theStacc Ads · every platform · proposal · 44 screens · example data
Ads on every platform: find, test and run the platforms that bring real customers
A proposal on top of the v4 Ads mock and the lead-target proposal. theStacc ranks every ad platform for one business, runs small capped tests where a test can actually give an answer, counts each lead once across every platform, and keeps moving money toward what brings booked jobs. Ads come from one approved kit built by theStacc's existing writers and image service.
What changes from v4
- Plan replaces Create and the setup steps: the goal, proven spend, tests with their keep-or-stop line, and every other platform ranked with a plain reason. One approval.
- One Test page for every platform. Steps become states: set up, in review, live, result. Platform differences fill four slots: how you connect, what only you can do there, the preview, and who sees it.
- Ad kit: approve 7 facts and 4 ideas once; every size on every platform inherits that approval.
- To do replaces Fixes and collects every approval, decision and owner-only step. Leads counts each person once and sends booked jobs back to every platform.
- Menu: Overview, To do, Plan, Campaigns, Leads, History, Settings. Reports open History.
- Main example: Northside Plumbing & Drains, Columbus, Ohio, testing ChatGPT ads on water heaters while Google verifies it for Local Services Ads. Variants: a Pune dental clinic, a US online store, a small salon, and a Hindi-first coaching institute in Indore for tier 2 and 3 towns.
- Decided 25 Oct: tests turn on by themselves with live tracking against expectations; the owner chooses how autopilot starts (earned after 3 approvals by default); the video editor opens to customers; ChatGPT ads, Microsoft and new Google accounts test on theStacc's own ad accounts with test money paid upfront at cost.
What the 118 existing v4 and lead-target screens become
| First run and connect (18) | The Plan page on a first visit, and the Connect section of each Test page, shown when it is needed |
| Overview states (10) | Overview, now across platforms |
| Fixes (26) | To do, with the evidence panel |
| Create and draft review (27) | The Test page states, the Ad kit and the details panel |
| History and Reports (9) | History, each month opening with its report |
| Campaigns and Settings (20) | Campaigns grouped by platform; Settings tabs |
| Lead target and autopilot (8) | Plan, Overview, To do and the Autopilot tab |
Plan the month 6 screens
PL01Plan: the month, with testsGoal, proven spend and two tests paid from the gap under the cap, each with its keep-or-stop line. Below, every other platform ranked in four groups: later, coming later, track only, not for you. One approval.Needs backend: planner; test sizing from break-even and cost-per-lead priors; platform eligibility rulesPL02Plan: one platform openedChatGPT ads in a side panel: our verdict, how the ad looks, who it reaches, what it costs, what the owner does there, and the honest risks.Needs backend: OpenAI Ads API connector; chat-card previewPL05Plan: approved, now a launch boardAfter approval the same page tracks each test: the owner's steps on ChatGPT ads, and Google's checks for Local Services Ads.Needs backend: test status across platformsPL03Plan for a dental clinic in PuneThe same page for an Indian clinic. ChatGPT ads, TikTok, Local Services Ads and AppLovin are out, each with the reason. Practo and Justdial have no API, so they are tracked side by side.Needs backend: India eligibility rules; track-only platformsPL04Plan for an online storeAn online store ranks differently: orders a day, catalogue and video decide. No new test during Black Friday week.Needs backend: Shopify signals: orders per day, catalogue, video countPL06Plan when a test cannot answerA small salon at $700 a month gets no test: the cheapest readable one would take 80% of its budget. The page says what would change that.Needs backend: minimum readable test rule
Ad kit 3 screens
AK01Ad kit: facts and ideas, approved onceSeven facts and four ideas. Every size for every platform comes from these, so adding ChatGPT ads or Microsoft needs no new approval.Needs backend: ad spec table per platform; claim ledger; text as a layer over text-free images in 1:1, 4:5, 9:16, 1.91:1AK02Ad kit: two things need youA line taken from the website that is not yet an approved fact, and a photo of a customer with no consent on record.Needs backend: website-fact confirmation; photo consent ledger with revocationAK03Ad kit: edit one lineOne line with its length on every platform, and one photo with every crop following a single focus point.Needs backend: per-platform length checks; focus-point crops
A test, from set-up to result 10 screens
TS01ChatGPT ads test: set up on theStacc's accountThe default for ChatGPT ads: the test runs on theStacc's own ad account under the owner's name. The owner's part shrinks to about 3 minutes: check name and logo, agree once, pay the test money upfront.Needs backend: OpenAI Ads API connector; encrypted key storageTS19Let theStacc advertise for youOnce for every platform where a test runs on theStacc's account: seven plain points and one business document. This is the written authority the platforms' terms require.Needs backend: client authority record; business document vettingTS20Pay the test moneyTest money is paid upfront at cost, so theStacc carries no credit risk and nothing is spent that the owner has not paid for. Unused money comes back within 7 days.Needs backend: prepaid test wallet; refunds; sales tax and GST linesTS02ChatGPT ads test: ready to createName and logo checked, authority agreed, test money paid. OpenAI's brand check is still running; creating now is safe because it arrives paused.Needs backend: key check against GET /v1/ad_accountTS03ChatGPT ads test: detailsWho sees it and the money, editable. Below, exactly what gets created in OpenAI's own terms: campaign, ad group, four chat cards.Needs backend: geo lookup for markets; paused publish with read-backTS04ChatGPT ads test: in reviewCreated paused and read back. OpenAI publishes no review time, so we check every 30 minutes. Tests turn on by themselves once approved, by default; the owner can switch that off.Needs backend: review polling; serving issues; auto turn-on for testsTS05Turn on: one dialog for ready testsMoney starts here, so this is a dialog. It covers every ready test at once and says which one is still waiting.TS06ChatGPT ads test: live, too earlyDay 19. Real leads against the keep line, a likely range instead of a guess, and the platform's own count shown beside ours.Needs backend: live read on deduplicated leads; credible rangeTS17Live test behind what we expectedTests turn on by themselves, so theStacc watches them against what was expected. Here, day 9 with no leads: the ads and page work, but three Sunday calls went to voicemail. One suggested fix that leaves the test rules alone.Needs backend: expected-vs-actual bands per test; alert rules; call recordings; tracked-number schedulesTS07ChatGPT ads test: the resultThe test passed on theStacc's account. Keep it and move it to the owner's own account (recommended), keep it on ours for up to 60 days, or stop. How sure we are, in plain words.Needs backend: verdict engine on real leads and booked jobs
The same Test page on other platforms 10 screens
TS18The same test on the owner's own accountThe alternative for owners who want their own ChatGPT ads account from day one: about 25 minutes on OpenAI's site, a key to paste, and a business check that can take a week.Needs backend: OpenAI Ads API connector; encrypted key storageTS08Local Services Ads: Google's checksPay per lead, but only after Google's background, licence and insurance checks. The steps tick themselves as Google confirms them.Needs backend: Local Services API: verification artifacts, budget, lead feedbackTS09Microsoft: a copy of Google SearchEverything that changes on the way to Microsoft, and why we copy it ourselves instead of using Microsoft's scheduled import.Needs backend: Microsoft Advertising connector; campaign copy with differencesTS10Meta for a clinic in Pune: WhatsApp chatsImplants only, ads that open a WhatsApp chat. Meta's AI edits are off and the health-data limits are explained.Needs backend: Meta connector; click-to-WhatsApp; creative feature opt-outsTS11AppLovin for an online store: guidedAppLovin's campaign API is invite-only, so until it enables theStacc we prepare everything and the owner presses Create there. A fair test needs 5 to 10 times the cost per order each day, judged against states held back.Needs backend: AppLovin access for theStacc; AppLovin pixel and server events; customer videoTS13Google Demand Gen for the Pune clinicAnother Google format on the same page: a small reach test on Shorts, Discover and Maps, with every Google auto-edit listed and switched off, and the floor read live.Needs backend: Demand Gen create path; live minimum budget; asset automation opt-outsTS14Reddit for an online storeReddit through its open API: the owner's steps there, ads written like posts, and the three things Reddit adds by default that we switch off: comments, Redditor Highlights and expanded targeting.Needs backend: Reddit Ads API v3 connector; structured posts; Reddit Conversions APITS15Nextdoor for the plumber: coming laterHow Nextdoor will look once Nextdoor approves theStacc as a partner, and the guided setup until then. Nextdoor has no radius, so the radius becomes ZIP codes.Needs backend: Nextdoor partner access; radius to ZIP conversion; Nextdoor conversions APITS16TikTok for an online store: boost your best postsTikTok now runs every ad as the owner's real account, so the test boosts the store's own best posts, plus one new video from the video editor shown only as an ad.Needs backend: TikTok Marketing API v1.3 Smart+; Spark ads from organic posts; TikTok Events APITS12A stopped test, kept on recordAn earlier Meta test that failed: what it cost, the rule that stopped it, and what might work instead.
India: tier 2 and 3 towns 7 screens
PL07Plan for a coaching institute in IndoreHindi-first, for students from district towns. Google and WhatsApp ads stay proven; ShareChat + Moj and Dailyhunt are the tests; ChatGPT ads fail the affordable gate. Offline channels are counted with their own numbers.Needs backend: India tier 2/3 eligibility rules; assisted buying for platforms with no APITS21ShareChat + Moj test, in HindiNo API, so theStacc's team runs it in the owner's own ShareChat account, linked to theStacc's Business Centre: ShareChat's terms ban resale. Hindi ads with English meaning, the coaching rules that bind the words, settings, keep line and expected numbers.Needs backend: ops-assisted buying on theStacc's ShareChat account; Hindi video with Sarvam voice and AI labelST04Auto-rickshaw panels: you book, we countAn offline channel the owner already pays for, given its own number, missed-call reply and QR, so its cost per real enquiry sits next to Google's.Needs backend: tracked numbers and QR per offline channel; missed-call WhatsApp replyOV01Overview across platformsLeads against the target counted once, the three things that need the owner, every platform side by side with its own claim in grey, and what autopilot did.Needs backend: deduplicated lead count across platforms; pace forecastTD01To do: one queue for everythingApprovals, decisions, owner-only steps and leads to mark. The selected ask raises the month's total, so it needs the owner: before and after, why, how sure, and the safety line.Needs backend: autopilot asks across platforms; expiryPH01Phone: To doThe Ads home on a phone: target, then one card per thing waiting.Needs backend: push and email notificationsPH02Phone: decision sheetOne ask from a notification. Approve or Not now.
Overview and To do 3 screens
LD01Leads from every platformEvery real lead counted once, with where it came from. Calls under 60 seconds and repeat callers are shown but not counted.Needs backend: lead ledger; click IDs from every platform; call tracking; CRM syncLD02One lead, two platformsClicked a ChatGPT ad, later a Google ad, then called. Counted once. Reports credit the last click; judging a test splits it half and half. Both platforms are told it booked.Needs backend: attribution rule; offline conversion uploads per platformPH03Phone: mark leadsAfter a call, one tap. That is what every platform learns from.
Leads 5 screens
CA01Campaigns on every platformGrouped by platform with subtotals, tests tagged with their day, and campaigns made outside theStacc shown but not touched.HI01History with the monthly reportEach month opens with its report. Every change says who made it, before and after, and when it gets checked.ST01Settings: platforms and trackingEvery platform account, the owner's steps there, and whether leads reach it. One tag, tracked numbers, and where booked jobs are sent.Needs backend: connectors per platform; one tag sending to each conversions APIST03Settings: autopilot and approvalsThe owner chooses how autopilot starts. By default it is earned one kind of change at a time, after 3 approvals; it can be on from the start or off. Tests turning on by themselves is a switch here too.Needs backend: autopilot permissions per kind of change; earned-autopilot counterST02Add a platform with no APIYelp, Angi, Practo, Justdial: enter what you pay, get a tracked number, compare on cost per booked job.Needs backend: track-only platforms; tracked numbers per platform