funding note, a "we're drowning in tools" complaint. I approved the list. Then I got a queue of pre-written first-touch messages grounded in my own knowledge base, so they sounded like me and not like a template. I edited roughly one in three, approved the rest in a batch, and hit send. Every reply landed back in the same pipeline with a next step already drafted.

Here's the actual rhythm that produced the first 100:

1. Week 1–2: 40 manual touches a day, zero automation — just to learn the language of the problem.
2. Week 3: I fed those 200 conversations into the knowledge base so agent output matched the objections I was actually hearing.
3. Week 4–8: Scaled to 120 touches a day with human approval on every send, and a FOUNDING50 offer — lifetime founding pricing to the first 50 teams willing to give me a weekly call.
4. Week 9+: The first 50 referred the next 50, because the offer was structured around public feedback, not discounts.

The uncomfortable lesson in this first 100 customers launch playbook: the founder never leaves the room. The AI GTM operator just makes sure the room is full of the right people, with the right context, before you walk in.

building in public to first 100 customers

Building in public wasn't a marketing tactic for dolv — it was the only honest way to sell an AI GTM operator nobody had heard of. When you're a small team selling to small teams, you have no brand equity to borrow. So I turned the entire customer-acquisition process into the product's public surface area: every connector shipped, every failed campaign, every churn conversation became a post. The goal was never virality. It was compounding trust with the exact 2–20 person teams who were watching.

The mechanics were deliberately boring, and that's why they worked. Five times a week for the first 90 days I ran the same loop:

1. Ship one visible thing — a new integration, a Funnel Intelligence view, a lead-scoring tweak.
2. Post the raw artifact — screenshot or short clip, no polish, plus the number that moved.
3. Name the failure. "Our first broadcast got a 4% reply rate. Here's what we changed."
4. Invite a specific reader in. Not "sign up" — "if you're running demand gen solo, reply and I'll set it up with you."
5. Log every reply in the CRM, then publish the aggregate result a week later.

The post that produced the most customers was a teardown of my own onboarding. I published the funnel: 1,240 visitors, 61 signups, 9 activated, 3 paying. Then I showed the single onboarding step I deleted — and activation jumped to 19 in the next cohort. That one post drove roughly 40% of the first 100 customers. People don't buy features they can't picture; they buy evidence that someone is measuring the same things they are.

The FOUNDING50 offer came directly out of this. Readers kept asking for the setup they saw in the posts, so I packaged it: fifty founding seats, done-with-you onboarding, a direct line to me, in exchange for feedback and a public case study. That was

FOUNDING50 founding customer offer strategy

The FOUNDING50 offer wasn't a discount — it was a trade. I had 50 seats of capacity to onboard by hand, and 50 slots of patience from early users willing to tolerate rough edges. So I priced it as an exchange of value rather than a markdown: founding teams get locked-in pricing, direct founder access, and roadmap influence; I get feedback velocity, permission to publish their results, and one warm introduction each. That symmetry is what made the first 100 customers launch playbook work — nobody felt like a guinea pig.

Concretely, the offer had four parts. First, price lock: FOUNDING50 members pay a flat rate for life, even as the standard tier rises. Second, white-glove onboarding: a 45-minute session where I connect their stack (CRM, email, analytics) and configure their first campaign with them, live. Third, a monthly build slot where their feature request jumps the queue. Fourth, reciprocal asks — one 30-minute feedback call per month, a short testimonial at day 60, and one referral intro to a peer team.

The mechanics I used to run it:

1. Cap at 50 seats and display a live counter on the pricing page — scarcity was structural, not manufactured.
2. Require a short application ("What's your current GTM stack? What breaks first?") instead of open checkout — that filtered for teams who'd actually give feedback.
3. Publish each founding story publicly with the customer's permission, which doubled as social proof for seats 51–100.
4. Raise price at seat 26 and again at seat 41, so early movers saw their lock-in appreciating in real time.

The most useful detail: I never called it a beta. "Beta" signals unfinished; "founding" signals ownership. When seat 50 closed, I held the waitlist and opened a standard tier at three times the founding rate — which converted a surprising number of the fence-sitters simply because the reference price had moved.

If I were doing it again, I'd cap the cohort lower and start the public build log earlier. Fifty was right for onboarding quality; thirty would have been right for depth of relationship.

how to fix a broken B2B conversion funnel before scaling traffic

Most founders scaling a broken funnel are paying to accelerate a leak. Before you touch ad spend, measure the funnel stage by stage and find where the drop-off is abnormal — not where it's merely annoying.

Take a real pattern from the first 100 customers launch playbook: a lean B2B SaaS team was running 5,000 monthly TOFU sessions, converting 2% to demo requests (100), but only 12 of those became booked calls, and 3 closed. They assumed the problem was top-of-funnel volume and doubled traffic. Revenue barely moved, because the funnel was losing 88% of its intent at the MOFU-to-BOFU handoff — a form that asked for company size and budget before the buyer had any reason to trust them, plus a two-day reply lag.

Here's the sequence I use instead of scaling:

1. Instrument each stage separately. TOFU → MOFU → BOFU → closed, with a rate and a volume at every boundary. A single "conversion rate" number hides which stage is broken.
2. Compare each rate against a rough benchmark. 2–3% visitor-to-lead, 25–40% lead-to-meeting-booked, 20–30% meeting-to-opportunity. The stage furthest below benchmark is your bottleneck — fix that one first.
3. Fix the friction, not the copy. Long forms, no social proof at the decision point, no clear next step, slow human follow-up. Replace qualifying questions with a calendar link and let AI Sales & CRM score the lead after the fact.
4. Re-run for two weeks before adding traffic. If lead-to-booked moves from 12% to 30% on the same 100 leads, you've tripled pipeline with zero extra spend.

In the example above, shortening the form to one field and adding an instant booking link lifted lead-to-meeting from 12% to 28%. Only then did the team scale traffic — and every incremental session converted at the new, higher rate. Fix the funnel, then buy the volume.

LinkedIn DM sequence for founder-led B2B sales

When you are a two-person team selling an AI GTM operator to other two-person teams, LinkedIn is the highest-leverage channel you have — but only if it looks nothing like the automation everyone else is running. The version that carried me through the first 100 customers launch playbook was deliberately slow, manual, and built around the prospect's funnel problem instead of my feature list. No scheduling tool, no "quick question?" opener, and roughly twenty new conversations a day that I typed myself.

The sequence itself is four touches over about ten days:

1. Day 1 — Connection request. One line of context, zero pitch. "Saw your post about pipeline reporting — we're both building lean GTM. Worth connecting."
2. Day 3 — First message. Reference something specific they published, then ask a single diagnostic question. "You mentioned your team lives in spreadsheets. Is that still true for weekly pipeline reviews?"
3. Day 6 — Value drop. Send something useful with no ask: a teardown of their signup flow, three subject lines, a screenshot of what their funnel would look like in one dashboard.
4. Day 10 — Soft close with an exit. "If this isn't a priority right now, say so and I'll stop showing up in your inbox."

A concrete example: a four-person B2B SaaS founder had posted about copying numbers into a spreadsheet every Friday. I asked whether that was still how he reported, he replied within the hour, and the value drop was a one-screen funnel view of his own traffic. He booked a call the next day. Across the first 200 conversations the sequence returned about a 38% acceptance rate, a 22% reply rate, and roughly six booked calls per forty threads — enough to land the first handful of FOUNDING50 customers without a single paid impression, and every message human-approved before it left my inbox.

AI GTM operator founding launch playbook

The founding launch playbook I ran for dolv had less to do with a launch day and more to do with refusing to sell something I hadn't operated myself. Before a single FOUNDING50 seat went out, I pointed dolv's own AI agents at dolv: crawling the site, drafting content, scoring inbound, and pushing qualified leads into the CRM with human approval on every external send. The company became the first case study in its own funnel — which meant every bug, gap, and slow approval queue surfaced on my desk before it surfaced in front of a prospect.

The operating loop is deliberately boring and repeatable:

1. Connect — wired the read/write integrations so agents could read real signals (site analytics, reply threads, CRM fields) instead of guessing from vibes.
2. Execute — agents produced the weekly TOFU blog sections, MOFU comparison pages, and BOFU onboarding emails, all queued for one-click human approval.
3. Measure — Funnel Intelligence attributed each signup back to the specific asset and channel that produced it, so I could cut what didn't move.

Worked example: in week three, an agent flagged that visitors reading a "lean team, one operator, 25 tools" comparison page converted roughly 3x better than homepage traffic. I paused two planned posts and wrote five variants of that page instead. That single redirect produced 19 of the first 100 signups — nearly a fifth of the cohort — from content that took about four hours total to draft and approve.

The lesson that shaped the rest of the first 100 customers launch playbook: building in public isn't a marketing tactic bolted onto the product, it's the product demo. Every post showing an agent's actual output is a proof point no landing page can fake. When a four-person agency joined as a FOUNDING50 member, they weren't buying a promise —

##