The same request, two ways

10 ideas (5 practical, 5 unusual). All under $500, startable in ≤2 weeks, aimed at 25-45 commuters next to the station. No loyalty cards or discounting. Assumes morning rush is real, you can do quick service, and basic signage/sampling outside is allowed.

Practical

1. Station-exit A-frame — Portable sign at the main exit listing 2-3 grab-and-go items with prices and “2 min walk.” Commuters see it before they decide to leave the platform area.
2. Pre-bagged combos — Coffee + pastry sealed in bags ready at 6am. Cuts transaction time to seconds so rush-hour people actually stop instead of walking past.
3. QR pre-order form — Free Google Form or simple link on a window/sign; name + items, pay on pickup. Removes the “will there be a line?” friction that kills morning stops.
4. Nearby office sample drop — One person walks 5-10 nearby offices with a tray of samples and a one-page “order by 5pm for tomorrow morning pickup.” Turns bulk into predictable volume without delivery staff.
5. Geo-targeted morning ads — $150-300 on Instagram/Facebook aimed at 25-45 within 1 km of the station, 6-9am only, with a clear “next to the station” message. Puts the bakery in the feed of people already heading that way.

Unusual

6. Train-timed chalkboard — Daily board outside synced to the 3-4 busiest departure times (“7:12 train special: still hot”). Creates a small “I have to catch this” reason to detour.
7. Platform-exit sample sprint — One staffer stands at the exit for 20-30 minutes at peak with free tiny bites and a clear “bakery is 90 seconds that way” line. Direct interruption of the flow.
8. Mystery commute bag — Fixed-price sealed bag containing 2-3 random breakfast items. Novelty + zero decision time; people buy it as a game instead of choosing.
9. Free charge + QR playlist — Cheap power bank station or cables near the counter plus a QR to a short “morning energy” Spotify playlist. Gives a micro-reason to step inside while waiting for a train or coffee.
10. Commute fuel stamp card (non-loyalty) — One-time physical stamp on a paper “ticket” the first time someone buys before 8am; collecting 5 gets a public shout-out on the chalkboard or Instagram. Status/visibility play instead of points or discounts.

Top 3 ranked against cost < $500, start ≤2 weeks, actual commuter pull

1. Platform-exit sample sprint (unusual)
   Meets all three hardest: near-zero cost (just product), start tomorrow, directly hits people already walking past.
   Risk: Station staff or security shut it down after one day.

2. Station-exit A-frame + pre-bagged combos (practical combo)
   Sign is ~$50-80, bags use existing product, both live by end of week. Captures people who see the sign and want speed.
   Risk: Sign gets ignored or stolen/moved if not weighted and checked daily.

3. Nearby office sample drop (practical)
   Cost is product + 1-2 hours of walking; can start this week. Creates repeat bulk orders from people who already pass the station.
   Risk: Offices ignore the drop or want delivery you can’t provide, so volume stays low.

Budget drove the drop. It accounts for 158 of the 160 total lost repeat orders (−23%). Regular was nearly flat (−2%), Premium actually rose.

Data quality

• Only issue: missing avg_order_value for Budget in August (as flagged). Everything else is complete.
• 27 rows, Jan–Sep, three segments, no duplicates, no negatives, no obvious outliers.
• Numbers are internally consistent (customers always > repeat_orders).

Assumptions

• customers = unique customers who ordered that month.
• repeat_orders = count of orders from customers who had ordered before.
• Comparison is the sum of repeat_orders over Apr–Jun vs Jul–Sep.
• Budget post-period AOV average uses only July and September (August is blank).

Change in total repeat orders (Apr–Jun vs Jul–Sep)

Segment	Apr–Jun	Jul–Sep	Change	% change
Budget	683	525	−158	−23.1%
Regular	520	508	−12	−2.3%
Premium	276	286	+10	+3.6%
Total	1,479	1,319	−160	−10.8%

Calculations (for checking)

• Budget pre: 224 + 231 + 228 = 683
  post: 196 + 171 + 158 = 525 → −158
• Regular pre: 174 + 170 + 176 = 520
  post: 172 + 169 + 167 = 508 → −12
• Premium pre: 91 + 93 + 92 = 276
  post: 95 + 94 + 97 = 286 → +10

Budget monthly repeat rate also collapsed (from ~53% to ~43%). Customers fell 4% while repeat orders fell 23%, so the remaining buyers ordered less often.

AOV jumped for Budget at the same time:

• Budget: 24.43 → 27.00 (+10.5%)
• Regular: 41.57 → 43.23 (+4.0%)
• Premium: 80.10 → 82.63 (+3.2%)

The July price increase is visible in the data and the timing matches the drop. The hypothesis is supported.

5-point summary for operations manager

1. Almost the entire drop in repeat orders came from the Budget segment (−158 orders). The other two segments are basically flat or slightly up.
2. The drop started right when Budget prices rose ~10% in July. The other segments saw smaller price increases and held up.
3. Budget customers are ordering less often (repeat rate fell from about 1-in-2 to about 2-in-5). The customer base itself also shrank a bit.
4. Put the $5,000 win-back money on Budget. Recovering even a portion of those 158 orders has the highest upside. Premium does not need it.
5. These buyers appear price-sensitive. A generic “we miss you” offer is less likely to work than one that restores value (discount, bundle, or loyalty incentive). Track whether the recovered customers stay after the offer ends.

Done. Single HTML file. Phone browser. No app, no login. Add / tick / live sync across phones.

Assumptions (stated, not guessed):

• Both phones have internet in the store.
• You share one unique link (room code is the only secret). Anyone with the link can edit — fine for household test, no login as requested.
• List lives in each phone’s browser storage + live sync via free public relay (itty.ws). Messages are not stored by them. If both phones clear site data, the list resets. Good enough to test the idea.
• “Within 2 seconds” assumes both are online and the page is open.

How to run it (non-programmer steps):

1. Download the file below.
2. Go to https://app.netlify.com/drop on a computer.
3. Drag the HTML file onto the page. You get a URL (e.g. https://random-name.netlify.app).
4. Open that URL on phone A. It auto-creates a room and updates the address.
5. Tap “Copy link” and send it to phone B (text/WhatsApp).
6. Both open the same link. Phone A adds “milk” → phone B sees it in ~1–2 seconds. Tick works the same way.

The page is phone-first: big touch targets, input stays at the bottom, warm off-white paper look, one green accent.

What it does (first version):

• Add item (type + Add or Enter).
• Tick to mark done (strikethrough, moves to bottom).
• Live sync: any change broadcasts the full list. Other phone updates immediately.
• Room code + copy link at the top.
• Persists on each phone for that room.

Limitations (honest):

• Not production-grade. No offline queue, no conflict resolution if both edit the exact same item at the same millisecond, no permanent cloud backup beyond the browsers.
• If the free relay is down, local list still works but the other phone won’t see updates until it comes back.
• Hosting is free and temporary-style; for long-term you can move the same file to any static host later.

Open the downloaded file on one phone if you just want a quick local look, but both phones need the shared hosted URL for the live test.