Pain-qualified segmentation found message-market fit for a healthcare startup

By Artyom Jurkevich, Founder & CEO, RevSculpt

The client is an AI concierge for clinics. The product answers inbound patient calls, books appointments, and routes callers to the right person, taking load off the front desk. The stage is early but healthy: paying users, a working product, product-market fit found. The team is small: sales is essentially the founder and a first sales rep. They know their ICP, but they don’t yet know how to reach it systematically.

Their marketing relied almost entirely on industry trade shows. The question was simple: could this product be sold through a digital channel, and if so, to which clinic segments, and with what message? We set a goal that's realistic for this stage: in six weeks, validate the channel, find the segments that respond, and pin down message-market fit. Not “meetings at any cost,” but precise knowledge of which pain drives a clinic to look for a solution.

Results:

• CAC cut from ~$340 to ~$120 (roughly two-thirds)

• Working segments found and message-market fit confirmed in 6 weeks

• 25 meetings at a single trade show: a ~1,500-attendee list worked in a week

• 25 message hypotheses, 5 segments × 5 angles, tested across ~12,000 emails

• The client left with a clear ICP, verified pain points, and a tested segment-to-message pairing.

Why you can't test a hypothesis at a trade show, and why outreach is the better channel here

Trade shows are a good channel, and they happen often enough: quarterly, twice a year. But a show has a limitation rarely discussed: it's almost impossible to test hypotheses cleanly there. Too many uncontrolled variables, who walked up to the booth, in what mood, after what. The signal is murky: you can't tell whether the message worked or the person was simply in the mood to talk.

When you need to test how to sell, you have three paths, and two are weak:

• At a trade show, signal purity is low and many variables are outside your control.

• Running hypotheses through AI on large samples is fast, but still a simulation. It cannot show exactly how the live market will react.

• Outreach goes straight to the actual audience. The variables are controlled, the response is real, and the test cycle takes days, not quarters.

That's why outreach is the best instrument for this job. It's a controlled, clean test: one variable changes, the rest are fixed, and you see exactly what produced the answer.

What we were testing: which pain triggers a clinic

The product solves several pains at once, and early on it isn't obvious which one makes a clinic act. That was the subject of the test: which pain switches on the search for a solution.

We segmented not by company size but by the specific pain visible from the outside. The source: public patient reviews.

What the “pain orchestrator” is

It's a layer that holds a map of the product: each feature is tied to the specific pain it solves and to how that pain shows up in reviews. When the agent reads a clinic's Google reviews, the orchestrator does three things:

• Classify the pain using the patient’s own words: “couldn’t get through,” “never called back,” “hung up on me.”

• Map that pain to the use case the product solves.

• Choose the message angle that connects that pain to the right feature.

Each clinic gets a message based on its specific pain, the feature that solves it, and the situation it is already facing. Instead of a product pitch, it hears: “You’re losing roughly X% of inquiries at this point. We help close that gap.”

The list: Apollo's broad sample, refined to 60,000 live contacts

The target: medical facilities of 10-50 people in the US. The starting point was Apollo, but only the starting point: Apollo gives a very broad raw sample, tens of thousands of facilities with almost no context. It's a list that's useless until you segment and enrich it.

So on top of Apollo we:

• Added an industry-specific source for clinics, to avoid depending on one database.

• Enriched every record, filling in the clinic profile, role-level contacts, and the pain signal from reviews.

• Only then segmented.

The funnel started with about 40,000 facilities, two relevant contacts per facility, and roughly 60,000 valid contacts after cleanup. The decision here is distributed: the front-desk tech specialist feels the pain, the Chief Revenue Officer counts the lost revenue, and the final word sits with the clinic's leader. Each role got its own emphasis.

The message is segment × angle: 25 versions instead of one template

In this project the message isn't a text you write once. It's a matrix, and building it is our core work.

Segments

Five pain-based segments aren't an Apollo filter, they're the product of review analysis and enrichment. Building the right segments is harder than it looks: you have to figure out which pains are even visible from the outside and which of them are worth separating.

Angles define what the argument is actually built around

An angle is which pain you lead with and why it will land. The same product sells from different angles, and each angle speaks to a different buyer motive:

• Lost revenue. “You're losing money on missed calls,” the CRO's language.

• Overloaded front desk. “Your admins can't keep up with the volume,” the operations language.

• Patient experience. “Patients leave when they can't get through,” the language of a leader who thinks about reputation.

Those are three different angles on one product. Within each angle we write the actual message: a specific email for the “segment + angle” pairing. That's why 5 segments × 5 angles = 25 distinct emails: we're testing not “which wording is prettier” but which argument switches on a clinic's desire to solve the problem at all.

The matrix and fast kill/scale

All 25 live hypotheses run at once, ~1,000 contacts per variant for significance. Then a simple rule:

• If the reply rate is below the threshold, the segment is dropped..

• If the response is middling, the wording is reworked.

• If the response is strong, the segment is scaled.

The main result was the contrast between segments:

Across ~12,000 emails over six weeks, that's modest in absolute terms, and for an early stage it's normal: one or two quality conversations a week on a product like this is already an event. The value isn't in the email count but in knowing exactly which segment and which wording produce the response.

We don't drop trade shows, we squeeze them

Any decent rep prepares for a show. But by hand it runs into time. The window when an attendee will still answer and agree to meet closes about three days before the event: by three-to-one days out, almost no one replies, everyone's flying, packing, busy. So you have to reach them earlier, around a week before, and cover the whole list fast. By hand, you can't get through ~1,000 people in that window, you process half the list at best, and the other half never even learns you'll be there.

Outbound removes both limits:

• Speed and reach. In a short window the system covers the entire attendee list, not half of it.

• Channels. We find people not only in the show's app or platform but also via LinkedIn and email.

• High acceptance. A message like “I know you're heading to [show] - let's meet up” lands well: people go to shows precisely to talk. So acceptance is high, and the contact is warm before the meeting.

Result at a single show: 25 meetings arranged in advance from a ~1,500-attendee list worked in a week.

What the client walked away with in six weeks

CAC: cut from ~$340 to ~$120, roughly two-thirds.

Working segments: we know which clinics respond, which don't, and why.

Message-market fit: a working “segment → pain → message” pairing, validated on real replies rather than a guess.

ICP and pain: verified. We don't disclose the message or the pain here: that's what the client paid for.

Readiness to scale: the in-house team can rewrite its marketing around a validated pairing instead of a hypothesis.

Key lessons: outbound for a healthcare startup

At an early stage, you're buying answers, not meetings. The main result of six weeks is knowing which segment responds and which wording works. That removes a startup's most expensive risk, scaling the wrong message.

Segmenting by pain beats segmenting by firmographics. “Clinics of 10–50 people” is a size, not a segment. A segment is “a clinic that loses calls.” The first is visible in any database; the second is visible in reviews and the second predicts response.

The contrast between segments matters more than the average. 5 replies per 1,000 versus 0 is a map of the market.

A digital channel fixes the trade show, it doesn't replace it. The same signal logic, tied to an event, turns a trip into 25 prepared meetings.

When this approach doesn't apply

No product-market fit. Here the client came with a working product and users. If the product is still finding itself, outbound will only show faster that the offer is raw.

You need large volume right now. This is a validation engine, not a volume machine: in six weeks on a narrow ICP you get clarity, not hundreds of meetings.

The pain isn't visible from the outside. Segmentation worked because the pain reads in public reviews. With no surfaced signal, you need another way to extract it.

Book a Strategy Call

FAQ

How can a startup quickly find message-market fit through outbound? Test narrow and in parallel. We ran 5 segments × 5 angles (25 hypotheses at once, ~1,000 contacts per variant) and killed weak segments as soon as the data showed it. Over six weeks that gives you not a flow of meetings but an answer: which segment and which message work.

What does it mean to segment by pain instead of firmographics? Firmographics is size and industry; pain is a specific problem visible from the outside. We sliced clinics by what shows in their Google reviews, for example, losing inbound calls, and spoke to exactly that situation. Pain predicts response better than company size.

Can outbound lower customer acquisition cost? Yes, if you stop paying for irrelevant touches. By focusing the message on responding segments, we cut the client's CAC from ~$340 to ~$120 in six weeks.

Why bother with outbound if we already have trade shows? A show is a good channel for conversation but a poor one for testing hypotheses: signal purity is low. Outbound gives a controlled test between events and strengthens the shows themselves: for one of them we prepared 25 meetings from the attendee list in a week.

Want to validate a channel without burning budget?

Book a 30-minute strategy call. We'll look at your product and market and show what a signal-based test can deliver in the first six weeks.

Book a Strategy Call