The Ad Account Naming and Structure Cheat Sheet

What is the offer? Who is the audience? What is the date or iteration? A name that cannot answer all three is not a name, it is a label. Format: [Offer/Product] | [Audience] | [Month-Year]. Example: LeadMagnet-AIAudit | ColdLookalike | May2026.

Pick one separator and use it everywhere. Hyphens inside a label, pipes between label groups. Never mix underscores and hyphens in the same account. Inconsistency breaks filter and search functions in every reporting tool.

Names like 'PAUSED-Retargeting-v3' create confusion when you reactivate. Status lives in the platform. The name lives forever in your historical data. Keep it clean and permanent.

Conversions campaigns optimise for conversions. Traffic campaigns optimise for clicks. Running a conversions campaign for a page with fewer than 50 conversion events per week will trigger the learning phase permanently. Use traffic or engagement objectives until your pixel has enough data.

Cold, warm, and retargeting audiences must live in separate campaigns. Mixing them in one campaign tells the algorithm to optimise across stages, which usually means it ignores cold entirely and spends on your warmest audience, inflating your results while your funnel starves.

Campaign Budget Optimisation (CBO) lets the algorithm decide where to spend across ad sets. Use CBO when you trust the algorithm and your audiences are similarly sized. Use ad set level budgets (ABO) when you are testing new audiences and need equal spend across all of them to get a fair read.

If you change the audience AND the placement in the same ad set, you do not know which change caused the result. One variable per test. Name the variable in the ad set name: [Audience-Type] | [Interest or Lookalike] | [Placement]. Example: Cold | Lookalike1pct-Leads | Feed-Only.

Two ad sets targeting overlapping audiences will compete in the same auction and skew your results. Use the Audience Overlap tool before you publish. Exclude your warm audiences from cold ad sets. Exclude purchasers from lead-gen ad sets.

Each ad set needs 50 optimisation events within 7 days to exit the learning phase. Below that number, the data is not statistically meaningful and the platform is still guessing. Do not make budget or creative decisions during the learning phase.

Advantage+ placements let Meta distribute budget across all surfaces. This often reduces cost per result at scale. Manual placements give you control and cleaner data per surface. Use manual placements when testing creative to know exactly where it performed. Use Advantage+ when you are scaling a proven creative.

Use: [Creative-Type] | [Hook-Reference] | [Version]. Example: Video | PainHook-TimeWaste | v2. The hook reference does not need to be the full copy. It needs to be enough to pull up the right creative in three seconds when you are reviewing a report at 7am.

Every time you iterate on a creative, increment the version number. v1, v2, v3. Do not duplicate an ad and give it the same name. You will lose track of which version is live and which is paused within one week.

Editing a live ad resets its delivery history and can push it back into the learning phase. Duplicate the ad, make your change in the duplicate, publish the duplicate, then pause the original. This preserves your historical performance data.

For conversion campaigns, wait until you have at least 50 conversions per ad set before comparing performance. For traffic or awareness campaigns, wait until you have at least 1,000 impressions per ad. Deciding earlier than this is reading noise, not signal.

Creative is the biggest lever in paid social. Before you narrow your audience, run three to five creative variations to the same broad audience. Find the creative that wins. Then test audiences. The order matters because winning creative often works across audiences, but a winning audience rarely saves a weak creative.

More than five ads in a single ad set means the algorithm rarely serves the weaker ads enough impressions to get a fair read. It concentrates spend on the early front-runner, which may not be the best performer at scale. Three to five ads gives you a test with enough data on each to make a call.

Do not run tests during sales events, major news cycles, or platform algorithm updates. Your baseline shifts and you cannot attribute results to your creative or audience changes. Note in your records when any external event overlapped with a test period.

A common mistake is allocating budget by ad or ad set without a top-level funnel view. Set your funnel allocation first: for example, 60 percent cold, 25 percent warm, 15 percent retargeting. Then divide within each stage. Changing this ratio is a strategic decision, not a creative decision.

Doubling a budget resets the learning phase and often spikes your cost per result temporarily. Increase budgets in increments of 20 percent or less, every two to three days, to maintain delivery stability while scaling.

Restricting your ads to specific hours of the day limits the algorithm's ability to find your audience. Only use dayparting if you have strong, account-specific data showing your conversions spike in narrow windows. Without that data, run all-day and let the algorithm optimise delivery time.

Deleted campaigns, ad sets, and ads remove their performance data from your account history. Archive instead. Archived items retain their data but do not distract your daily view. Set a naming rule for archives: add ARCHIVE | [Date] to the beginning of any campaign you are retiring.

Every person who touches the account must follow the same naming system. Write it down. A one-page document with your conventions, separator rules, and version numbering system prevents one well-meaning team member from introducing a format that breaks your entire reporting setup.

Both Meta and Google Ads support campaign labels and tags. Use them. Tag by funnel stage, offer type, and quarter. Tags let you filter performance across accounts and campaigns without relying on your naming conventions to do all the work.

Pixel events drift. Pages get redesigned, thank-you pages move, URL parameters change. A quarterly pixel audit confirms your conversion events are still firing correctly and your optimisation data is still clean. A campaign optimising on a broken pixel event is optimising for nothing.

Meta's algorithm is now strong enough that very broad targeting (age, gender, country only) often outperforms tight interest stacks for conversion campaigns with enough pixel data. Broad does not mean lazy. It means letting the algorithm do the targeting work when you have given it enough conversion signal to learn from.

Running broad match, phrase match, and exact match keywords in the same ad group makes it impossible to read search term data cleanly. Separate match types into separate ad groups, or use separate campaigns entirely. This is not optional if you want to know what is triggering your ads.

LinkedIn audiences below 300,000 members will see inflated frequency and rising CPMs quickly. If your target is narrower than that, use account-based targeting with a contact list upload rather than layered audience filters. Layered filters below the floor deplete fast and produce misleading results.

TikTok audiences experience creative fatigue significantly faster than Meta audiences. Plan to refresh your creative every seven to ten days on active campaigns rather than the two to four week window that works on Facebook and Instagram. Budget for a higher volume of raw creative assets before you launch.