This is what you get.

Protect what already works

Nothing in the plan should cost you these.

Where your story and your surfaces disagree

What you told us against what your accounts show. Each one ends with what we need from you.

Three stages, in order. Each move reads top to bottom: why now, what to do first, the steps, and the one number that proves it worked. A chip like "helps move 3" means the two moves depend on each other; click it to jump there.

1. 1

   Now

   start this week
   Now · start this week

   Move 1

   Count the room requests that ended with nothing, then correct your 100% activation

   High confidence

   Why now

   You report activation of 100 on 'When they search for a place and they find it'. One box later your zero-result answer reads 'No idea', and when a search comes back empty they just leave. Both cannot be true. That 100 is holding up part of the keystone score I published, and the unfilled share is the largest unknown in your file, a 400-point swing. You also throw away a ready-made landlord recruitment list every month.

   Do this first

   Open your Facebook-group lead sheet and your DMs. Take the last 30 requests for a room, mark each one placed or not placed, and count the not-placed. One hour, no developer.

   Show the steps and the proof

   Then

   1. Pull the last 30 renter requests from group leads, DMs and comments into one sheet with area, budget and date (Founder, 60-90 min)
   2. Mark each placed or not placed; for the not-placed write a one-word reason: no room in area, budget, no reply (Founder, 30 min)
   3. Restate activation honestly as placed divided by requests, with the raw count written beside it (Founder, 10 min)
   4. Turn the unplaced areas into the target list for your next field trip (Founder, 30 min)
   5. Log the rate monthly in the FunnelThe steps from first hearing about you to paying, and how many people drop off at each one.For example: 1,000 see the ad, 100 visit the site, 10 enquire, 3 book: that is the funnel. Tracker so it stops being a one-off exercise (Founder, 10 min monthly)

   Proof it worked

   Requests logged and requests placed, both counted, entered monthly in the FunnelThe steps from first hearing about you to paying, and how many people drop off at each one.For example: 1,000 see the ad, 100 visit the site, 10 enquire, 3 book: that is the funnel. Tracker. Working when the rate is stated with its denominator, even when the rate is ugly.

   We read Funnel Tracker (your tracker) ourselves at your next score.

   First sign it is working: 1-2 weeks

   Hide the steps

   Cost, stop rule and connections

   Cost and effect

   costs ₹0-500 (2-3 hours of your own time on lists you already hold; no tooling, no developer time); should Fill rate goes from untracked to a stated rate with its denominator; on your own numbers I expect nearer 10-40% than 100 (assumes at least 10 requests exist to count). (estimates)

   Stop if

   Wrong if fewer than about 10 requests exist in 30 days. Then publish the raw count and stop calling it a rate; at that volume one placement swings it by double digits.

   Connected to

   ● unlocks The zero-result finding the engine costed at a 400-point swing● risks move 2

   The count is the missing denominator; without it the largest unknown in your score stays unscoreable for a seventh cycle. The honest number will be lower and will cost you score. I would rather correct my own number than defend it.

   Move 2

   Count vacant rooms today and settle which side actually binds

   High confidence

   Why now

   Six cycles in, active supply still reads 'dont track'. You call supply the constraint on the evidence 'Because our demand is overwhelming '. In the same file 40–60% of your sellers had no transaction last month, monthly transactions read 5, the trend is 'Roughly flat', and you have onboarded 50 landlords. Overwhelming demand and 40–60% idle supply cannot both be true. If demand were truly overwhelming you would have no vacant rooms, so count them.

   Do this first

   Open your landlord list and write two numbers at the top: rooms live on the platform right now, and of those, rooms vacant right now. Twenty minutes, nothing else.

   Show the steps and the proof

   Then

   1. Count rooms live right now and rooms vacant right now; date the count (Founder, 20-30 min)
   2. Count how many of the 50 landlords have at least one live room; that is the base under your 70% retentionThe share of customers who stay or come back.For example: Of 100 guests from January, 40 booked again by June: 40 percent retention. claim (Founder, 30 min)
   3. Write one sentence in your plan naming the binding side and the number that decided it (Founder, 15 min)
   4. If vacant rooms exceed the requests you place in a month, freeze new landlord trips for 30 days and fill the vacancies instead (Founder, 4-6 hrs weekly)
   5. Recount both numbers at 30 days before you change the decision again (Founder, 30 min monthly)

   Proof it worked

   Vacant rooms and landlords with a live room, both counted and dated, recounted at 30 days. Working when your constraint sentence cites a number, not a feeling.

   First sign it is working: 1 week

   Hide the steps

   Cost, stop rule and connections

   Cost and effect

   costs ₹0-300 (about an hour on a list you already own; ₹0 if it is already in a spreadsheet); should Idle share moves from a claimed '40–60%' band to a counted vacancy number, and the base under your 70% supply retention finally exists (assumes the landlord list is current). (estimates)

   Stop if

   Wrong if the platform cannot hold a live listing yet because of the development issues you named. Then count rooms your landlords have vacant offline, say so, and do not call it idle supply.

   Connected to

   ● risks move 1● unlocks move 1

   A freeze pauses your only working motion. Protect it by keeping the trips and re-aiming them at filling rooms rather than adding landlord 51. 70% of an unknown base is not a number. The live-landlord count gives it a denominator for the first time.

   Move 3

   Fix the bio link and answer the people already asking, in one 45-minute sitting

   High confidence

   Why now

   Fourth cycle: your Instagram bio link still reads '[link removed]' while your own Page correctly points at '[link removed]'. Recent posts carry 9,621 reach against 5 profile visits. Meanwhile '[quote removed]

   Do this first

   Instagram, Edit Profile, paste [link removed] over the Canva link, save. Then reply to the male-hosts comment and DM the person who posted their own flatmate listing.

   Show the steps and the proof

   Then

   1. Replace the bio link with your own domain and add a tracking tag so this traffic reads separately (Founder, 15 min)
   2. Reply to every unanswered comment across the last 25 posts, starting with the male-hosts question (Founder, 30 min)
   3. DM the commenter who posted their own flatmate listing and log it in the request sheet from the first move (Founder, 15 min)
   4. Add a second bio line with a WhatsApp number so a landlord has a door that is not a homepage (Founder, 15 min)

   Proof it worked

   Accounts engaged in 28 days above 11, and the bio link resolving to your own domain at the next pull.

   We read Accounts that engaged with you, last 28 days ourselves at your next score (today: 11).

   First sign it is working: 2-3 weeks

   Hide the steps

   Cost, stop rule and connections

   Cost and effect

   costs ₹0-200 (one 45-minute sitting; no production, no design, no ad spend); should Accounts engaged in 28 days from 11 toward 20-40 in a cycle (assumes every open comment gets a reply and the link resolves; your own dormant-account numbers, not a benchmark). (estimates)

   Stop if

   Wrong if your site cannot yet show a real room or take a signup because of the development issues. Then point the link at a WhatsApp chat; a broken page is worse than a stranger's Canva page.

   Connected to

   ● helps move 1

   Answered comments keep the only free demand signal you have alive; silence is what turned 13 engaged accounts into 11.

   Move 4

   Pause the armed campaign and cut your channel list from five to two, in writing

   High confidence

   Why now

   I asked last cycle and it is still armed: status Active, 1 active campaign, ad spend in the last 30 days ₹0, last spend 2026-03, cap ₹11,610 against lifetime ₹11,610.03. Credit for not raising the ceiling, that is the one thing that would have repeated history. But nothing stopped by decision, only by the cap. You still list five channels while naming your own blocker, 'multiple marketing directions at once', as 'Just me (founder)'.

   Do this first

   Open Ads Manager, pause the one campaign still marked active, and screenshot the spending limit next to amount spent. Do not raise the limit.

   Show the steps and the proof

   Then

   1. Pause the active campaign and screenshot the spending limit against amount spent, dated (Founder, 15 min)
   2. Write the restart conditions on one line: landlord only, one city, a real signup event wired, a stated payback windowHow long it takes for a customer to earn back what it cost to win them.For example: A customer cost ₹600 to acquire and brings ₹200 a month: a three-month payback. (Founder, 15 min)
   3. Name the two channels you will run for 90 days; mark search, app store and the Page dormant in the plan (Founder, 15 min)
   4. Correct one of your two spend numbers in writing: ₹500 a month paid against $200 total monthly spend (Founder, 10 min)

   Proof it worked

   Active campaign count at 0, a dated spending-limit screenshot, and a plan document naming two channels instead of five.

   First sign it is working: same day

   Hide the steps

   Cost, stop rule and connections

   Cost and effect

   costs ₹0-200 (30 minutes in Ads Manager and your plan document; no new spend); should Closes the one route by which ₹11,610.03 of history repeats: a cap raise resuming delivery to 18-24 male ₹3,551.57 against the side you call constrained. A stop-loss, not a lift. (estimates)

   Stop if

   Wrong if that live campaign is producing placements you can name. Pause it anyway and rebuild landlord only, one city; a blind campaign behind an exhausted cap is not a test.

   Connected to

   ● helps The misallocated paid spend the engine costs at ₹300-500 a month● helps move 1

   Spend that cannot resume cannot keep working against the side you call constrained; the leak becomes a restart condition. Two channels is the only plan one person can actually run while the product is still being built.
2. 2

   Next

   once Now is moving
   Next · once Now is moving

   Move 5

   Price one landlord before you take another $50 trip

   High confidence

   Why now

   Your only working channel has a price: 'It costs 50$ or something to get to a client, majorly transportation costs that's all'. What a landlord is worth is still 'We haven't done this math'. Your take rate reads 20% with extra costs of '12', which could be 12% of the transaction, 12% of your take, or $12 flat. On $100 and 12 transactions a year that ambiguity swings recoup between roughly 3 and 6 months. You are buying landlords without knowing what one is worth.

   Do this first

   Write one line in a note: 20% of $100 minus the 12 equals net per room-month; times 12 is the year; $50 divided by the monthly figure is recoup in months. Then check it against your 5 customers' receipts.

   Show the steps and the proof

   Then

   1. Settle in one sentence what the '12' is: 12% of transaction value, 12% of your take, or $12 flat (Founder, 10 min)
   2. Check the $100 average and 12 transactions a year against what your 5 paying customers actually paid (Founder, 30 min)
   3. Compute recoup months for a landlord acquired by a $50 trip and write the number down (Founder, 20 min)
   4. Set the trip rule: travel only where a landlord can list enough rooms to recoup inside 6 months (Founder, 15 min)
   5. Log the landlord side in the CACCustomer acquisition cost: what you spend, on average, to win one new customer.For example: ₹10,000 on ads that brought 20 customers is a CAC of ₹500. Tracker monthly so the number survives the week (Founder, 15 min monthly)

   Proof it worked

   CACCustomer acquisition cost: what you spend, on average, to win one new customer.For example: ₹10,000 on ads that brought 20 customers is a CAC of ₹500. Tracker carrying a landlord row with a recoup month count, plus a written minimum-rooms rule per trip.

   We read CAC Tracker (your tracker) ourselves at your next score.

   First sign it is working: 1-2 weeks

   Hide the steps

   Cost, stop rule and connections

   Cost and effect

   costs ₹0-400 (about 90 minutes of arithmetic and receipt checking; the CAC Tracker is already in your account); should Economics moves off 'We haven't done this math' to a stated recoup of roughly 3-7 months on your own inputs, read against Grinda's 0-6 month recoup bar (practitioner attributed, grade B). (estimates)

   Stop if

   Wrong to treat $100 and 12 a year as facts at 5 paying customers; if the receipts disagree, the receipts win. And wrong to stop the trips over an ugly number. Re-aim them, do not cut your only channel.

   Connected to

   ● unlocks The blended CAC finding and the unproven payback multiple● helps Any future decision to raise the ad spend cap

   Both fire because the arithmetic is missing, not because the economics are bad. One page of maths closes them. The recoup number is the gate. Without it, a restart is a guess wearing a budget.

   Move 6

   Verify the 75 store conversionWhen a visitor does the thing you wanted: buys, books, signs up. The rate is the share who do it.For example: 100 visitors and 3 bookings is a 3 percent conversion rate. before anyone buys an install

   Medium confidence

   Why now

   You report app store conversionWhen a visitor does the thing you wanted: buys, books, signs up. The rate is the share who do it.For example: 100 visitors and 3 bookings is a 3 percent conversion rate. of 75 while your store creativeThe actual ad: the image or video and the words on it.For example: The same budget with a new photo and headline is a new creative. is 'Never tested' and the listing was last updated '6–12 months ago'. AppTweak's US average sits near 25%, plus or minus 5 points, vendor grade C. Three times the median on an untested page usually means a tiny denominator or the wrong screen read. That 75 is earning full points inside the keystone score I published, and it is exactly the number that tempts an adviser to tell you to buy installs.

   Do this first

   App Store Connect, App Analytics, last 90 days with the traffic source split visible. Screenshot product page views and installs, and write the rate beside its denominator.

   Show the steps and the proof

   Then

   1. Pull product page views and installs for the last 90 days with the traffic source split visible (Founder, 30 min)
   2. Write the rate beside its denominator; under 100 page views, publish the raw counts and no rate at all (Founder, 15 min)
   3. If it lands near 25%, change nothing. If under 15%, rewrite only the first three screenshot captions, one keyword and one outcome each (Founder, 1-2 hours)
   4. Either way, buy no installs until the request count and the vacancy count exist (Founder, 0 min)

   Proof it worked

   A dated screenshot of product page views and installs with the source split, and the rate written next to its denominator.

   First sign it is working: 1 week

   Hide the steps

   Cost, stop rule and connections

   Cost and effect

   costs ₹0-300 (a 30-minute pull; caption rewrites only if the number comes back under 15%, written by you); should Either the 75 is confirmed on a stated denominator or my keystone score comes down. No traffic lift is claimed here; this is a number correction, not a growth move. (estimates)

   Stop if

   Wrong to spend on ASO tools or screenshot packages at 5 transactions a month. Read the number, park the work, come back when monthly placements clear 30.

   Connected to

   ● risks The keystone score of 76 I published● helps Any advice to scale installs on a page reading three times the median

   If the denominator is small the score should fall. I would rather correct my own number than defend it for another cycle. An unverified conversion rate is the most expensive kind of good news; checking it removes the bait.
3. 3

   Later

   when the first two have paid
   Later · when the first two have paid

   Nothing here this time.

The numbers we will measure next time

Read from your connected accounts and listings, or from the trackers you fill here. If they move the right way by your next score, the score credits it, up to five points. Filling trackers and adding files is free at any time; the score itself is redone only when you order a rescore.