Decide every ad move by how many conversions the account gets
Google and Meta learn only after enough results. Small businesses rarely hit that, so our system has to plan and change things around that limit.
The method in 5 steps
- Pick the result the budget can afford about 15 times a month (a booked call, a lead, a sale).
- Size the budget backwards from Keyword Planner cost per click. Google needs about 30 conversions a month to bid well, Meta about 50 a week.
- Keep it to one campaign with a few themed ad groups. Splitting starves the learning.
- Write only from approved facts, checked against a rule list per business type (doctors, finance, real estate, education).
- Change slowly, and only after the data says so. Example: pause a keyword after it spends 3× the target cost with zero results. That's about 95% sure it's a loser.
What our plan gets right and what to change
Right
- Google Search first for most businesses
- Ad text only from approved facts (also a legal need in India)
- Publish paused, every change approved
Change
- Online stores start on Shopping, not Search
- Two ads per ad group, not one
- Bidding on conversions only once tracking works
- Don't set a target cost in Google before about 30 conversions
- Suggestions show how sure they are
The open-source repos
None of them plans campaigns. They're useful for safety: exact-change approvals, budget caps, spotting a campaign that suddenly stops. Their numbers are unsourced and none are set up for India.
Watch out
- Google and Meta changed several things in 2026 that break older tools.
- India's new medical ad rules (from 6 Oct 2026) ban testimonials, before/after photos and offers for doctors.
- There are no reliable Indian cost benchmarks. We'll learn them from our own accounts.
About 9,000 words with sources, tables and thresholds.