Why this matters
When new and returning people share a campaign, its results blend two very different audiences. Past visitors and members convert cheaply, so the campaign's average cost per result looks healthy while the prospecting inside it may be failing, or starved. Nobody can cap the follow-up, test it or move money between the two jobs, because they are not separate anywhere.
This exercise applies two readings. The prospecting vs remarketing comparison explains why RemarketingAdvertising to previous visitors or engagers. reports flattering results and must be capped and judged separately. The campaign structure guide says a split earns its place only when it needs its own budget, target or report, and that names should encode the splits in a fixed order. The Google Ads vs Meta Ads comparison is useful background on why search and social play different parts in the same sequence.
The studio and its campaigns
Stillwater Yoga Loft is a fictional independent yoga studio above a high-street shop. New students start with an introductory offer bought online; members pay monthly. The owner set up both campaigns herself and wants more new students without paying to advertise to people who are already members.
Stillwater Yoga Loft: the current account, last 30 days
| Campaign | Platform | Targeting | Spend | Intro offers reported | Reported cost per intro offer |
|---|---|---|---|---|---|
| Yoga Ads | Search | Keywords: "yoga classes [town]", "beginner yoga near me", "hot yoga [town]", "stillwater yoga", "stillwater yoga timetable" | £420 | 21 | £20 |
| FB Yoga | Social | Everyone aged 18–65 within 5 miles, including site visitors and the member list | £360 | 18 | £20 |
- The studio's own booking system recorded 24 new intro offer students in the same period, not 39.
- Of the 21 search conversions, 12 came from the two brand keywords.
- Social delivery report: about 40% of spend reached people who had visited the site in the last 30 days or were on the member list; average frequency among them was 7 in the month.
- Site visitors in the last 30 days: about 2,500. Members: about 300, with emails on file.
- Target from the owner's economics: £30 per new intro offer student, judged in the booking system.
Giving each campaign one job
Start from the decisions. The studio needs to know what new students cost, to cap how much it spends following up visitors, and to stop paying to reach members at all. That gives two stage layers on each platform and nothing finer: the budget is too small to split further without starving each campaign of conversions. Exclusions do the separating; names make the separation visible in reports and analytics.
Task
- Say what is wrong with each current campaign and why its reported cost per result cannot be trusted.
- Design the prospecting and remarketing campaigns for search, including what happens to the brand keywords.
- Design the prospecting and remarketing campaigns for social, with the audience for each.
- Write the exclusions for every campaign, including how members are handled.
- Decide how to cap the remarketing layer on each platform, and what would make you change the cap.
- Write a naming convention and the four campaign names, so reports and tracking tags read the same on both platforms.
- Say what each campaign will be judged on and against which target.
Worked answer
What is wrong now. Both campaigns mix jobs. Most search conversions come from brand searches by people who already know the studio, and much of the social spend reaches visitors and members. The platforms together claim far more intro offers than the booking system recorded, so the reported CPACost of each conversion. is flattered on both.
Search. A prospecting campaign holds the generic keywords and excludes the site-visitor and member lists. A remarketing campaign holds the brand keywords, which are searched by people who already know the studio, plus the same generic keywords restricted to the site-visitor list. Members are excluded from both.
Social. A prospecting campaign targets the local area broadly, letting delivery find likely students, with visitors, members and intro offer buyers excluded. A remarketing campaign targets recent visitors who have not bought, with creative that answers the questions a nervous beginner has: what to bring, what the first class is like.
Exclusions. Members are excluded everywhere, from the uploaded email list. Intro offer buyers are excluded from both layers once they buy. Recent visitors are excluded from prospecting so the two layers never bid on the same person.
Caps. Remarketing gets a fixed budget sized to the visitor pool, not to its reported return, and the social layer gets a FrequencyAverage times a person saw an ad. cap. Change the cap when the visitor pool grows or shrinks, or after a holdout test shows what remarketing actually adds.
Names. Use objective, market, stage, descriptor in that order, with the same value as the campaign tag in analytics.
The rebuilt account
| Campaign name | Platform | Job | Budget per month | Judged on |
|---|---|---|---|---|
| sales_town_pro_intro | Search | New students from generic searches | £330 | Cost per new student in the booking system, target £30 |
| sales_town_rem_intro | Search | Brand and returning searchers | £90 | Cost per student, held stricter at £15 |
| sales_town_pro_intro | Social | New students in the local area | £300 | Cost per new student, target £30 |
| sales_town_rem_intro | Social | Visitors who did not buy, capped at 2 impressions a week | £60 | Cost per student at £15; holdout planned |
- Total budget is unchanged at £780; remarketing is about a fifth of it, set from the pool size.
- The same name on both platforms lets one report filter group each job across channels, with the platform recorded as the tag's source.
Common mistakes
- Leaving brand keywords in prospecting. They make the new-student campaign look cheap with people who already knew the studio.
- Excluding visitors from prospecting but not members from remarketing. Members then become the cheapest "conversions" in the account.
- Sizing remarketing on its reported return. Its budget should follow the pool, or frequency climbs.
- Splitting further by class type. Each extra campaign would learn from too few conversions.
- Different names on each platform. Reports cannot join, and the structure disappears in analytics.
When the rebuild makes sense, try the cross-channel campaign structure checkpoint, or return to the beginner performance marketing level.
