An introduction to our Ideal Customer Profile — the one definition that product, sales, CS and marketing all work from.
We keep the ICP deliberately narrow. Saying no to the edges is a feature of the definition, not a limitation of it — it's what lets us go deep instead of spreading thin.
Product, sales, CS and marketing optimise for the same customer instead of four different ones.
Find and qualify the right leads, run a message that resonates, keep acquisition cost down.
A focused roadmap wins on fit rather than feature count.
Built from 1,360 deals analysed. Most importantly it gives us a working decision filter — something we can measure plans and priorities against, rather than a description we file away.
An owner-led tours, workshops, or Food & Beverage operator in DK, NO, SE, IT, or ES selling a scheduled, catalog-based experience, from Seed through Accelerate.
Five clauses. The rest of this deck unpacks one at a time — then we'll test them.
We group them by how a host earns their living, not by industry.
| Code | Category | Where the value sits | The guest asks |
|---|---|---|---|
| TOUR | Guided Tours | The journey, destination, or environment | "Where are we going?" |
| INSTR | Instructor-led Workshops | Doing or learning something, with a defined end point | "What will I learn?" |
| FB | Food & Beverage | The product is the anchor | "What will I taste?" |
If you can answer the question in the last column, you can usually place the host.

Growth comes from adding departures, not seats.
Biggest gap: OTA channel depth, and commission we mostly cannot see live.

The line that matters: value is delivered fully in one sitting. That's what separates a workshop from a yoga studio or a language course.
Biggest gap: the private and B2B flow — most operators still run it in email.

Biggest gap: everything after the ticket — bottle sales, clubs, and the B2B invoice.
Not what industry they are in — how the transaction is shaped. Four checks, and all four have to hold.
There is a time slot.
The same thing runs again.
A set of experiences, not one-off builds.
Any qualifying guest can book it.
Supporting signals: a live website, taking bookings online, operating for a year or more.
Recurring classes are the one people ask about most, because we do have real customers there. They stay supported — we are just not going out to win more until the billing work is scheduled. Memberships are off-roadmap with no ship date, so don't promise it's coming.
Anything outside these five is out — however good the fit looks on every other axis. This is the easiest clause to check and the one people most want to make an exception for.
Nordics are projected to reach roughly 69% of net-new potential by end of 2027. Italy and Spain run into 2030–31, which is why they are a longer game rather than a quick win.
Once the operating model holds, the second read is complexity. Fit does not fall off a cliff — it erodes as complexity grows.
A solo guide on Squarespace is low-complexity. A multi-location catering school on Business Central and Planday, with a buying committee, is high-complexity — at the same revenue. The full complexity model is still being built; until it is ready, we use business volume as the gate.
Business volume is a hint that complexity is worth a closer look, because a bigger operator usually has more moving parts. It is not the gate itself. There is no revenue floor in the product ICP, and buyer structure never disqualifies on its own.
Both sit far above anything we would once have called a ceiling. Both are customers we serve well.
Well past €1M, and a good fit — because the operational complexity is low and the product does what they need.
Same story. Large by volume, straightforward to run, squarely inside what we build for.
Had we drawn the ICP on size, we would have excluded both. Size tells you how much an operator books. Complexity tells you whether we can serve them.
There is one Understory ICP — the product ICP. On top of it sit functional overlays, one per product or motion. Every paid feature has a target audience somewhere inside the ICP, so the set of overlays grows over time.
Industry, operating model, complexity, market, product fit. Business Volume is deliberately excluded here.
A set of rules for one product within the ICP. Overlays narrow. They never expand. No overlay can bring in an operator the ICP excludes.
This is why the ICP has no revenue floor but Sales-Led does. So when someone asks "is this customer in scope for Pay?", the first question is whether they are in the ICP at all.
| Overlay | The gate | Why that gate |
|---|---|---|
| Sales-Led | Business Volume ≥ €25k, plus geography and spoken language | Protects CAC — below it, winning the customer doesn't pay back |
| Pay | Operational complexity, plus geography | Complexity proxies the payment agreement; geography decides currency and local methods |
| Meta AdsBloom | BV ≥ €50k · INSTR and FB only · AOV ≥ €40 · under 80–90% occupancy | Creates demand from a cold audience, so it needs headroom and order value. TOUR is effectively excluded — OTA dependency is structural |
| Google Ads | Checkout revenue ≥ €25k Branded, ≥ €40k Smart · 30 bookings in 90 days | The customer has to earn back a fixed monthly cost |
| Reviews | No floor — open to the full ICP | Usage-based pricing absorbs the range a fixed fee would have to exclude |
| Websites | A judgment call: who is going to build this? | Delivery capacity, not customer size, is the constraint |
| None yet | Conversion is unmeasured, and it swings break-even from 30 to 2,222 bookings a month |
Sales-Led and Google Ads Branded Search both sit at €25k. That's a coincidence. Sales-Led uses it to protect the cost of winning a customer; Google Ads uses it as the point where a customer earns back €189 a month.
They also measure different things — Sales-Led uses estimated total BV, Google Ads uses revenue observed through our checkout. Track them separately. If one moves, the other doesn't automatically follow.
Some criteria look like customer traits but are actually driven by our own product — consent rate, review collection rate, how much of a business has moved into our checkout.
If a customer fails on a real trait, they're disqualified. If one of our levers is underperforming, that's a roadmap item — not a reason to walk away from a customer we could serve.
Six operators. For each one, decide before the answer appears — and say which rule decided it, not just yes or no.
Click or press → to reveal each answer.
Decide, then reveal →
A textbook TOUR operator in every respect except one: Portugal is not among DK, NO, SE, IT, ES. Right business, wrong map — and this is the near-miss you'll see most often.
Decide, then reveal →
Two failures. Membership is access, not a booking. And a class you come back to each week isn't finished in one sitting — that's the exact line separating a workshop from a recurring class.
Worth remembering: memberships are off-roadmap with no planned date. This isn't "not yet".
Decide, then reveal →
A producer hosting tastings on-site is core FB — sell it. The wedding-venue layer is the sub-category we watch rather than build for: if unrelated venue hire becomes the main event, fit erodes.
Not every answer is binary. "Yes, and keep an eye on X" is a real outcome.
Decide, then reveal →
The booking requires a named individual before a time can be chosen. Our unit is a scheduled slot open to any qualifying customer. Everything else here looks fine, which is what makes it a good trap.
Decide, then reveal →
Enterprise on business volume — but that is not what decides it. The real problem is complexity: multiple entities, and legacy systems that would need integrating or migrating.
Note what did not decide it. Not the four stakeholders — buyer structure is context, never a disqualifier. And not the €1.4M: we serve customers well past that.
Decide, then reveal →
They clear the product ICP and the Sales-Led floor of €25k. But Bloom needs €50k in Business Volume, so this operator sits inside the ICP and outside that overlay.
"In the ICP" never means "qualifies for everything we sell." Check the overlay for the specific product.
Fail any one and it isn't our ICP, however good the rest looks. Then, separately: check the overlay for whichever product you're proposing.
| Open item | Where it stands |
|---|---|
| Complexity model | The real fit signal, but the framework for inferring it from public information is still being built. Business Volume stands in until it lands. |
| BV & segment model | Segments are assigned by hand today, from a rep's estimate. A data-driven model is in build. |
| Per-package minimum price | The H2 pricing project, expected end of Q3 or in Q4. It's what lets Seed stay in scope without a revenue floor. |
| HubSpot taxonomy | Industry values in HubSpot don't yet mirror the ICP doc, so the two can disagree. |
| Wineries | Deliberately paused until H1 2027. Italian food producers are in scope now. |
The ICP is a living document on a roughly weekly cadence. Where it says TBD, it genuinely means TBD.
Who to serve (ICP) — the canonical definition.
ICP profiles — firmographics, pains, capabilities per category.
ICP Overlays V1 — every overlay and its thresholds.
Segments × Value Props × Capabilities — the Alignment² map.
Lars & Sari — product: definitions, category profiles, who qualifies.
Christian — overlays and pricing.
Harboe — growth and lead sourcing.
Kjær — sales playbook and enablement.
Cecilie — customer success. Joakim — RevOps and ICP scoring.
If the ICP and a system disagree, the ICP document wins — and tell whoever owns that section.