Budget Allocation for Paid Social: A Practical DTC Framework

The best way to allocate a paid social budget is to work backward from the purchases you need, not forward from a percentage of revenue. Set a total from your target volume and current CPA. Then split it into four buckets: prospecting, creative testing, retargeting and brand reach. Keep the account structure simple enough that Meta’s delivery system gets the data it needs.

I’m Stefano Mazzei, a performance marketer and Co-Founder of Cosma. Most budget questions I get from DTC founders aren’t really about percentages. They’re about why money in one ad set stalls while another can’t spend fast enough. This guide covers both the top-level split and the account-level mechanics.

TL;DR

• Build the total bottom-up: target purchases multiplied by your real CPA, not “what we can afford this month.”
• Split spend by job (prospecting, testing, retargeting, brand), not by audience slice or placement.
• Consolidate. Fewer campaigns and ad sets pool conversion signals and exit the learning phase faster.
• Reserve a fixed share for creative testing, because creative is the lever that moves CPA most.
• Re-read the split every 2 to 4 weeks against blended results, not just in-platform ROAS.

Table of Contents

• Start with the number of purchases, not a percentage
• What paid social costs right now
• The four buckets I use to split a paid social budget
• A worked example: $60,000 a month
• Don’t slice the budget too thin inside the account
• Measure the whole funnel, not just Ads Manager
• Perspective: why we separate brand from performance, then merge the system
• Want a second pair of eyes on your split?
• A simple monthly review checklist
• FAQ
• Sources

Start with the number of purchases, not a percentage

The usual advice is to spend a set percentage of revenue on marketing. It’s easy to remember and it hides the real question: how many new customers do you need, and what does each one cost?

Academic work on marketing resource allocation, such as this Harvard Business School paper on allocating marketing resources, frames the manager’s task as distributing resources optimally against response, not matching a fixed ratio. That’s the logic I use with founders.

The bottom-up method takes four steps:

1. Define the growth target. For example, 1,500 new-customer orders next month.
2. Use your current blended acquisition cost. Take the CPA from the last 30 to 60 days, not the best week.
3. Multiply. Target orders times CPA gives the media investment the plan implies.
4. Stress-test against margin. If that figure breaks your contribution margin, either the target or the CPA assumption has to change. Raising the budget won’t fix it.

Here is an illustrative example. These are made-up round numbers, not client data. A brand wants 1,000 new customers a month and its blended new-customer CPA is $55. The implied acquisition spend is $55,000. If a 40% gross margin and a $70 average order value mean each first order only breaks even near $28 of ad cost, the plan needs either a higher AOV, repeat purchases that carry the economics, or a smaller target. That conversation is more useful than debating whether paid social should get 30% or 40% of the total.

What paid social costs right now

Costs keep moving, which is another reason to recompute the plan regularly. According to eMarketer’s 2025 forecast, CPM prices are rising year over year on every major US social network, although AI-driven ad products are helping advertisers justify the higher costs through better outcomes.

On lead cost, Search Engine Land reported WordStream LocaliQ benchmarks showing Facebook’s average cost per lead up 21% year over year to $27.66, against $70.11 on Google. Those are lead-gen figures for small businesses, not ecommerce purchase costs, so treat them as a direction of travel, not a CPA target for your store.

For purchase-based benchmarks, I keep a separate breakdown of Meta Ads ecommerce benchmarks. Use those to sanity-check your own numbers, then trust your own account data over any average.

The four buckets I use to split a paid social budget

I don’t allocate by audience, by placement or by product line first. I allocate by the job the dollars do. The shares below are starting points I’d adjust to your account, not rules.

These ranges are my working judgment from running accounts, not an industry statistic. A brand with a small catalog and a tight CPA target will skew to prospecting. A brand with a high price point and a long consideration window will need more brand and retargeting.

Prospecting gets the largest share

Prospecting is where growth comes from, so it takes the biggest slice. The mistake I see is starving it to feed retargeting, because retargeting reports a great ROAS. That ROAS is flattered by people who were already coming back. Cut prospecting and retargeting pools dry up within weeks.

On targeting, I default to broad audiences for most ecommerce accounts and let creative do the segmentation. I explain the reasoning in why broad targeting should be your default.

Creative testing gets its own protected line

If testing spend comes out of whatever is left over, it disappears the first month results dip. That is precisely when you need new concepts. I ring-fence a testing share and judge it separately, because test ads will often look worse than scaled winners. They’re supposed to. The point is to find the 1 in 5 or 1 in 10 that beats your control.

Creative variety matters more than most budget conversations admit. If you need formats to test, start with four creative-first Facebook ad formats that scale.

Retargeting is a cap, not a goal

Retargeting should be sized to the warm audience you actually have. If you’re pushing frequency past what your site traffic can support, you’re paying to show the same ad to the same people. I’d rather cap retargeting at a modest share and let the rest work upstream.

Brand and reach: the long game

This is the bucket founders cut first. The research behind that instinct is worth knowing. Binet and Field’s work suggests the strongest growth comes from putting around 60% of budget into brand building and 40% into activation. But the researchers themselves treat it as nuanced. As Field says in The Irish Times, it was never meant as a hard and fast rule.

That 60/40 comes from large, mostly established brands. A growth-stage DTC brand that needs payback within weeks can’t copy it. Still, Nielsen’s analysis of marketing mix decisions describes a pendulum that has swung hard from brand toward performance. Skipping brand entirely has a cost, and it shows up later as rising CPAs.

There is a separate 60/40 you may see cited. Research by Analytic Partners commissioned by Google proposed putting 60% into digital and 40% into traditional for omnichannel retailers. It’s a different split answering a different question (channel type, not brand versus performance), and it was written for retailers with physical stores. Don’t mix the two up.

A worked example: $60,000 a month

Here is one illustrative split for a hypothetical DTC brand spending $60,000 a month on Meta. The numbers are an example of the method, not a benchmark.

To run it, I would use:

• One or two prospecting campaigns with broad targeting and your best-performing creative.
• One testing campaign where new concepts get a fair, equal chance at spend.
• One retargeting campaign sized to your site traffic.
• A small reach or video campaign, judged on branded search and new-customer share, not on last-click ROAS.

If prospecting CPA drifts 20% above target for two weeks, I don’t immediately cut the budget. I check creative fatigue first, then offer, then landing page. Cutting spend on a fatigued account just makes it a smaller fatigued account.

Don’t slice the budget too thin inside the account

The top-level split is only half of it. How you structure ad sets decides whether the money actually works.

Meta’s delivery system needs conversion volume to stabilize. According to Meta’s own guidance for smaller advertisers, ad sets can stay in the learning phase when the system hasn’t found the best delivery path and hasn’t reached roughly 50 conversions a week. That raises spend in the learning phase and pushes up cost per result.

That’s the trap of hyper-segmentation. Ten ad sets at $50 a day each may never exit learning. Two ad sets at $250 a day often do. My rule is to consolidate to pool signals, and I wouldn’t obsess over the 50-a-week threshold at the ad set level on a small account. The learning phase is a stability indicator, not a score.

For how to set this up, see my guides on Meta Ads campaign structure with ASC and Pixel plus CAPI and, for a broader take on layouts, this walkthrough of Meta ads campaign structure.

Campaign-level budgets and Advantage+

Advantage+ campaign budget lets Meta shift spend between ad sets in real time. For most DTC brands I prefer that to manual ad set budgets, because it follows performance faster than a human checking the dashboard each morning. The tradeoff is control: the system may favor one ad set heavily. That’s fine if all ad sets contain creative you’d happily scale.

Advantage+ Sales campaigns work best when you have enough purchase volume. I lay out the threshold in when DTC brands should use Advantage+ Sales. If you’re below it, keep structure tighter and put more weight on testing.

Scaling rules that protect efficiency

• Raise budgets in steps rather than doubling overnight. I usually move 20 to 30% at a time and wait for results to settle.
• Add budget to campaigns that have proven CPA, not to the ones with a good single day.
• Duplicate winners only when the existing campaign is capped by audience or creative volume, not by habit.

If you want a two-week routine to apply this, my 14-day Meta campaign optimization playbook walks through it.

Measure the whole funnel, not just Ads Manager

In-platform ROAS only sees what Meta can attribute. It misses the influence of email, organic social, creators, retail and word of mouth, and it tends to overcredit retargeting. If you set budgets purely on platform numbers, you’ll over-invest in the bottom of the funnel and under-invest in what fills it.

I check four things together every week:

1. Blended MER (total revenue divided by total ad spend).
2. New-customer CPA or acquisition cost.
3. Platform-reported ROAS and CPA, as a directional signal.
4. Branded search and direct traffic, as a proxy for brand lift.

When those disagree, I trust the blended view. For more on how funnel stages work together, this overview of marketing funnels and lead generation is a reasonable companion.

Perspective: why we separate brand from performance, then merge the system

One mistake I see constantly is asking a single ad to do two jobs. A founder wants an ad that builds the brand and also drives a purchase this afternoon. It does neither well. Brand ads and performance ads have different goals, timescales and metrics, and judging both on same-day ROAS punishes the brand ad and flatters the retargeting ad.

So we separate the jobs in the budget, but connect them in the system. The same customer research feeds both: the language real buyers use, the objections they raise and the moments that make them buy. Creative testing then tells us which of those angles deserves more spend, and the budget follows the evidence instead of a template.

The second thing we hold to is simplicity. Most accounts I audit have too many ad sets, too little spend per ad set and too many manual overrides. Cleaning that up often lowers CPA before any new creative ships. Budget allocation is rarely about clever splits. It’s about giving a good system enough data and good ads to work with.

— Stefano Mazzei

Want a second pair of eyes on your split?

If you’re spending on Meta and aren’t sure your allocation matches your economics, my team at Cosma can review your account structure, creative testing and spend split with you. You can see what Cosma has done for other brands, check Cosma’s pricing, or book a call with my team at Cosma.

A simple monthly review checklist

• Recompute target purchases and implied spend from your latest CPA.
• Check each bucket’s share against the plan and note why any drifted.
• Retire fatigued creative and fund the next round of tests.
• Compare platform ROAS with blended MER and new-customer share.
• Change one major variable at a time so you can read the result.

FAQ

How much of my marketing budget should go to paid social?

There’s no universal figure. It depends on your margins, your other channels and how many new customers you need. I start from target purchases times CPA, then check that the total fits contribution margin. Percent-of-revenue rules are a sanity check at best.

What is the 60/40 rule in advertising?

It comes from Binet and Field’s research, which found the strongest long-term growth when around 60% of budget goes to brand building and 40% to activation. Field has said it was never a hard rule. For growth-stage DTC brands, treat it as a reference point and set your mix from your own payback window.

How much should I spend on creative testing?

I usually reserve 10 to 20% of paid social spend for testing, and I protect it when results dip. Smaller accounts can run fewer concurrent tests but should keep testing continuous. If testing never gets its own line, it becomes the first thing cut.

Should I use campaign-level budgets or ad set budgets?

For most DTC brands, campaign-level budgets with Advantage+ work better because Meta shifts spend toward what’s performing in real time. Use ad set budgets when you need to force spend onto a specific test. Either way, keep the number of ad sets low so each gets enough conversions.

Sources

• Allocating Marketing Resources (Harvard Business School)
• US Social Ad CPMs Forecast 2025 (eMarketer)
• Facebook ad costs jump 21% in 2025, but still beat Google (Search Engine Land)
• Binet Rules (ThinkTV)
• How best to balance the long and the short of marketing (The Irish Times)
• Are you investing in performance marketing for the right reasons? (Nielsen)
• The new 60/40 model (Think with Google)
• Meta guidance for small and mid-size advertisers (Meta for Business, via WeChat)

Recommended

• Creative First DTC Facebook Ads From Cosma: 4 Formats That Scale
• $9K to $61K: Creative Driven DTC Facebook Ads Case Study
• Creative First Facebook Ad Testing for DTC: The 50 Event Rule
• Hook, Proof, Offer: Facebook Ad Examples for DTC Brands