Key takeaways
- Three levels: campaign, ad group, ad. LinkedIn renamed them, older guides still call them campaign group, campaign and ad.
- One campaign per funnel stage. That way you immediately see how budget spreads across the funnel.
- One ad group carries exactly one audience and one objective. Otherwise nothing is attributable afterward.
- At least five ads per ad group, otherwise the creative wears out too fast.
- The most common structural mistake is cutting too fine: many ad groups, none with enough data.
Campaign structure isn't busywork for tidy people. It's the precondition for learning anything at all. An account where three audiences, two offers and four formats sit in the same campaign produces one number at month end that no decision can be derived from. And because rebuilding always costs performance, structure is one of the few decisions you're better off making before the first dollar. How to build your first campaign in practice is covered under how to set up LinkedIn Ads. This page is about the scaffolding above it.
The three levels, and what they're really called
Campaign Manager knows exactly three levels:
- Campaign. The outer bracket, usually one funnel stage. It barely controls anything, it organizes.
- Ad group. The working level. Objective, audience, format, daily budget and bid all sit here. Almost every decision that drives your cost is made at this level.
- Ad. Creative, copy and call-to-action, meaning what your audience actually sees.
There is a trap here that most guides online still get wrong: LinkedIn renamed these levels. If you know Campaign Manager from earlier years, or you are reading an older guide, you will find different names for the same things.
| Former name | Current name | What sits at this level |
|---|---|---|
| Campaign group | Campaign | The bracket, usually one funnel stage |
| Campaign | Ad group | Objective, audience, format, daily budget, bid |
| Ad | Ad | Creative, copy, call-to-action |
Two practical consequences. When an older text tells you "one campaign per audience", today that means "one ad group per audience". The rule still holds, only the level's name changed. And anyone coming from Google Ads or Meta now finds their way around faster than before: the level called an ad group there is called the same thing on LinkedIn now.
One campaign per funnel stage
The simplest structure that holds follows the funnel, not the offer and not the format. Three campaigns cover the normal case:
- Cold audience. People who don't know you yet. The goal is attention and trust, usually through video and Thought Leader Ads. Soft calls to action, no selling.
- Retargeting. Everyone who has already reacted. More concrete formats and sharper claims run here, because the groundwork is done. Details under retargeting.
- Harvest. A lead form or landing page, aimed at the audiences with the highest buying intent.
The practical value of this split is the budget view: you open the account and see in one row how much money is working in which stage. That distribution is the single most common flaw in B2B accounts, because almost everything ends up in harvest and the cold stage starves. Why that gets expensive is explained by the shadow funnel, the split itself is covered under budget.
One technical detail: if you give the campaign its own objective, you restrict which formats can run inside it. So leave the campaign objective open when you want to test several formats against each other within one stage.
That is exactly what we build with B2B teams: not an agency mandate, but your own system. The coaching walks your team through it. The overview shows how we work.
What belongs in one ad group, and what doesn't
The rule is short: one ad group, one audience, one objective. Anything else breaks your analysis.
The moment two different audiences sit in the same ad group, LinkedIn decides which one gets more delivery, and you only see the blended result. Run them separately and after two weeks you can say which delivers cheaper and shift budget accordingly. How to cut audiences sensibly is in the targeting guide.
The same goes for the objective. One objective per ad group, because the objective determines what LinkedIn optimizes delivery for and which formats and bid types are even available. A campaign meant to produce both reach and leads produces both badly.
On budget: use a daily budget, not a lifetime budget. LinkedIn spreads a lifetime budget across the period itself, and you lose control over daily spend. But daily spend is precisely your control instrument for the bid, see bidding strategy.
Save every finished audience as a template before you launch the ad group. It costs one click and saves you rebuilding it from memory later, when the same audience is needed in a second stage. Rebuilt audiences almost always differ slightly, and then you're comparing two campaigns that aren't comparable.
How many ads belong in one ad group
At least five. The reason isn't statistics, it's frequency: with only a handful of creatives running, the same person sees the same image too often, click-through rate drops and cost per click rises. Five ads give LinkedIn enough material to rotate.
Your budget sets the upper limit. Every ad needs enough delivery to be judgeable, otherwise you have twelve creatives and no reliable number for any of them. How much budget an ad group needs for that is under budget. Which creatives belong in the rotation at all is covered in the creatives guide.
The campaign optimization playbook
Structure is the foundation, the routine on top of it decides the outcome. The benchmark traffic light, the order of levers and the fixed weekly rhythm come bundled in one playbook.
Names that still read six months from now
Campaign names like "test 3 new final" are why accounts become unusable. A naming scheme costs ten minutes of agreement once and carries for the entire run. What works is a fixed order of five fields, separated by a consistent character:
- Region, if you run several markets.
- Funnel stage, meaning cold, retargeting or harvest.
- Audience, named so a colleague understands it without asking.
- Offer, meaning the lead magnet, the webinar or the demo.
- Format, so you can compare per format.
Which five fields you choose is secondary. What matters is that the order never changes and everyone on the team uses the same scheme. A date doesn't belong in the name, it's already in the data and goes stale in the name immediately.
The same discipline is needed for retargeting audiences, and there it matters even more. An audience called "website 90 days" tells nobody two months later who is inside it. Put the source first and the condition with it, for example "website: pricing page, 90 days" or "video: watched 75 percent, 90 days". Since these audiences take some time to fill, it pays to create all foreseeable ones right at the start. That requires a correctly installed Insight Tag.
The mistake that collapses any structure
Almost every structural problem we see in accounts is the same problem: cut too fine. The logic behind it sounds reasonable, since every extra separation promises an extra insight. In practice the opposite happens. Budget spreads across so many campaigns that none of them collects enough data to be judgeable. At month end you have twelve campaigns with a handful of clicks each and not one reliable statement.
So the test question before any new separation isn't "is this interesting?", it's "will this ad group get enough budget to reach the data threshold within the test window?". Where that threshold sits is under budget. If the answer is no, the separation doesn't belong in the account, it belongs on the list for later.
This applies especially at the start. A new account doesn't need sophisticated architecture, it needs one that works: one campaign per stage, and inside it a few ad groups with clearly separated audiences. You branch out later, once the budget carries it.
When to rebuild, and when to leave it alone
A rebuild is never free. Every structural change throws the affected ad group back into its learning phase, and during that time the numbers are worse and not comparable. So there are two clear triggers:
Rebuild when you can no longer answer a question about your own account. If you can't say which audience delivers cheaper leads or which format carries the cold stage, a separation is missing. Same goes if a stage doesn't structurally exist, for example because all budget sits in harvest.
Don't rebuild because of day-to-day fluctuation. An expensive Tuesday is not a structural problem. Neither is one weak ad, you swap that out without touching the campaign. And never change several things at once, or you won't know afterward which change did the work. Which levers belong in which order is covered under campaign optimization. Log every change with a date: that single line is the difference between an account you learn from and one you guess in.