Crumbl built a business on a menu that expires every Sunday. The chain rotates four to six cookie flavors each week alongside its classic Milk Chocolate Chip, announces the new lineup on Sunday evening for the week starting Monday, and has served more than 300 flavors across more than 950 locations (Crumbl weekly menu). That weekly reset is the whole reason a small cookie chain generates a genuine data market: fan sites publish the lineup the moment it drops, food media rank the flavors, and resellers and suppliers track which cookies land where and at what price. All of it means reading a lot of stores, every week, on the clock.
Crumbl's site sits behind Cloudflare, and in our test the front door passed our request rather than challenging it, which makes the real work the store-by-store read behind it. This page covers Crumbl as a target, the jobs a proxy is right for, the ones it is not, the door as measured on 24 August 2026, and the type, setup, sizing and limits.
Which proxy fits
US residential proxies pinned to the metro of the stores you read, driving a real browser, one session per store. A static ISP proxy per account for any login. Datacenter exits loaded the homepage, as our test showed, and are the first addresses metered once the store menus are read on a weekly sweep.
Crumbl as a target
The weekly flavor lineup, the same nationally in broad strokes but with per-store differences in availability, price and pickup or delivery options. Store-level prices, which vary by market. The store locator, at more than 950 locations and growing. And the app, which carries loyalty and ordering and is not a data job.
Crumbl runs a JavaScript application behind Cloudflare. Past the door, the menu and prices load around a chosen store from Crumbl's own API calls.
Jobs a proxy is right for
- Weekly lineup tracking. Which flavors are live this week, where, and at what price, captured the moment the Sunday drop hits. This is the job the fan sites are built for, and it is a race against the clock every week.
- Store-level price monitoring. Crumbl prices differ by market; analysts and resellers track them per store.
- Availability and sell-through. Which stores have which flavors in stock through the week, read on a schedule.
- Location data. The store list with hours, growing fast, for site selection and competitor mapping.
- Cross-store comparison. How the same week's lineup is priced across metros, read from an exit in each.
Jobs a proxy is not for
Loyalty accounts. They attach to a phone number and the app on a device; farming offers runs against Crumbl's terms.
Ordering. A real pickup or delivery address and a payment method, not a network property.
The door, measured
On 24 August 2026 a single request to crumblcookies.com from a datacenter server in Germany, with a normal browser user-agent, returned HTTP 200 and the real homepage, served through Cloudflare, which passed the request rather than challenging it. The twenty elite HTTP proxies from our free proxy list that followed, one call each, split 15 that never connected and five that all loaded the homepage. So the door passed our client, and what defends the site is the metering behind it on the store-specific menu, which a shopper opens a few times and a script opens for every store every week. We did not measure that limit; log the first rising 403 or 429 rate per exit and treat it as the number. Fixing 403 errors through a proxy covers the response you are most likely to meet.
A note on timing that is specific to Crumbl: because the lineup drops Sunday evening and the interest peaks in the first hours, a weekly tracker hits the site hardest exactly when everyone else does, which is when a shared address is most likely to be metered. Pace and pool accordingly.
Metro-pinned residential exit
same city as the store
Store chosen
one session
This week's flavors + prices
record store and week
Rotate
next store, fresh exit
Type by job
| Crumbl job | Proxy type | Why |
|---|---|---|
| Weekly lineup tracking | Rotating US residential, metro-pinned, session per store | Reads each store as a local; survives the Sunday-night peak with a pool |
| Store-level price monitoring | Rotating US residential, metro-pinned | Prices are per market |
| Availability and sell-through | Rotating US residential, session per store | Stock is per store through the week |
| Locator and location data | US residential, slow pacing | Paged queries from one address is the metered pattern |
| Any login | US ISP (static residential) | One fixed, trusted location per account |
| One manual look at one store | Datacenter, or a fresh free proxy | The homepage passed; the weekly sweep will not survive |
For why the two address types are treated so differently, datacenter vs residential proxies.
Free proxies here
Five of five connecting proxies loaded the homepage, so Crumbl is friendly to a one-off look and no friendlier to a weekly sweep of hundreds of stores. Use a free proxy to test that your parser reads a store menu correctly, verify it in the proxy checker first, and read when free proxies are fine for the line. For the weekly tracker, our US residential proxies start at $0.44/GB, pay as you go, no KYC, pinned to the metro.
Setup
- Metro-pinned residential exit, real browser. A Dallas store from a Dallas exit; the store choice does the localizing. Launch flags are in the Playwright proxy guide.
- One session per store, timed to the drop. Read the lineup as soon after the Sunday-evening announcement as your schedule allows, record store, week and prices, rotate. Sticky vs rotating proxy sessions covers the two modes.
- Diff week over week, not day over day. The lineup is stable within a week; read it once at the drop and once mid-week for sell-through, and diff.
- Pool up for Sunday night. The peak is predictable, so scale the exit pool for that window and shrink it after. Avoiding IP bans while scraping has the rest.
Sizing
National lineup + prices, 200 stores, weekly:
store sessions per drop 200
pages per session 2-3 (menu, prices)
sessions per exit per hour ~10, paced
exits in the Sunday window ~20-30 US residential, metro-pinned
Residential meters bandwidth; menu pages with images blocked are small, so the store count sets the bill. Pricing is pay as you go and the balance stays, so the pool can grow for the Sunday window and shrink after.
The boundary
A metro address is the proxy's whole part; Crumbl's terms prohibit scraping through any address, loyalty accounts stay off the table, and an order still needs a real pickup address. Within those lines, a metro-pinned residential exit with a real browser lets a legitimate data job read the public weekly lineup and store prices the way customers in each city see them, one store per session, at a human pace, watching the block rate.
The chain whose value is in its app-only items rather than a weekly reset is in proxies for Chipotle; the general method behind any scheduled store read is in proxies for price monitoring.
Sources
- Crumbl weekly menu overview (rotating four to six flavors plus the classic Milk Chocolate Chip; new lineup announced Sunday evening; more than 300 flavors; more than 950 locations): crumblmenu.com
- HProxy measurement, 24 August 2026: one direct request from a datacenter server (HTTP 200, homepage, Cloudflare passed) and 20 elite HTTP proxies from our free proxy list (15 no connection, 5 homepage).