Ads module · proposal on top of mock v4

Lead target and autopilot

The owner states one number: how many leads they want a month. theStacc works out the budget, which platforms to use and how many ads, then keeps adjusting toward that number inside limits the owner sets. These 8 screens are built inside the v4 app shell with the same classes and example data (Brightside Family Dental, Pune), plus one online-store variant to show the plan changing by industry and country.

Start at LT01 →

What changes from v4

SU12 Goal and budget becomes Lead target (LT01, LT02). The target sits next to value and the cap; the side panel shows the budget the target needs and whether it is reachable.
New step: the plan (LT03, LT05). One row per platform with budget, number of ads and a plain reason, including platforms we skip and why (for example AppLovin does not accept clinic ads).
New step: autopilot rules (LT04). Three tiers: automatic with undo, ask first, never (enforced in code).
Overview leads with the target (LT06): leads so far against target, counted once from call tracking and forms rather than platform numbers.
Autopilot approvals and History (LT07, LT08) reuse the Fixes before/after pattern and the History log, including decisions it held back on.

Rules the planner follows

Budget the target needs = target leads × expected cost per lead (Keyword Planner for the city, then the account's own numbers after 30 days). Flag when it exceeds the cap or the city's search volume.
Concentrate before spreading: one platform until it has about 30 conversions a month, then test the next with about 20%.
Eligibility per platform by industry, country and volume. AppLovin only for apps and online stores with about 10 purchases a day and enough 30 to 60 second video; never for healthcare facilities.
Autopilot checks daily, acts about once a week per campaign, moves at most 20% of budget a week, never exceeds the cap, never edits during platform learning.
Optimise on deduplicated real leads (and leads that became customers), not platform-reported conversions.

Screens

LT01Setup 3: lead targetCost-per-lead estimate by industry and city (Keyword Planner and benchmarks); feasibility maths LT02Setup 3: target needs more than the capSearch-volume ceiling per city; second-platform cost estimate LT03Setup 5: the planPlanner: target to budget split per platform; platform eligibility rules per industry and country LT04Setup 5: autopilot rulesAutopilot rule engine: auto-apply tier on top of existing suggestions; approval tier; hard limits in code LT05Plan variant: online store (US)AppLovin connector (needs AppLovin API whitelisting); eligibility by industry, country and purchase volume; creative count check LT06Overview: progress against the targetDeduplicated lead count from call tracking and forms; pace forecast; autopilot activity feed LT07Autopilot asks: start testing MetaApproval tier for platform start; before/after forecast; Meta connector LT08History: what autopilot didAutopilot decision log incl. held-back decisions; 30-day undo

Orange text on each row is the backend work that screen relies on. Example data only; nothing here talks to an ad platform.