Once you can build a campaign, the harder question is what shape the whole thing should take. Most brands never decide this on purpose. They boost a post, then another, then set up a retargeting campaign because someone said to, and end up with a pile of overlapping ad sets that nobody can read. The result is spend without a structure, and structure is most of what separates paid social that compounds from paid social that just recurs.
This guide lays out the handful of strategies brands actually use (prospecting versus retargeting, the full-funnel model, always-on versus burst, creative testing, and budget allocation) and for each one, what it is, when it fits, and how to set it up. Every worked number here is labelled illustrative, because the honest ones depend entirely on your business.
Prospecting versus retargeting
The most fundamental split in paid social is between reaching people who do not know you and reaching people who already do.
Prospecting targets a cold audience: core, lookalike or broad, as covered in how to pick your audience for ads. This is where you find new people, where most of your budget belongs, and where your creative is genuinely being tested, because a cold viewer extends you none of the goodwill your existing audience does.
Retargeting targets people who already engaged with you or visited your site: warm custom audiences. It is smaller, cheaper per result, and reliably reports excellent returns.
That last fact is a trap, and it is worth stating plainly because it is the single most common mistake in the whole discipline. Retargeting looks efficient partly because some of the people in it were going to convert anyway, so the ad takes credit for a sale it did not cause. Scaling budget onto retargeting on the strength of its reported return is therefore usually a misallocation: you are paying to reach people who were already yours. Reading ad results without fooling yourself explains the incrementality problem underneath this in full. The rule of thumb: prospecting is the engine, retargeting is the catch-net, and the catch-net should not get the engine's budget.
The full-funnel model
Prospecting and retargeting are the two ends of a fuller picture: the funnel, usually described as cold, warm and hot, or in the vocabulary of top, middle and bottom of funnel.
- Cold: people who do not know you. The job is a first impression. Creative introduces; the objective is awareness or engagement; the metric is reach and attention, never sales.
- Warm: people who have engaged once. The job is to deepen interest. Creative gives a reason to look closer; the objective is traffic or leads.
- Hot: people who have shown real intent, like visiting a product page. The job is to close. Creative addresses the specific hesitation; the objective is the conversion itself.
The model fits any brand with a considered purchase, where nobody buys on the first exposure and the point is to move people along. It fits less well for pure impulse buys, where the funnel collapses into a single step. The reason to build this way is readability: when each stage is its own campaign with its own job, objective and metric, you can see which stage is actually the constraint. A brand whose cold reach is fine but whose hot conversion is weak has a very different problem from one that cannot get anyone into the funnel at all, and a full-funnel structure is what lets you tell those two apart instead of guessing.
Setting it up means separating the stages so their results stay clean: one campaign per job, distinct audiences with the warm pool excluded from the cold one so they do not bid against each other, and different creative for each because an introduction and a closing argument are not the same ad.
Always-on versus burst
The other structural choice is about time: do you run continuously, or in concentrated bursts.
Always-on is a steady, continuous presence: prospecting and retargeting running all the time at a sustainable level. It fits brands with an ongoing offer and a reason to be in front of new people every week. Its advantage is that delivery stays out of the noisy learning phase, because ad sets that run continuously settle and stay settled. Its risk is drift: an always-on campaign left unattended quietly wears out its creative and its audience.
Burst is concentrated spend around a moment: a launch, a sale, a season. It fits when there is a genuine event to build around and nothing much to say in between. Its advantage is impact; its cost is that every burst pays the learning-phase tax again, because ad sets that stop and restart re-enter learning each time.
Most mature brands run a hybrid: an always-on prospecting and retargeting base that keeps delivery settled and the pipeline fed, with bursts layered on top for real moments. The mistake is treating every ordinary week as a burst, which means you are permanently paying the restart cost and never letting anything stabilise.
Creative testing done cleanly
Paid social is the fastest, cleanest way to learn what resonates, but only if the test is built so the answer means something. A messy test is worse than no test, because it produces a confident conclusion that is wrong.
A clean creative test follows a few rules:
- Change one thing at a time. If two ads differ in hook, image and caption, a difference in results tells you nothing about which change caused it. Vary one variable per comparison. Testing hooks specifically is covered in testing hook variants.
- Keep everything else constant. Same audience, same placement, same objective, same budget, same window. The confounds only cancel when the conditions match.
- Give each variant enough volume to be readable. A variant that gets a few hundred impressions produces a number indistinguishable from noise. Decide in advance what counts as enough.
- Decide the winning metric before you look. Otherwise you will pick the metric that makes the result you already wanted look correct.
The reason the same-conditions rule works is that the platform's reporting is trustworthy for relative comparison within an ad set even where its absolute claims are inflated: creative A beating creative B by a clear margin in the same ad set is solid evidence, which is exactly why creative testing is where paid earns its keep.
Budget allocation across the funnel
How you divide money across the funnel is a real decision, not a formula, and it depends on your business, your margins and where your constraint actually is. There is no universal split, and anyone who quotes you one as a rule is inventing it.
What you can hold onto are principles. The bulk of budget usually belongs in prospecting, because that is where new demand comes from and where the creative is genuinely tested; retargeting takes a smaller share because its job is to catch, not to generate. A ring-fenced testing slice (a small, deliberate amount always running creative tests) is what keeps the pipeline of proven creative full, and it pays for itself by finding the winners you then scale. And the split should follow your evidence, shifting towards whichever stage your own results show is the constraint, rather than sitting fixed because you set it once.
Budget splits explained works through the trade-offs in more depth, including why the intuitive move (pouring budget into the efficient-looking retargeting) is usually the wrong one.
Where Acumin fits
These strategies live in the ad manager, but the raw material for them lives upstream, and that is where Acumin works.
Studio makes the creative each stage needs. Full-funnel done properly needs several distinct pieces, not one ad reused, so a tool that makes variants quickly is what makes clean testing affordable. Snapshot reads your public organic numbers, which is how you find the over-performing pieces worth promoting into a prospecting campaign in the first place. And the Content Strategy cockpit holds the plan the whole paid structure serves, so your always-on base and your bursts are funding a strategy rather than reacting to whatever felt urgent that week.
The honest limit, as the product states it: reading Meta ad results back automatically is pending Meta's app review. Until it lands, Acumin prepares and drafts while you read results and run campaigns in Meta's own tools. Acumin drafts; you send.
Do this today
Take whatever paid you are running now and sort every ad set into one box: prospecting, retargeting, or testing. If you have no testing box, that is the first thing to fix, because it is the box that keeps the other two supplied. If retargeting is taking the largest share, that is the second, because it almost certainly should not. You do not need a new campaign to start: you need to be able to say, for every pound you spend, which job it is doing.