Skip to content
Faisal Mahmud TusherDigital marketing, SEO & analytics
with Faisal Tusher

Meta Ads › Intermediate

Consolidating Fragmented Ad Sets for a Skincare Brand

You take an invented skincare brand whose many small ad sets never leave the learning phase, work out why from its own figures, and rebuild the account into a few ad sets that can learn. The exercise practises sizing ad sets to the exit threshold, choosing between campaign and ad set budgets, and stating what should change without promising results.
ExerciseIntermediate levelModule 1: Structure, budgets and the learning phase5 minUpdated

Why this matters

Fragmented accounts rarely look broken from the inside: each ad set was added for a sensible reason, each one delivers, and the account never settles. The learning phase guide explains why: the exit condition is a count of optimisation events per ad set, so splitting one budget across more ad sets makes the threshold harder for all of them. Here you apply that arithmetic to a whole account.

The account summary

Dewpoint Skincare is an invented brand selling cleansers and moisturisers for sensitive skin. All its campaigns optimise for purchases, and the founder wants the total budget held where it is.

Illustrative example — not a real result

Dewpoint Skincare: ad sets over the last 28 days

Daily budget across the account: £200. Target cost per purchase (CPA): £35. For the threshold, use the figure the learning phase guide gives from Meta's documentation: roughly 50 optimisation events in the 7 days after the last significant edit.

CampaignAd setDaily budgetSpendPurchasesPurchases per weekCost per purchase7-day frequency
Prospecting – InterestsSkincare enthusiasts£25 (ad set)£700164£43.751.7
Prospecting – InterestsClean beauty£25 (ad set)£700143.5£50.001.9
Prospecting – InterestsYoga and wellness£20 (ad set)£56092.25£62.221.6
Prospecting – InterestsSensitive skin£20 (ad set)£560153.75£37.331.8
Prospecting – LookalikesPurchaser lookalike 1%£80 (campaign)£1,120348.5£32.942.4
Prospecting – LookalikesEmail list lookalike 1%(shared)£700184.5£38.892.1
Prospecting – LookalikesPurchaser lookalike 3%(shared)£42092.25£46.671.8
RetargetingSite visitors, 30 days£15 (ad set)£420143.5£30.004.1
RetargetingAdded to cart, 14 days£15 (ad set)£420112.75£38.186.2
Total£200£5,60014035£40.00

Notes from the account:

  • Every ad set shows learning limited.
  • Initiate checkout fired 760 times in the same 28 days, about 190 a week.
  • The audience overlap tool shows heavy overlap between Skincare enthusiasts and Clean beauty, and the 3% purchaser lookalike contains the 1% one. The cart audience sits inside the site visitor audience.
  • Neither prospecting campaign excludes existing customers.
  • Change history shows interest ad set budgets edited two or three times a week.

Deciding what to merge

Start with the threshold, not the audiences. Compare each ad set's weekly events with the exit figure, then ask what the account could reach if its events were pooled. If even the pooled total falls short, the optimisation event has to change as well. A separate ad set earns its place when spend must be fixed for a reason outside the auction, or when the audience is the point, as with retargeting; a split made to see which interest wins belongs in a test. The Advantage+ audiences guide shows how a Lookalike Audience can live on as a suggestion inside a broad ad set, and the campaign versus ad set budgets comparison sets out which budget type suits which ad sets.

Task

Answer before reading on.

  1. Compare each ad set's weekly purchases with the threshold, then say whether one pooled prospecting ad set on purchases could exit learning.
  2. Decide which ad sets to merge, which to keep and which to stop, with a reason for each.
  3. Choose the optimisation event for prospecting and say when you would review it.
  4. Choose a budget type and daily figure for each new campaign, within the current total.
  5. List the exclusions the new structure needs.
  6. Say how you would switch to the new structure and what you expect to change, as direction, not figures.

Worked answer

The arithmetic first.

Illustrative example — not a real result

Dewpoint's threshold arithmetic and new structure

  • Best ad set: 8.5 purchases a week against roughly 50 needed. No ad set comes close.
  • Pooled prospecting: 54 + 61 = 115 purchases in 28 days, about 29 a week. Even one prospecting ad set on purchases would stay learning limited.
  • Pooled checkouts: about 190 a week across the account. Prospecting at £180 a day would carry most of them, comfortably above the threshold even if split across two ad sets.

New structure:

CampaignAd setBudgetOptimisation event
ProspectingAdvantage+ audience, both 1% lookalikes entered as suggestions, customers excludedCampaign budget £180 a dayInitiate checkout
ProspectingPurchaser lookalike 1%, original audience options, customers excluded(shared)Initiate checkout
RetargetingSite visitors, 30 days, recent purchasers excludedAd set budget £20 a dayPurchase

Merges. The interest ad sets, the email-list lookalike and the wider purchaser lookalike go: they overlap with each other, so they bid against themselves, and none needs fixed spend. Their audiences survive as suggestions in the broad ad set. The narrow purchaser lookalike, the strongest performer, stays as a second ad set so delivery has a defined starting point while the broad one learns.

Event. Structure alone cannot clear learning, so prospecting moves to initiate checkout for now, trading precision for learnability. Review it once weekly prospecting purchases approach the threshold, and judge the campaign on cost per purchase throughout.

Budgets. The prospecting ad sets share a goal and an event, so campaign budget lets delivery follow the cheaper source and stops manual edits resetting each one. Expect uneven spend; that is the tool working, not a test. Retargeting has its own economics and an over-served audience, so it gets its own, smaller ad set budget, merged into one ad set with the cart audience inside it.

Exclusions. Customers out of both prospecting ad sets, as a control rather than a suggestion. Recent purchasers out of retargeting.

Switching. No old ad set ever exited learning, so there is nothing stable to protect. Launch the new structure, pause the old one the same day and leave it alone through the first week.

Expected change. Fewer ad sets in learning, then prospecting exiting it; a noisy first week; lower Frequency in retargeting; and less tidy per-ad-set reporting, because spend is no longer equal.

Common mistakes

  • Merging but keeping the purchase event. The pooled total is still below the threshold.
  • Putting retargeting under the prospecting campaign budget. The cheap results pull spend into an audience that is already saturated.
  • Keeping interest ad sets to find out which wins. That is a question for a test, not a structure.

When you are done, test yourself with the structure and learning phase checkpoint, or return to the intermediate Meta Ads level.

Pages this lesson applies

Terms used on this page

Also related

Related