Example evaluation of a sample idea (Mesa). Real Firmgrove output, sample data. Back to The idea
Mesa applies a proven revenue management concept from airlines and hotels to a segment that has never had access to it: the independent urban restaurant. The core insight is strong and the analogy is tight. Restaurant tables are perishable inventory, no-shows and off-peak troughs are structural and painful, and the existing workarounds are genuinely bad. The founder has thought carefully about the wedge customer, the pricing logic, and the detection problem, and has arrived at clever answers to each. The human-as-sensor framing for onboarding non-integrated restaurants is particularly good: it makes the hardest venues the easiest to start with, which is exactly the right instinct for a cold-start marketplace product. The main idea risks are the two-sided cold-start problem, TheFork platform dependency, GDPR friction on the diner opt-in pool, and the as-yet-unproven willingness of restaurants to pay real money rather than just endorse the concept. None of these are fatal to the idea; they are the right challenges to work through next.
The problem is stated with precision and the economic mechanism is tight. Empty covers are perishable, fixed costs do not fall on slow nights, and the existing workarounds are low-effort, unmeasured, and goodwill-burning. The founder can name the exact moment the pain is felt and why it is sharp rather than abstract.
The wedge customer is described with uncommon specificity: 40-80 covers, owner-operated, mid-market, dense Madrid neighbourhood, already taking online bookings, already improvising fixes. That last detail is particularly strong as a buying-signal indicator. The diner side of the marketplace is less fully articulated but the founder points to credible analogues.
The early SAM in Iberia is real but modest at 15-45M euros annually, which is appropriate for a focused wedge strategy. Western Europe extension could reach 10x that, but execution complexity scales with it. The global restaurant tech TAM is large but loosely estimated. The market is large enough to build a meaningful business if the model works; it is not a lottery-ticket size market.
The white space is genuine and well-located. No incumbent currently offers real-time, rule-based yield management for sub-50-cover independents. TheFork is the dominant reservation infrastructure in Southern Europe and has experimented with promotions but has not productised this. The gap between enterprise revenue management and the paper-notebook restaurateur is real and currently unoccupied.
Demand validation is the single most important open question for the idea. The founder correctly identifies that verbal interest is cheap and that retention on a paid pilot is the real signal. Neither side of the marketplace has been tested with real money or real behaviour yet. The founder has a clear and credible plan to test it cheaply, which is exactly the right posture.
Value creation is clearly on the restaurant side: recovered revenue from inventory that would otherwise expire at zero value. The pricing model (low base plus commission only on filled tables) aligns Mesa's revenue with the restaurant's outcome, which is structurally honest and sales-friendly. The ROI framing is clean: one filled four-top covers several months of subscription.
Two regulatory risks are material and not trivial. GDPR consent architecture for a location-based opt-in diner pool requires careful design from day one, not a retrofit. Spain's consumer protection rules around dynamic pricing in food service could constrain how offers are framed. These are manageable but need to be baked into the design early.
The motivation case for restaurants is strong and well-argued. The founder correctly distinguishes between verbal interest and real willingness to pay and points to the right validation mechanism. The diner motivation is plausible and supported by analogues but not yet tested in this specific form.
The founder demonstrates a clear and nuanced understanding of the restaurant operator's world: fixed cost structure, weekday trough dynamics, the emotional reality of an owner watching empty tables, and the failure modes of existing workarounds. The airline and hotel yield management analogy is applied with precision rather than as a loose metaphor. The human-as-sensor reframe for the detection problem shows genuine product thinking.
Mesa is a marketplace. Restaurants need a local diner pool large enough to fill tables before the tool is useful. Diners need enough live offers to make opting in worthwhile. In a new neighbourhood, both sides start at zero, and the chicken-and-egg dynamic can stall before either side sees value. This is the central structural risk of the idea.
How to address it: Design the pilot so the restaurant side does not depend on a cold diner pool to see early value. One approach: seed the diner pool manually before approaching restaurants, recruiting opted-in locals through community channels, neighbourhood apps, or personal network before a single restaurant signs. Alternatively, launch with a small cohort of restaurants simultaneously in one tight neighbourhood so the pool density is reached faster. The goal is to ensure the first restaurant sees at least one table filled before it has to decide whether to keep paying.
TheFork owns reservation infrastructure in Southern Europe and could restrict API access, build a competing promotion feature, or simply change its terms in ways that break the integration layer. A business that sits on top of TheFork's data is structurally exposed to TheFork's strategic decisions.
How to address it: The human-as-sensor architecture the founder has already designed is the right hedge: it means Mesa does not need TheFork to function and can onboard restaurants that never use TheFork. Treat the TheFork integration as a convenience layer for a segment of restaurants, not the foundation. Over time, direct POS integrations reduce dependency further. Avoid building any feature that only works if TheFork cooperates.
A location-based opt-in diner pool is a GDPR-sensitive data structure. Getting consent architecture wrong at launch creates legal exposure and could require a costly rebuild later. Separately, Spanish consumer protection rules may require that dynamic discount offers are clearly framed to avoid claims of misleading pricing, particularly if the 'original' price and the discounted price are both displayed.
How to address it: Engage a GDPR-specialist lawyer before the first diner opt-in, not after. Design the consent flow as a genuine feature, not fine print: make it clear, revocable, and local. On pricing display, model the offer framing on TooGoodToGo (which has navigated this in Spain) rather than inventing a new approach. Frame offers as 'tonight only' availability at a specific price, not as a discount off a published price, which may reduce the advertised-versus-charged-price exposure.
The ROI math is compelling and the founder argues it well, but no restaurant has yet paid real money and renewed. The gap between 'this makes obvious sense' and 'I will put my card on file' is real, particularly for owner-operators who are cautious about new monthly commitments and may revert to free hacks if the first month is slow.
How to address it: Run the paid pilot the founder has already described: charge from day one, even at a reduced rate, and treat month-two renewal as the only signal that matters. Make the commission structure so transparent that the owner can see in real time exactly what Mesa earned them. The owner watching a filled table walk in and knowing it came from Mesa is the strongest retention mechanism: make that moment as visible as possible.
A diner opting into a local offer pool is a low-friction commitment. Actually walking to a restaurant in 45 minutes in response to a push notification is a high-friction behaviour. Conversion from opt-in to seated diner may be lower than the TooGoodToGo analogy suggests, because TooGoodToGo involves planning a pickup while Mesa requires immediate spontaneous action.
How to address it: The hand-run pilot the founder has described is exactly the right test for this. Measure not just opt-ins but actual show-up rate. Test whether certain offer structures (shorter time windows, deeper discounts, specific times of day) drive meaningfully higher show-up. If conversion is low, the idea may need a different diner-side mechanic, such as a reservable slot rather than a pure walk-in offer.
That a sufficient number of nearby opted-in diners will actually show up within 45 minutes of receiving a push notification, at a rate high enough to make the tool reliably useful to restaurants
The entire restaurant-side value proposition collapses if the diner pool converts poorly. A restaurant that pushes three offers in a month and fills zero tables will not renew, regardless of how good the pricing model is. This assumption sits beneath everything else and has not been tested in this specific form. It is also the assumption that is hardest to reason about from first principles, because spontaneous 45-minute dining behaviour is genuinely different from the TooGoodToGo and off-peak TheFork analogues the founder points to.
Recruit 50-100 opted-in diners in one Malasaña street via a WhatsApp group or a local community channel. Partner with one or two restaurants informally. When a real table opens, manually send the offer as a WhatsApp message and measure how many diners respond and how many actually show up and are seated within the time window. Run this five to ten times over four to six weeks. A show-up rate above 15-20% on a pool of that size would validate the behaviour. A rate near zero would tell you the diner mechanic needs to change before building anything else.
Mesa has the structural characteristics that make a venture bet interesting: it targets a large, fragmented, and chronically underserved market with a model that gets more defensible as it scales. The diner pool is a network effect moat that compounds with density, and the SaaS-plus-marketplace revenue structure can produce strong unit economics at scale. The airline yield-management analogy is not just a pitch flourish; it maps cleanly onto the problem, and the incumbents (TheFork, OpenTable) have no incentive to build this because it would cannibalise their own discovery fees. That is a real wedge. The Spain-first geography is smart timing: high digital reservation penetration, dense urban dining culture, and a population comfortable with last-minute consumer apps. The principal risk for venture fundability is whether the diner pool can be built to a density that makes the real-time matching reliable across cities, because the value proposition to restaurants collapses without it.
A qualitative read, not a valuation or a promise of funding.
The core idea is sound, the timing is right, and the wedge is genuinely differentiated from what exists. The two-sided cold-start is the single most important thing to get right before anything else, and the good news is it is solvable through deliberate sequencing: seed one neighbourhood in one city densely rather than spreading thin. Madrid's Malasana or Chueca, for example, could be a self-contained proof of concept where diner pool density and restaurant coverage reinforce each other in a small geographic radius. If that neighbourhood works, the playbook is repeatable and fundable.
Run a single-neighbourhood pilot in Madrid with 8 to 12 restaurants and a manually recruited diner opt-in pool of 500 to 1000 locals. Measure two things only: the rate at which pushed offers convert to seated diners within the time window, and whether restaurants renew or expand their rules after the first month. Those two numbers will either validate the core loop or tell you exactly what to fix before you scale.
This is a recommendation, not a verdict. The call is yours.
This idea evaluation was generated by Firmgrove. The Firmgrove attribution is part of the document.