Good local guide, good tourist. Curate your city. Earn the mandate. Win the room.

Brief

Two-page version

The calls I made on the ambiguous parts

Core loop — two loops, one identity

Grind loop (daily, free). Open Scout, see what the guides you follow reviewed today, write your own review of a place you went, earn XP, and climb your city's leaderboard for a category (wine bars, coffee, dinner). This is the retention engine and the source of taste-graph signal. Boss loop (4–8×/year, paid). When a trip is detected, the app builds a mandate from the last 90 days of reviews and asks the user to confirm. The agent then bids in a private auction for boutique distressed inventory. The hotel is only revealed at close. The grind is the home mode. The auction is the travel mode. Same identity carries both.

Monetisation — two surfaces, no hotel commission

Free daily action that drives retention

Writing a constructive review earns XP. It's free, takes 30–60 seconds, and is the act that already happens in WhatsApp groups and Instagram stories. Scout just gives it a home and a leaderboard. Every review sharpens the taste graph that bids on hotels later, so the daily action and the eventual auction win share the same currency. Target: DAU/MAU ≥ 40% in seeded cities by month 2. That's a non-negotiable gate before paid acquisition opens.

Social mechanic for organic growth

Guide ranking and outcome cards. Public leaderboards, city by city and category by category (top wine bar guide in Lisbon, top coffee guide in CDMX), create status competition. Locals invite friends because "you'd be #1 for natural wine if you joined." On the auction side, every win produces a shareable outcome card ("Mandate met · saved €127 · Bairro Alto"). It hides the floor and the hotel rate, so the hotel side isn't damaged but the user side is brag-worthy. Full ranking mechanics in §07.

Positioning

Not Booking with an AI skin. Not Priceline's opaque category. Not "Beli with hotels." Scout is the first taste-graph-powered travel platform: a local curation game on the front, a consumer-side hotel auction on the back. The wedge is hotels because the perishability and commission spread is biggest there. The moat is the taste graph because no OTA can replicate months of daily local signal. The brief's Priceline reference is a useful contrast, not a category to copy. Priceline buys on price alone; Scout bids on taste-matched fit at a price ceiling.

Section 00

Market context — the hotel market where supply incentive is unusually strong

Picture a 90-room boutique on a Tuesday. Twelve rooms sit empty tonight, and tomorrow that revenue is gone for good. The obvious move is to drop the price — and it's the one thing the General Manager can't do publicly. A hotel sells one inventory of perishable rooms through six channels, and almost every channel is locked. OTAs like Booking and Expedia carry roughly 85% of online volume. Rate parity contracts cap any visible discount^[1][3]. Wholesale operators like Hotelbeds and WebBeds leak even cheaper rates into long-tail sites the hotel never agreed to^[9]. The pressure all points one way: hide discounts inside closed channels, stop showing them publicly.

Hotels are really two different markets. At chains like Marriott, Hilton, IHG, and Hyatt, pricing is set by automated revenue-management software (IDeaS, Duetto) and parity is strictly enforced. The property GM can't say yes to anything; selling in is a twelve-month enterprise sale to corporate. Boutiques are different. A single property, or a small group of three to fifteen, runs on basic property-management software (Cloudbeds, Mews). The GM still controls pricing, and parity is loosely policed at best. The GM personally feels every empty Tuesday, and personally signs the contract. That's the only side of the market where Scout can move at startup speed.

Why the existing distressed channels haven't fixed this. HotelTonight, Hotwire, and Booking Genius all do roughly the same thing. They stack a hotel commission on top of an already-discounted room, leak the outcome into public price trackers, and send the hotel whoever happens to click. But rate parity has a contractual carve-out for closed-user-group rates. That's the same legal slot Marriott Bonvoy, AAA, and AmEx Fine Hotels & Resorts already occupy. It's the structural opening Scout fills, and the reason it has to be members-only.

The slice Scout is choosing to operate in.

Section 01

The mechanic call — why a city-curation grind

The frequency gap is the structural problem. A hotel-only mandate fires 4–8 times a year. The auction works the day you use it, but the rest of the year you're a dormant account. Without a daily reason to open the app, the taste graph never deepens, the social graph never forms, and the auction becomes a thin product wrapped around a search box.

So the mechanic call is no longer "what kind of agent competition?" That one is settled (auction house, see §06). The new call is upstream: what daily behaviour earns a user a seat at the auction? There are three readings. Only one survives.

Why the other readings lose.

Priceline proved the auction mechanic works twenty years ago^[10][11]. The bidder was just blind. Scout's bid agent isn't blind because it carries a taste graph the user spent months building. The grind is the moat. The auction is the payoff.

Section 02

The user — Local Guide by default, Travelling Scout on trips

Scout has one user, in two modes. At home, they're a Local Guide: the friend everyone texts before a date night, the one with strong opinions on the new wine bar three streets over. On trips, they're a Travelling Scout: the same taste-confident person, just dropped into a city where their taste graph is empty and someone else's needs to fill in. Most users are both, alternating depending on the day. The grind is the home mode. The auction is the travel mode. The same identity carries both.

Today, the Local mode runs entirely on free labour spread across Instagram, Maps, group chats, and Beli. None of it compounds into anything the user owns. The Travel mode opens Booking.com and applies filters that don't really work ("boutique" is a marketing tag, not a real filter). After two hours they either give up or book something they'll second-guess. Mr & Mrs Smith curates well but at full rack rate. Scout's claim: route the daily local opinions into an asset (the taste graph), and that asset finally makes the Travel mode work.

The friend everyone texts before a date night.

Visits 2–4 new places a week. Has opinions on all of them.

On trips: boutique by default. Tired of the search.

Would compete for "best guide in this city" if it counted for something.

Other segments considered.

Two adjacent personas accelerate the wedge once locals are in. The first is the HotelTonight refugee. They've already made peace with not knowing the hotel name in advance, and they only need a trust layer that says "this place actually fits you." The second is the Going.com native: books the cheap flight first, now needs a hotel handled fast. Both are halfway to the boss tier already.

Section 03

User's mandate — derived from the taste graph, not typed in

The whole product hinges on one substitution. The user no longer chooses a specific hotel; they hand the agent a mandate, and the agent chooses. If the mandate is expressive enough, every auction win feels deserved. If it's too crude, every win feels like a gamble, and the trust falls apart. (Priceline's "star rating plus ceiling" was too crude; §01 covers why.)

The conventional fix is progressive disclosure: ask three questions now, learn the rest from rejection reactions. That's better, but it has its own problem. The mandate is still being authored at the moment of travel — when the user has the least patience and the least information about their taste in a new city. Reactions only refine the mandate if the user sticks around long enough to react, and the cold-start period is exactly when they don't.

The real fix: derive the mandate from the daily grind. By the time a Local Guide travels, they've already written 25–80 constructive reviews of places in their home city. Each one carries structured signal: "warm lighting," "natural wine list," "loud but in a good way," "third-wave coffee," "no service charge," "neighbourhood not centre." The taste graph isn't a wizard. It's a residue of months of opinions the user gave for reasons unrelated to travel. At trip time, Scout's agent — Mara — reads that residue, assembles the mandate, and enters the private auction without the user sitting down to write anything.

The design question collapses to a single sentence: "Has the user done enough local curation that the agent can bid in their voice?" If yes, the mandate is one screen of confirmation. If not, fall back to the 3-decision starter plus the reaction-learning path — and flag the auction outcome with lower confidence, because the agent is bidding on borrowed taste.

Section 04

Core loop — daily grind, travel boss tier

The product runs on two loops that compose. Loop A, the grind loop, is the daily local-curation cycle. Most users will spend most of their time here. Loop B, the boss loop, is the per-trip hotel auction. High stakes, low frequency — and the moment everything the user built becomes economically real. The daily loop is the actual product. The travel loop is the payoff.

Loop A · The grind loop (daily)

The grind loop is the daily product. Most users will run it dozens of times before their first trip. XP isn't a vanity number. It's the input weight on the user's taste graph and the signal that ranks them on guide leaderboards. Decay enforces recency. Without it, a user who reviewed 50 places three years ago would outrank someone reviewing three a week today.

Loop B · The boss loop (per trip)

Next: how Scout makes money from both loops, without quietly recreating the same misalignment it sells against.

Section 05

Monetization — two surfaces, no hotel commission

Scout has two revenue surfaces matching the two loops. The grind loop runs a creator economy. Local Guides earn from activations when their reviews drive real visits. The boss loop runs a savings-aligned auction. Scout takes a cut of verified savings; hotels pay zero commission. The structural rule across both: no revenue from hotels, no revenue from venues. All revenue from users (subscriptions plus savings share). If the user doesn't get value, Scout doesn't earn.

(Platform fees shown in USD; the running Lisbon example uses EUR to match the scenario's local currency.)

Surface A · Local creator economy (grind loop)

Surface B · Boss-tier auction economics (boss loop)

Rejected revenue models (hotel commission, venue commission, paid placement in feed, ads, savings-share without deployment credit)

Validation gates — must clear before paid acquisition opens

Section 06

How the auction works — the boss tier mechanic

The auction has two sides. On the supply side, a hotel quietly lists rooms it expects not to sell, with a secret floor below which it will not go. On the demand side, agents bid against that hidden number — each one carrying a taste-graph-derived mandate from a real user. When a bid clears the floor, the deal closes. The hotel's identity is only revealed at the close, and neither side ever sees the other's full hand.

Four moving parts. The listing is a distressed room plus a sealed floor, with an optional sweetener like breakfast or a room upgrade that only appears at the close. Floor estimation is the work the agent does to guess that hidden floor — triangulating from past closings, occupancy, days to check-in, and what's normal for the neighbourhood. Bid selection is how the hotel picks among bids that clear the floor: it takes the most committed, not the cheapest. And floor movement reflects the fact that floors drop as the room gets closer to unsellable. The agent's Patience setting decides whether to wait for that drop or claim the room now.

How the auction matures. The name on the door doesn't change. What matures is how literal the competition gets, from a few internal bid strategies on day one to a real two-sided live auction once enough hotels and agents are around to support it.

Why use AI agents instead of a simple rules engine?

You could in theory write the bidding as a rule engine: "if occupancy drops below 70% and check-in is within 48 hours, bid within 8% of the estimated floor." Rules like that are easy to audit, easy to reason about, and fast to run. They're also brittle in exactly the way this auction punishes. A fixed ruleset is a published strategy. Hotels read its bid pattern over a few weeks, then shift their floor timing to defeat it.

What Scout offers hotels that existing distressed channels do not^[6].

The math from the hotel's side. Take a €130 distressed night. Through HotelTonight, the hotel keeps about €110 after commission. Through Hotwire, somewhere between €100 and €110. Through Booking Genius, €111 to €117. Through Scout, the hotel keeps the full €130 — roughly 30% more per booking at the same headline price to the traveller. The user is still saving against the public rate; Scout's cut comes out of those savings. Lift on the hotel side, savings on the user side, no commission in between. That's the entire spread.

Section 07

Social layer — guide ranking, asymmetric follow, endorsement

The grind loop needs a social shape that compounds. Three primitives do the heavy lifting. Asymmetric following: locals follow locals; travellers follow locals. Guide ranking: a public leaderboard per city × category, with an Active-Guide cap that creates scarcity. Post-trip endorsement: the only social action a traveller can take, and the one that trains the matching algorithm. The social object isn't a feed post. It's the guide profile — a living taste graph that other people use as a mandate.

Section 08

Positioning — the bridge between local curation and travel commerce

What Scout is not: not another reviews app, not "Beli with hotels," not an OTA, not "Priceline with AI." Each of those framings borrows a category but discards the part that makes Scout structurally different.

Category claim: the first taste-graph-powered travel platform. A local curation game on the front, a consumer-side hotel auction on the back. Locals build a public taste graph by reviewing the places they actually go. That graph becomes a mandate when they travel, deploying into a private auction for boutique hotels' distressed inventory. The platform earns from subscriptions on the local side and verified savings on the travel side. Never from hotels, never from venues.

Why category creation over repositioning.

Scout's defensible pieces — labelled taste graphs, the city-by-city guide network, private hotel inventory, bid intelligence — only behave like network effects inside the frame "bridge from local curation to travel commerce." Inside the frame "better reviews app," the auction is irrelevant. Inside the frame "better OTA," the daily grind is irrelevant. Neither Beli nor Booking can copy this without becoming a different company. Beli would need a hotel auction stack and supply relationships. Booking would need to fire its hotel sales team and walk away from public search. That's the moat.

Section 09

GTM — local guides first, hotels second, paid acquisition last

The sequencing inverts the conventional hotel-first playbook. Hotels are no longer the seed; Local Guides are. A boutique GM partnership in Lisbon is worthless without a critical mass of taste-confident locals whose mandates will eventually deploy there. So we seed locals first (cheap, viral, no commercial conversations), prove the grind retains, and then bring hotel supply online for cities where we have the demand. Paid acquisition stays off until the three gates from §05 clear: grind DAU/MAU ≥40%, Pass conversion ≥10% of Active Guides, and ≥30% bridge rate from Level 5+ guides to first-trip mandate.

City order is unchanged: Lisbon, then Porto, Mexico City, Lyon, Edinburgh. The wedge cities have to be ones where (a) the boutique GM has pricing authority on the supply side, and (b) the local food, bar, and coffee scene is dense and identity-forming enough to support a competitive guide leaderboard. London, New York, Paris, and Barcelona stay out of Year 1. They're chain-saturated on the hotel side and so over-reviewed that no new guide can credibly claim "I know this city."

Section 10

What's next — once the bridge is real

"Successful" means three things are simultaneously true. The grind is a real habit (DAU/MAU ≥40%, Pass conversion ≥10%). The boss tier is live in at least two cities beyond Lisbon. And outcome cards are generating more inbound than referral codes. None of that is assumed. But if all three hold, the question stops being "does the bridge work?" and becomes "what is the taste graph for?" The answer determines whether Scout is a hotel-deals app, a city-curation game, or something more general.

Year-3 north star: Scout is the taste-graph primitive that powers consumer-side auctions in any perishable, mandate-fit category. Hotels are the wedge. The same mechanic — grind loop, closed group, mandate, savings share — re-runs against any supply where (a) inventory is perishable, (b) the operator has real pricing authority, (c) a closed buyer group can be carved out of public price parity, and (d) the user's daily taste graph carries usable signal for the bid. The 3-bet sequence below is ordered by how much of Scout's existing infrastructure each one inherits.

References

Primary set — distribution, rate parity, NYOP economics

1. Lighthouse — Uncover common hotel rate parity issues to combat disparity
2. Techspian — How OTA Inventory Systems Revolutionize Travel Operations
3. idhotelier — Strategic Governance and Implementation of Rate Parity in Hotel Distribution
4. AltexSoft — Travel Agency Inventory System: Main Sources and Management
5. EHL Insights — How Revenue Management Works in Hotels
6. SiteMinder — Online Travel Agencies (OTAs): Complete Guide for Hotels
7. SiteMinder — What Is a Global Distribution System (GDS)?
8. SiteMinder — Hotel dynamic pricing: Complete guide with examples
9. Lighthouse — 3 ways to tackle rate disparity with wholesalers
10. Wikipedia — Name your own price
11. Unique Business Models (Substack) — How does Priceline make money?
12. UCLA Econ (Sboard) — Priceline: Name Your Own Price Introduction (PDF)
13. Review of Economic Studies — Name Your Own Price at Priceline.com: Strategic Bidding and Lockout Periods
14. Journal of Retailing and Consumer Services — Posted price and name-your-own-price in a product line design problem
15. SiteMinder (YouTube) — The Basics of Online Travel Agents for Hotels

Revenue management — videos & primers

16. YouTube — Hotel Revenue Management (overview)
17. Hotel-Spider (YouTube) — Revenue management in the hotel industry — Basics
18. Little Hotelier (YouTube) — What is hotel revenue management?
19. SiteMinder (YouTube) — Hotel Revenue Management — Simplified!
20. techtalk.travel + Expedia (YouTube) — Basics & Adoption of Revenue Performance Management Principles for Small, Independent Hotels

OTA supply chain, inventory, bedbanks

21. AltexSoft (YouTube) — Hotel Q&A: OTA Connection With Hotels
22. Gimmonix (YouTube) — Gimmonix reveals how OTA, bedbanks and wholesalers manage their hotel offerings
23. AltexSoft — Bed banks 101: Wholesaler Role in Travel Distribution
24. SiteMinder — What is a bed bank? List of examples for hotels
25. Mize — The Most Complete Hotel Wholesalers and Bed-Banks List
26. HotelRunner — Top 5 Global Bedbanks You Must List Your Property
27. Hotelbeds — Hotelbeds: Travel Technology Provider
28. Chekin — How OTAs Work: What Every Hotelier Needs to Know

Opaque pricing & Priceline economics

29. YouTube — Complete Guide to Opaque Pricing in Hotels
30. OccupancyBoost (YouTube) — Opaque bookings explained for Hoteliers and why you need them
31. Corporate Finance Institute — Opaque Pricing — Overview, How It Works, Benefits, Example
32. YouTube — Priceline: a Startup That Won Travel
33. Phocuswire — Opaque bookings on the wane: Priceline, Hotwire, and a changing landscape

Market players — OTAs, channel managers, market share

34. Mize — Online Travel Agencies Market Share Across the World
35. Cloudbeds — The 10 Best Online Travel Agencies in 2025 for Hotels
36. Lighthouse — Top 10 Online Travel Agencies (OTAs) for your hotel
37. SiteMinder — Best Channel Manager for Hotels
38. SiteMinder — The World's Leading Hotel Channel Manager
39. Eviivo — 8 Best Channel Managers for Hotels to Know in 2026
40. Statista — Top online travel companies by market cap 2025
41. Wikipedia — Trip.com Group

Comparable channel discount & CAC benchmarks (S05 unit econ)

42. The Traveler — Booking.com vs. HotelTonight vs. Priceline: Who Has the Biggest Discounts? (HotelTonight ~16% average off published rate)
43. SimplyCodes / Hotwire — Hotwire Promo Codes & Coupons (Hot Rate hotel bookings average ~37% more savings than standard bookings)
44. Booking.com — Genius loyalty program (10-20% by tier)
45. Userpilot — Average Customer Acquisition Cost (CAC) Industry Benchmarks (2026)
46. Business of Apps — App User Acquisition Costs (2025)

GTM benchmarks & cross-category perishability (S09)

47. Viral Loops — How Robinhood's Referral Built a 1M User Waiting List (~3.0 viral coefficient; 2/3 of waitlist from referrals)
48. kwokchain — Notes on Superhuman's Acquisition Loops (70% of weekly new users from referrals)
49. Revenue Hub — RevPASH for Improved Restaurant Revenue Management (restaurant tables as perishable inventory; direct hotel-RM analog)