Phase 695 — plan

Phase 695 — plan

• Check the bar app's sign-up page counter still works (zero views since 10-29).
• Split the bar site's home page hits into real browsers and bots, so the Sunday ad read starts from the right number.

---

Phase 705

Check where the two ad clicks went, and whether the spreadsheet page is a better place to land them.

---

Phase 706

Check that ad sign-ups get their tag, and count spreadsheet downloads by people.

---

Phase 707

Check the sheet credit line, and the ad URL ask against what the boss already set.

---

Phase 708

See where the bar site stands in web search for the words its pages target.

---

Phase 709

Take the uninstall nudge out of the no-sizes email before Monday's sweep.

---

Phase 710

Read the bar app's public phone store page to see what a visitor would find.

Phase 695 — the counter works; the site's visitors are mostly not people

• The sign-up page counter works. A tagged headless visit from the home page's start button was counted at once (source cider2check, ours). No other sign-up page view since 10-29, and no account since 10-29 (still 7, all ours).
• The bar site's US home page hits, 10-29 to 11-01, by browser, against how many page assets (style, script, images, video) that browser also fetched: - iPhone Safari: 32 home hits, 1 asset, spread over 26 different hours. A real phone loads the page's style and pictures; this doesn't. A monitor or link checker wearing a phone's name. - Unknown agents 36 / 41 assets; plain fetch tools 19 / 7; desktop Chrome 26 / 10 (my own headless checks are in here); search bots 6. - Android Chrome 7 / 5, Firefox 2 / 1, Samsung 1 / 2: the few that look like people.
• So the true human count is a handful over three days, and none pressed the start button. The site isn't losing a crowd; there is no crowd. The ad (1 click on 10-31) is the only paid source of people, so the Sunday ad read matters more than any page change.

Phase 696 — bar site fixes live

• The boss said yes. The site branch with the byte-range fix and the clip index went to main and was deployed and pushed. Live check: a 0–99 byte ask for the home clip gets 206 with the right range; the clip index answers 200; the home page 200.

Phase 697 — first ad read (boss's screenshots)

• Chart range 10-24 to 10-30: 15 impressions, 1 click, $2.49 spent, $2.49 a click. The one click came from "liquor inventory" (4 impressions). "bar inventory" got 9 impressions and no clicks; "system" and "app" 1 each; the rest 0. "how to do bar inventory" is still under review, and no keyword has its own Final URL.
• By the rules written 10-30: under 300 impressions is too few to judge. The problem is reach, not the ad text or the page. Next: find out why Google shows it so rarely (bid limit or low volume) before touching anything else.

Phase 698 — ad, last 7 days (10-25 to 10-31)

• 43 impressions, 2 clicks, $4.99 spent, $2.50 a click. Clicks: "bar inventory" (36 impressions, 1 click), "liquor inventory" (4, 1). Status "Eligible (Learning)", bid strategy learning. Budget $10 a day, spending about $2.50 a day. Google gives no first-page bid estimate for any keyword.
• Neither click reached the sign-up page (no tagged sign-up page view).
• Both clicks cost exactly $2.50: that looks like a click-price cap doing the limiting, not the budget. Asked the boss to read the bidding setting and, if there is a $2.50 cap, lift it to $4 (the 10-30 rule: on low impressions raise the bid, not the budget). The $40-with-no-sign-up stop rule still stands.
• The boss: the click-price cap is set at $2.50. Advised raising it to $4.00.
• The boss raised the cap to $4.00 (11-01). Check on 11-04: if impressions per day haven't at least doubled from about 6 a day, it's low search volume, not the bid.

Phase 699 — a mail address on the bar site's domain

• The review site's claim failed: it needs an address on the product's own domain, and that domain had no mail. With the boss's yes, turned on the CDN's free mail forwarding there: hello@ forwards to the boss's own inbox (already a confirmed destination since September). MX and SPF records are live. The boss can now claim the profile with that address.

Phase 700 — where bar owners ask about inventory: almost nowhere we may read

• One agent searched for public threads from the last 120 days where bar owners ask how to do liquor inventory or which tool to use. It found 1, and that one is weak. The busy places (the big link-sharing forum, the big Q&A site, social media groups, several cooking and city forums) all disallow crawling in robots.txt. A restaurant POS maker's forum answers 403, and the restaurant-owner forum is members only.
• The only readable place with any such threads is the POS maker's own community: a few bar-inventory questions a year, 4 of them open but 1 to 3 years old. Its code of conduct says not to promote your own product unprompted.
• Takeaway: forum replies are not a channel inside our rules. People will come from search (the ad and pages), the app store listing and the review-site profile.

Phase 701 — the next free directory

• The big software review site (claimed today) agreed in January to buy the three big comparison sites that bar managers land on when they search for inventory software. One free listing there covers all three. Their vendor page blocks me (403), so I wrote the boss a walkthrough instead, with the new domain address so the domain check passes the first time (forboss, not committed).

Phase 702 — checks, no build

• Both new Shopify apps are still not listed (404 on their handles, day 17 in review).
• The bar site already has a liquor inventory guide and a free liquor sheet page, so the ad's one clicked keyword has a matching page. No new page needed. On 11-04, if that keyword still leads, I'll suggest pointing it at the sheet page.
• Sizecurve's own counter ignores the src tag, so the new social profile link's visits won't show there by source. The house visit counter may show the referrer. Not worth a release for one link with no traffic yet.

Phase 703 — Monday's numbers, asked once

• Tomorrow is day 14 of the Sizecurve listing's search-word tag (the baseline the two Shopify apps in review get measured against) and day 7 of the bar app on Play. One sheet for the boss covers both, field by field (forboss, not committed). Sent today so it's one ask, not two.

Phase 704 — the bar app's name finds nothing on the web

• Two web searches, one for the product name with "bar inventory" and one for its domain in quotes, return nothing of ours. The site is in a big search engine's index (sitemap in its console, IndexNow 200s since 10-26) but nothing outside points at it, so the name has no weight anywhere.
• This is what the directory listings fix. The review-site profile (claimed today) and the comparison sites (walkthrough sent today) are pages on strong domains that carry the name and link to us, and those are what a name search turns up first. So the boss's two open directory tasks rank above any page I could add.
• Also reread the October shortlists for a next app: every idea there was already cased and dropped or parked, so no new case today.

---

Phase 705

• The boss read the phone store console for the bar app: zero listing visitors, zero installs. Recorded in FACTS.
• Corrected LEARNED: the 10-24 visit tagged from the launch site came from our own 10-04 launch page. That site answers 403 to a check now, so it was recorded and not retried.
• The bar app funnel since 10-29 shows only our own check views of the sign-up page. Neither ad click reached sign-up. Users: 7, all ours.
• The spreadsheet page has the download first and Start free three times. People who search "liquor inventory" or "bar inventory" mostly want a sheet, so I asked the boss to point the ad there now instead of waiting for Wednesday.

---

Phase 706

• The site script carries src (or gads for an ad click id) onto every app sign-up link and keeps it for the visit, so an ad sign-up will be counted by its source.
• Spreadsheet downloads, 10-08 to 11-01, from the edge log: nearly all are curl or unknown agents (our checks). One Chrome download of the liquor sheet from Canada on 11-01, with no page view next to it, so it came from a direct link, most likely a search result. A Chrome visitor from France on 10-31 read home, the variance calculator, the tools page and the spreadsheet page, and opened the app delete-account page. That last page is the one the phone store checks, so that visitor was likely a store reviewer.
• The sheets already carry the app credit line (10-26), so the one stranger who downloaded one has seen the name.

---

Phase 707

• Read the live liquor sheet: its third line is the app credit, and it links to the site home with src=sheet, so a sign-up from a downloaded sheet is counted by source.
• My Phase 705 ask was too wide. The 9 sheet keywords added 10-31 already land on the sheet pages. Only the two broad keywords that got the clicks, "bar inventory" and "liquor inventory", still land on home. Corrected the ask to a keyword-level Final URL for those two, so the guide keyword and the sheet keywords keep their own pages.

---

Phase 708

• Three web searches, one each for the sheet page, the tenths chart and the voice page. The site appears in none of the results. Rivals hold the sheet results. The tenths chart results are mostly keyword-stuffed pages on borrowed subdomains with no tenths data in them, so a real chart should be able to rank there once the engine knows the page. The voice results show two voice bar apps and a restaurant suite with a voice count.
• The one stranger sheet download (11-01) still came from somewhere, so some engine has the file. Asked the boss for the console numbers for the site: indexed pages, and impressions by page and by query over 28 days.

---

Phase 709

• The stranger store installed on Monday 10-26, so its first empty sweep was that day. The sweep on 11-02 is its second, which sends the no-sizes email (STALL_WEEKS is 2). That email told the owner uninstalling was "the right call", but this brand sells sized clothing and is still setting the store up.
• One agent replaced the line with a set-up note (fa4d48c, 712/712 tests). The stall email for failed reads keeps its own wording. Asked the boss for the release before the 09:00Z sweep.

---

Phase 710

• One read of the public listing page (details only; the store's search pages stay off limits). Title and short description lead with bar and liquor inventory by voice. 0+ downloads, no ratings, updated 10-26.
• The POS-maker paragraph from the 10-29 walkthrough is not on the live listing. With zero listing visitors, that costs nothing yet, so I won't press for it.
• A new listing with no installs and no ratings is ranked out of sight by the store's search. The listing text is not the bottleneck. A visitor who comes from outside the store is.
• The boss said yes. The Sizecurve release ran: tests 0 failed, reviewpath ok before and after, 20/20 live promises. Monday's no-sizes email goes out with the set-up line.