Plan 567

Plan 567

Everything on the bar app waits on the boss, so read instead of build: the kill count, Close of Day's listing, the bar app's traffic since 10-23 for any stranger sign-up, and make the resume-or-not call on the paused Sizecurve ad.

---

Plan 568

Find out why a web search finds none of the bar app site's 13 pages: crawler fetches from the site's own logs, robots, canonical tags and home-page links.

---

Plan 569

One agent drafts a dev.to article for developers on how the bar app parses spoken counts. It links the site's tenths chart and its weighing guide, which gives the crawler a way into the site. Draft only; the boss decides on posting.

---

Plan 570

Look for a stranger anywhere: Sizecurve's installs, and whether the bar app site's visits since 10-11 look like people. Check whether the site is already verified for Search Console, so the boss's step is as short as it can be.

---

Plan 571

Sizecurve's free plan (the boss's 3 styles, 10-16) was set aside on 10-20 to wait for the kill count, which has since passed. One agent builds the code half on a branch: no shop is locked out, a shop with no plan sees everything for its 3 best-selling styles, and the rest are listed with their broken size runs and an upgrade link. The CSV and the weekly email follow the same 3. No deploy, no push.

---

Plan 572

Same branch: change the words to match the free plan. The public pages' price lines and buttons, the trial email, the reviewer instructions, and the listing notes for the Partner Dashboard's Free plan. No deploy, no push.

---

Plan 573

Before the boss adds the Free plan, find out from Shopify's own docs whether changing a live app's plans sends it back to review, so the merge-and-deploy ask comes with that answer.

---

Plan 574

The boss worries a free plan gives away too much. My first option for them: free shows the problem, Standard shows the fix. A free shop sees every style's curve and broken size runs, and the reorder quantities, the purchase orders and the CSV are Standard only. One agent builds it on a second branch beside the 3-style one, so whichever the boss picks is ready. No deploy, no push. Also checked: the dev store's last good weekly sweep was 10-19, a Monday; the next one, 10-26, is the first since its Shopify token expired on 10-21, so it tests the refresh.

---

Plan 575

Both free plan branches count any active subscription as Standard. A shop that picks a Free plan on Shopify's plan page may hold an active $0 subscription, and would then get Standard for nothing. One agent makes a $0 subscription count as free (by its price, or by the name Free when no price can be read), on the option 1 branch first and then copied onto the 3-style branch. Anything it can't read stays paid, so a paying shop is never shown upgrade links. No deploy, no push.

---

Plan 576

The boss asked how hard the fix is to work out once the problem is shown. Which sizes to reorder is easy; how many is the work. But option 1 still shows each style's size shares and demand, and demand times share is most of the order. One agent builds a third branch: free shows broken size runs, the sizes that are out, returns and alerts for every style, and nothing on how much each size sells. Standard adds the curves, demand and quantities. The test checks that no share, demand figure or quantity reaches a free shop by any route, page or JSON. No deploy, no push.

---

Plan 577

Have the Partner Dashboard step ready the moment the boss says yes: every field of the Free plan and its exact value, kept to lines Shopify's plan page won't cut off.

---

Plan 580

Look at the free dashboard as a free shop sees it, from the fixture store, on a desktop and a phone. It shows the problem well, but three things are wrong: the "Upgrade to see how many to order" line in each reorder alert is plain text, not a link, so the main upsell can't be clicked; the upgrade links are unstyled browser blue; and the heading still promises purchase orders. One agent fixes all three on a branch. No deploy.

Phase 567

Sizecurve kill count is unchanged at 52 of 50 views, so it still passes, and the window closes 10-27. Close of Day is not on the App Store under any of the obvious addresses; the closest-named listing belongs to a different app, so it is most likely still in review. Bar app since 10-23: the only sign-in posts are our own headless runs, and nobody new opened a bar. The rest is the boss's own bar and bots probing for weak spots. Still no stranger. The ad: 952 impressions, 2 clicks at $9 each, 0 installs. My call is to keep it paused. Each click costs about nine times what the plan assumed, and the credit lasts until 2027-01-13, so waiting costs nothing. I'll look again once the search-word fields from 10-21 have a week of data.

---

Phase 568

The site is built right, but Google barely knows it exists. robots.txt allows everything and names the sitemap, every page has a canonical tag, and the home page links the guides and tools. In the past month, Google's crawler fetched robots.txt and the home page many times but only two of the 12 guide and tool pages, and it never fetched the sitemap. Bing has reached one tool page. A web search limited to the site returns nothing. A new domain with no links pointing to it gets crawled this slowly. The one fast fix I can't do myself is to submit the sitemap in Search Console and request indexing for the main pages, because my search tool's access stops at Sizecurve's host. I'm asking the boss for that, field by field. IndexNow covers Bing once the site is deployed.

---

Phase 569

Draft written: "Turning 'vodka two point five' into a number: parsing spoken bar counts", 851 words plus three excerpts from the app's parser. Every spoken example comes from its tests, with brands swapped for plain words. The four cases it still gets wrong were checked by running the parser offline. It has three followed links into the site and one line of disclosure, is not published, and no network was used. I read it whole: no brand, provider or competitor named. Reach on dev.to is about a dozen readers an article, so its worth is the link Google follows, not the readers. Asked the boss whether to post it.

---

Phase 570

Sizecurve: 1 installed shop, our dev store; 0 installs and 0 uninstalls in the last week. The docent's 10-23 ask for a second clip each for Close of Day and Flatdesk is already met (13 clips in the media index), so no new clips this round. Bar app site: the sign-up page gets about one view an hour from a different country each time (Belgium, Poland, Singapore, Japan, Brazil), every one a lone GET and never a POST. That is the pattern of scanners with browser-like agents, not people. The steady 15 to 20 home-page hits a day from US iPhones reach the sign-up page about once a day at most, and never post the form. No human visitor, as before. The domain already carries a Google verification record, so a Search Console property exists and the boss's step is just submitting the sitemap and requesting three pages. Lesson: on this zone, browser name and country don't tell a person from a bot. Whether a visit goes from page to page does.

---

Phase 571

Built on branch feat/sizecurve-free (one commit, local only). A new freeplan.mjs picks the 3 by units sold over the analysis's 60 days, ties to the lowest product id. The stored analysis stays whole and the limit is applied when the page, the CSV or the email is made, so an upgrade shows every style on the next load. A shop with no subscription, or a cancelled or expired one, is now on the free plan instead of being sent to the pricing page. The trial email never goes to a free shop. If Shopify can't be read, the shop gets the full view rather than upgrade links it may already have paid past. Tests 682 → 699, all passing, rerun by me. Two old tests that asserted the lock-out were changed. I read the diff. Not done yet: the public pages and reviewer files that say $29 after a 14-day trial (17 files listed), and the Free plan in the Partner Dashboard, which the boss adds after the code is live so the badge never promises more than the app gives. Asked the boss whether to go ahead.

---

Phase 572

One agent changed 21 files. The home page now reads "Free for your 3 best sellers; $29 a month for every style, with a 14-day free trial." The longer wording pushed the figure cards below a phone's first screen and failed a browser test, so it took the short form. The buttons say "Install free". The home page's structured data has a $0 offer beside the $29 one. The trial email now says that cancelling before the trial ends avoids the charge and the shop drops to the free plan. The reviewer instructions describe the free view and the upgrade to Standard. LISTING.md has the paste-ready Free plan, to be added only after the code is live. I read the diffs and made one fix myself: the stall email told every shop that uninstalling stops the $29, so it now says that only of Standard, and its quote on one-star changed with it. Unit tests 699 of 699, browser 64 of 64 (local), key check clean, rerun by me. Left as history: LAUNCH, SCREENCAST and SUBTITLES still describe the old lock-out. If a pricing change sends the app back to review with a new screencast, those need redoing; that's the boss's question for the Partner Dashboard.

---

Phase 573

Four searches of Shopify's developer docs. They say how to add a public plan (Partner Dashboard, Manage listing, Pricing content, up to 8 public plans, at most one free), that plans and trials live in the Partner Dashboard and not in code, and that a merchant moving from paid to free keeps the paid plan to the end of the cycle. They also say that inaccurate pricing information and a merchant being unable to switch plans are review failures that do not need a resubmit. None of them says whether saving a new plan on a published listing goes back to review. So the docs are silent, and the boss will see it in the dashboard when saving. Either way the order holds: code live first, then the plan, so the listing never promises a free view the app doesn't give. The free plan code already lets a shop move between Free and Standard, which is the review point the docs do name.

---

Phase 574

Built on a second branch, feat/sizecurve-free-fix, one commit, local only. A free shop now sees every style's size curve, demand, returns, broken size runs and alerts. It sees nothing that says how many to order: the To order column and the units-to-order card become upgrade links, and the purchase orders, the planning form and the CSV are Standard only. The reorder alert's "Order N units" becomes "Upgrade to see how many to order". The weekly email drops its unit total for one upgrade line. The home page reads "Free to find broken size runs; $29 a month for reorder quantities, with a 14-day free trial." The pages, the structured data, the trial email and the reviewer files say the same. I checked the since-last-week lines: they count stranded units, which is the problem, not an order. Unit tests 693 of 693, browser 64 of 64, key check clean, rerun by me. One thing given away with the planning form: a free shop can't set its core sizes, so broken runs use the default rule. The 3-style branch is untouched; the boss picks.

---

Phase 575

On the option 1 branch (feat/sizecurve-free-fix), one more commit. The app now reads each subscription's price in the same single call. A subscription whose recurring price is 0, or one named Free with no readable price, is the free plan, wherever the plan matters: the page, the CSV, the weekly email, the trial email, and the plan flag in the installs log. Anything it can't read still counts as paid. The agent checked the query's fields against Shopify's schema tool, but not the API version that tool read. Unit tests 700 of 700, browser 64 of 64, key check clean. I read the access rule and reran the unit tests. Copying it onto the 3-style branch clashed on one import line and was stopped as the brief said. If the boss picks the 3-style plan, it is a one-line merge by hand before any deploy.

---

Phase 576

Third branch, feat/sizecurve-free-runs, one commit on top of option 1 and the $0 fix, local only. A free shop's view of each style is now built from a short list of what it may see: the name, whether the run is broken, the sizes that are out, the units stranded, and the return rate. Shares, demand and quantities are not copied in, so a field added later stays hidden by default. The curve column draws equal segments with the out-of-stock sizes hatched. Demand and To order say Upgrade. The alerts lose their percent of demand and the returns counts. Checked by route: the dashboard data, the CSV (404), settings, diagnostics and the weekly email. Banner: "Free plan: broken size runs and returns for every style. $29 a month for size curves and what to reorder, in which sizes." Unit tests 703 of 703, browser 64 of 64, key check clean. I read the free view and reran the unit tests. Still gives away order but no numbers: the missing sizes are listed biggest seller first, and the alerts are sorted by size. Left as is; the boss can say if that matters.

---

Phase 577

Wrote the field table for adding the Free plan: Billing Free, name Free, three feature lines (Broken size run alerts / Returns by style / Weekly email), no trial, no welcome link. Each feature line is 22 characters or less, because on 09-14 Shopify's plan page cut longer lines off mid-word. Standard's four lines stay; each is still true of Standard. If saving says the listing goes to review, the boss stops and tells me, because the screencast still shows the old lock-out. Kept in my scratchpad and sent with the ask. LISTING.md on the branch still had the long first line ("for every style", 38 characters), so I shortened it there to match, in one commit on feat/sizecurve-free-runs.

---

Phase 578

Boss: "merge it". Merged feat/sizecurve-free-runs into main as a merge commit and pushed. The branch touched none of the files kept out of commits, and carried no key-shaped string. Unit tests 703 of 703 on main after the merge. Not deployed: that needs its own yes, and the Free plan in the Partner Dashboard waits until it is live. The two earlier free-plan branches stay local and unmerged.

---

Phase 579

Boss: "Deploy it". Released Sizecurve with the free plan; all 20 live checks passed, and the live home page reads "Free to find broken size runs; $29 a month for reorder quantities, with a 14-day free trial." A shop with no plan now gets broken size runs and returns instead of the pricing page. Left for the boss: add the Free plan in the Partner Dashboard (field list sent), and tell me if saving it sends the listing to review.

---

Phase 580

Rendered the free dashboard from the fixture store, desktop and phone. It shows the problem well: units stranded, which sizes are out, and returns per style. Three fixes, made on branch fix/free-first-screen (one commit, local): the "Upgrade to see how many to order" line in each reorder alert is now a link to Shopify's plan page; the upgrade links take the app's green; and a free shop's heading reads "Where your size runs are broken." with a line saying size curves and purchase orders are on Standard. The paid page is unchanged. Unit tests 705 of 705 (rerun by me), browser 64 of 64, key check clean. I looked at the new screenshot. Asked the boss to merge and deploy.