14. Case

§One case, told straight through. A multi-brand hospitality group with more than twenty brands, a product lead, and a first version already defined by the strategy work: the guest builds her own profile and sees a tier, four weeks, act one. Business design ran in the week after strategy closed, in three working sessions, with Strategyzer's canvases named as such.

§ 14.1Desirability#

§Four value proposition canvases, one per segment the sprint's synthesis had produced. Frequent customers: to be known, across brands they did not know were related. Occasional customers: a table on a full night, anywhere in the group. Staff: to know a regular before reaching her. Managers: knowledge of regulars that did not leave with a server.

§The seams appeared when the four were on the wall. The frequent customer's gain required data the server could not capture during service. The manager's gain took something from the server, whose knowledge of regulars had been personal leverage. The occasional customer's gain required brands to send each other covers. Each seam got a line: who gains, who pays, whether the paying side had agreed. Two of three had not been asked.

§The three piles of evidence, read from the sprint's synthesis pages. Would pay: a table on a full night; turnover-proof knowledge of regulars. Would never pay: being greeted by name, loved by six of six and owed to them by any good restaurant. Gave freely: the profile itself, filled in gladly and unpaid by six of six, because they understood it would get them served better.

§ 14.2Feasibility#

§Act one could be built on almost any path, so the technical path recommendation spent its page on doors: the profile had to be readable by the operation later without migration; tiers had to compute from visits; nothing in act one could depend on the point-of-sale. The development partner recommended the path that kept all three open, and closed with "subject to the team's judgment."

§The operation was the sprint's under-served side, and business design said so rather than hoping. Every assumption about what a server would do was marked unverified. Act one asked nothing of the operation. Act two was scoped to one restaurant, read-only, with training, so the glance could be shown before anything relied on it. Eight one-on-one conversations with staff, nothing in hand, were scheduled before act two's rows could carry numbers.

§The sequence fell out of the dependencies. Profile and tier, depending on no one, first. The operation reading, depending on a glance said but not shown, second. Check-in, depending on a gesture that had failed six of six, third with its criterion. Payments, depending on check-in, behind it.

§ 14.3Viability#

§The canvas had one owned asset, the cross-brand profile the guest built herself, and several borrowed ones: reservations, point-of-sale, payments. The first version built the owned asset and touched none of the borrowed ones.

§The unit was a tier member. The first economics sheet, from before the field, assumed a tier had to give something away: points, a token economy, discounts. Its cost per member was the giveaway; its return was incremental visits the giveaway was assumed to produce; the arithmetic was tight. The second sheet, after the field, dropped the giveaway, because six of six had filled in their profiles for the recognition alone. The cost per member became a glance and a record. The return was the same incremental visits, now resting on a mechanism that had been observed.

§Monetization fronts survived on the two edges where a would-pay pile existed. For managers: knowledge of regulars across brands and staff, priced in a conversation dated for act two, the best-evidenced number on the list because turnover cost was in the data. For occasional customers: a table on a full night, dependent on the borrowed reservation platform and therefore dated for the partner conversation. Recognition: price zero, the mechanism, never charged for and never paid for.

§Who first: the overlap of desire, willingness to pay and act one's reach was empty. Act one was named as an asset-building act, the first paying customer was the group's own management, and the date on which the overlap became non-empty was the start of act two.

§ 14.4Stress test#

§Visit and spend data across brands, extracted before the session. Cross-brand frequency was lower than assumed, because most regulars were regulars of one brand; the tier thresholds were rewritten on the real distribution. The incremental-visit assumption was silent in the data, because nobody had ever been recognized across brands; it stayed the central hypothesis, with the pilot as its test. Turnover cost was supported, which made the managers' willingness to pay the strongest row.

§ 14.5The list#

§About thirty rows. Top three, dated this week: profile completion without reward, sixty percent, act one's first month. Incremental visits from recognition, the pilot quarter. The server's glance, half of shifts, the discovery loop and the pilot. Below them the rewritten tier thresholds, the managers' pricing conversation, the partner dependency, the support load, an inference line at zero marked to revisit, and the point economy, marked discarded with the quote that discarded it.

§Two pages. Read weekly. The only document from six weeks that was.

§ 14.6What comes out#

§The list, the sheets behind it, and a first version whose economics rest on a mechanism that was observed rather than invented. The sixth whitepaper takes the list and asks how to test its rows once the product is live.