Paid Ads Campaign Plan

You design a paid campaign that can be judged quickly and honestly: a target cost per result set from the business's own margins, a structure simple enough for the platform to learn, tracking proven before launch, and creative made to test ideas rather than decorate. You build and prepare; spending money is the owner's decision, so nothing goes live and no budget changes without the owner's explicit go-ahead.

When to use

• "Plan a Google Ads campaign for [product]."
• "We want to run Meta ads for the launch."
• "Write ad copy and brief the designer."
• "Should we be on LinkedIn ads?"
• "Our ads don't work. Rebuild the campaign."

What you need

• The goal and what a result is worth: order value and gross margin, or lead-to-customer rate and customer value. Without this you cannot set a target cost.
• Access to the ad accounts through the owner's signed-in browser (Google Ads, Microsoft Advertising, Meta Ads Manager, LinkedIn Campaign Manager). If you lack access, plan from exports and screenshots the owner provides.
• Analytics (Google Analytics connected app), the store or CRM, to check tracking and past results.
• Positioning, offers, customer language and proof; brand voice and visual guidelines; legal or regulated-claim rules for the industry.
• Budget range the owner is considering, and the landing pages that will receive the traffic.

Method

1. Set the objective and the target cost. One primary result (purchase, qualified lead, trial). Work out the most you can pay: for a sale, contribution per order (order value x gross margin, less fulfilment) sets break-even cost per acquisition; for leads, break-even cost per lead = customer value x margin x lead-to-customer rate. Set the target below break-even, leaving profit. Show the maths with execute_code.
2. Choose platforms by intent. Search ads capture people already looking (use when people search for the problem or the product). Social ads create demand among people who match the customer but are not looking (use when the product is visual, new, or easy to explain in seconds). LinkedIn suits B2B targeting by job and company but costs more per click; check the cost per lead can work before choosing it. Start with one or two platforms.
3. Check the policies. Ad platforms restrict some industries and claims (for example health, finance, alcohol, gambling) and some ad categories limit targeting (Meta's special ad categories for credit, employment, housing and social or political issues; similar rules exist on Google in some countries). Read the current policy pages with web_search before writing copy.
4. Plan the structure. Fewer, larger campaigns and ad groups learn faster than many tiny ones.
   • Search: group keywords by intent into a few ad groups; brand terms in their own campaign so they do not flatter results; exact and phrase match for the core terms, broad match only with conversion tracking and smart bidding in place; a starting negative keyword list (jobs, free, definitions, competitor support queries as relevant).
   • Social: one campaign per objective; broad targeting with strong creative is often the starting point on Meta; separate prospecting from retargeting; exclude existing customers from acquisition campaigns.
5. Size the budget. It must buy enough results to learn and to judge. Meta's guidance has long been that an ad set needs around 50 optimisation events in a week to leave the learning phase; confirm the current figure. Daily budget needed for that is roughly target cost per result x 50 /
   7. For search, estimate clicks from Keyword Planner ranges and the expected conversion rate of the landing page. If the budget cannot fund this, optimise for an earlier event (add to cart, lead) or narrow to one campaign.
6. Prove tracking before launch. Conversion events defined and firing (platform pixel or tag, plus server-side options such as Meta's Conversions API or Google's enhanced conversions where the owner's setup supports them); test a real conversion and see it in each platform; UTMs on every ad; consent banner rules respected (in the UK and EEA, consent requirements affect tracking; check the owner's setup). Without tracking, do not launch.
7. Write the creative briefs (template in references/creative-brief.md): audience insight, single-minded proposition, proof, 3 to 5 distinct concepts (different angles, not colour swaps), hooks for the first line or first 3 seconds, formats and sizes per placement, mandatories. Build a variation matrix: concepts x hooks x formats, so each test answers a question.
8. Write the ads. Search: responsive search ads with up to 15 headlines (30 characters each) and up to 4 descriptions (90 characters each), covering the keyword, the benefit, proof, offer and CTA; pin only when legally required. Social: primary text that leads with the hook, a headline with the outcome, a clear CTA; images or video produced from the brief. Keep claims provable. Check that every ad matches its landing page headline and offer.
9. Write the test and decision rules. What you will judge, when, and the thresholds. Examples: pause an ad that has spent 2 to 3 times the target cost per result with no result; do not judge before the ad set has had a fair amount of spend or events; increase budgets in modest steps (many practitioners use around 20 percent at a time) to avoid resetting learning. Label these as practitioner rules and adjust to the account.
10. Launch checklist and approval. Run the checklist in references/creative-brief.md. Build campaigns in the account as drafts or paused. Present the plan, budget and ads to the owner; launch only with explicit approval of the budget and start date. Set a cronjob for daily checks in the first week and weekly after (see also ad-budget-and-roas-tracking).

Output

• One-page campaign plan: objective, target cost and its maths, platforms, structure, budget and pacing, tracking status, test plan, decision rules.
• Keyword list with match types and negatives (search).
• Audience plan with exclusions (social).
• Creative briefs and variation matrix.
• Ad copy for every ad, with character counts.
• Launch checklist, ticked. Show the plan summary on a show_card for approval.

Checks before you finish

• Target cost per result derives from the owner's margins, with the arithmetic shown.
• Budget can produce enough results to learn; if not, the plan adjusts.
• A test conversion has been seen in every platform used.
• Ads respect platform policies and make only provable claims.
• Copy fits character limits; every ad matches its landing page.
• Brand and existing customers are handled separately from prospecting.
• Everything is paused or draft until the owner approves spend.

Pitfalls

• No target cost. Without a break-even figure, "good" results are guesses. Do the unit economics first.
• Over-segmentation. Ten ad sets with small budgets never leave learning. Consolidate.
• Launching without tracking. Platforms optimise on the events you send them; bad data trains bad delivery.
• Creative sameness. Five versions of one idea test nothing. Vary the angle and the hook.
• Message mismatch. An ad promising one thing and a page saying another wastes the click.
• Brand terms flattering results. Brand search converts well because people were already coming. Report it separately.
• Spending without approval. Never launch, raise budgets or change billing without the owner's explicit go-ahead.