Case study — Software, hospitality, France

€1.6M selling AI to hotels that did not want AI.

AI Host Bot answers guest messages for hotels in France. The product worked. The selling did not. Founders were doing every demo themselves.

€1.6MContracted revenue added

12 monthsStart to run rate

58%Net margin held

Results at a glance

New B2B pipeline generated€0 before€2.3M

Close rate28%37%

Qualified leads, first 30 days150+

Primary acquisition channelPaid adsB2B outbound by day 90

Marketing budgetCut to 1/10 of previous spend

Avg. cost per qualified lead — cold email vs paid

Google Ads€390

Meta Ads€345

Cold email€36

Cold email brought in qualified buyers at roughly a tenth of paid CPA — and every euro not spent on ads goes straight to the net. That gap is how the marketing budget fell to 1/10 of previous spend while pipeline kept growing.

Part 01 — The challenge

A great product sold by one exhausted founder.

Every deal ran through the founder. He demoed, he closed, he onboarded. When he took a week off, the month died.

The market made it harder. French hoteliers are not early adopters, and 2025 taught them to distrust anything with AI in the name. Cold emails about artificial intelligence got deleted on sight.

The product was not the problem. Guests message at midnight. Staff answer at nine. That gap costs bookings, and AI Host Bot closed it.

But nobody sells a gap. They sell a machine, and the machine had a scary name.

Part 02 — The solution

Never lead with the technology.

We took the word AI out of the first message entirely and led with the number that a hotel manager already knows hurts.

Step 01

Sell the missed night, not the model

Opening line was about messages arriving after the desk closes and what those cost. No mention of AI until the second conversation.

Step 02

Write in the buyer’s language

Every sequence in French, written the way a hotel manager speaks, not translated from English. In this market that is the difference between a reply and a delete.

Step 03

Target the group, not the property

One independent hotel is one contract. A group is twenty. We went at groups and management companies where a single yes covers every property.

Step 04

Prove it on one property first

A pilot on a single site, with the numbers reported back. Then the rollout. Nobody signs twenty properties on a promise.

Part 03 — The outcome

The founder stopped being the sales team.

€1.6M contracted in 12 months. Net margin 58%, which is what software looks like once acquisition stops running through one person’s calendar.

210 properties live. Most arrived in groups of ten or more, from 14 parent accounts.

The founder now takes the calls worth taking. The pipeline fills without him.

Before you decide

If you are thinking…

“We sell software, not physical product. This will not apply.”

The channel is the same. A defined buyer, an offer built around their problem, and infrastructure that lands in the inbox. AI Host Bot is software and it worked in twelve months.

“Our market is conservative and will not buy AI.”

Then do not sell AI. Sell the missed booking. Conservative markets buy outcomes faster than anyone, because they are not distracted by the technology.

“Our founder is the only one who can sell it.”

That was true here too. It stops being true the moment the pitch is written down, tested and handed to a sequence. The founder is a bottleneck, not a feature.