How we 4x’d paid social spend at expanding ROAS for a DTC outdoor brand.
6-month engagement (ongoing)·Meta (Facebook & Instagram) + TikTok
A Scandinavian-inspired DTC outdoor lifestyle brand was running $32K/month on Meta at a reported 2.7x ROAS and stuck on a six-month plateau. Their previous agency was rotating two new ads a week and blaming iOS for everything else. We rebuilt Meta CAPI, consolidated the account into broad Advantage+ Shopping campaigns, launched TikTok from zero, and stood up a creative velocity engine producing 12–15 new ads per week. In six months, spend scaled 4x and ROAS expanded to 5.2x because the tracking was finally honest and the creative pipeline finally matched what the algorithm rewards.
5.2x
Blended ROAS
up from 2.7x reported
4x
Spend scaled
$32K → $128K / mo
7.7x
Revenue growth
on paid social
We had two Meta agencies before SMB Circus and both told us creative wasn't the problem. SMB Circus shipped 60 ads in the first month and we tripled by week six.
What we walked into
The brand had been running paid social for two years. Revenue was real, the product was excellent, customer reviews were strong. Paid social was the bottleneck, not the demand.
Five findings from the first-week audit:
Meta CAPI was implemented but broken.
Server-side events were firing, but dataset deduplication was misconfigured, event match quality was 4.2 out of 10, and roughly 38% of purchase events had no user identifiers passing through. Meta was running ad delivery on a half-blind signal. True ROAS was almost certainly higher than what the dashboard showed.
42 active ad sets across 11 campaigns.
Lookalike 1%, Lookalike 3%, Lookalike 1–5%, eight separate interest stacks, three retargeting buckets at different windows, a "test" campaign that had been running for nine months. The account was over-segmented to the point that no audience had enough conversion volume for the algorithm to learn. Most ad sets had fewer than 5 conversions per week.
Two new ads per week, all static images.
No video. No UGC. No Reels-native content. The creative was beautifully art-directed lifestyle photography that would have crushed it in 2019 and was being ignored by the 2026 algorithm.
Zero TikTok presence.
The previous agency had pitched TikTok and been declined because "our customer isn't on TikTok." The brand's customer was 28–45, outdoor enthusiast, mostly urban-adjacent — exactly the demo that's now spending two hours a day on TikTok.
Landing pages were the brand's homepage.
Every ad routed to the same hero. No campaign-specific landing pages, no product-specific PDPs being tested as ad destinations, no mobile-optimized variants. Add-to-cart rate from paid traffic was 4.1%, less than half of what the same product was doing from organic traffic.
The previous agency wasn’t incompetent. They were running a 2021 playbook in 2026.
What we did
Audience targeting is commoditized. Creative volume and signal quality are the only variables left.
Rebuild the measurement signal
Before scaling, fix what Meta sees. Algorithm performance is a function of signal quality. Half-broken CAPI means half-broken delivery.
- Server-side GTM via Stape for first-party event capture that survives iOS, ITP, and third-party cookie deprecation.
- Meta CAPI rebuild. Full event taxonomy, dataset deduplication via event_id, proper customer information parameters (hashed email, phone, FBP, FBC, IP, user agent). Event Match Quality moved from 4.2 to 8.6 in three weeks.
- TikTok Events API implemented in parallel ahead of channel launch.
- GA4 reconciliation. Cross-validated Meta-reported conversions against GA4 and Shopify. Found that Meta had been under-attributing by roughly 28% on a 7-day click, 1-day view window.
Honest framing: The apparent ROAS jump from 2.7x reported to a true ~3.4x on the same spend was notperformance work. It was measurement work. We told the client this explicitly so they didn’t expect a 28% lift to compound on top of every subsequent gain.
Consolidate the account structure
Modern Meta delivery (Advantage+ Shopping, Advantage+ Audiences, broad targeting + creative testing) rewards consolidated structure with enough conversion volume per campaign for the algorithm to learn. We rebuilt the account from 11 campaigns down to 4.
| Campaign | Audience | Bid Strategy | Creative | % Spend |
|---|---|---|---|---|
| Advantage+ Shopping (Acquisition) | Broad, with named existing-customer exclusions | Highest Volume, ROAS floor | 60% of weekly creative | 65% |
| Advantage+ Audience (Top-of-funnel test) | Broad with audience signals from CRM customer list | Highest Volume | 20% of weekly creative | 15% |
| Retargeting | 7-day, 14-day, 30-day site visitors + add-to-cart | Cost cap | Sequenced (product, social proof, urgency) | 12% |
| Existing customer / cross-sell | Custom audience of 60-day purchasers | Highest Volume | Catalog ads + new product launches | 8% |
That’s four campaigns. From 11 down to 4. Each campaign now has enough weekly conversion volume for the delivery algorithm to optimize properly.
Specifically killed:
- All Lookalike audiences except as a fallback test pool. Advantage+ Audiences outperforms manual lookalikes on every account we’ve tested since Meta rolled out the new structure in late 2024.
- All interest-stack ad sets. The algorithm finds these audiences better than humans do.
- The nine-month “test” campaign that had spent $14K with no learnings documented.
TikTok launch
Argued and won the case for TikTok. The customer was on the platform; the previous agency was wrong.
Launch architecture:
- Smart+ Campaignsas the primary acquisition structure — TikTok’s Advantage+ equivalent, broad targeting with algorithmic delivery.
- Spark Ads from real creator content — not branded studio content reformatted for vertical. Authentic creator POV, product in context, 15–30 seconds.
- Native vertical creative only.No repurposed Meta ads. Every TikTok asset was shot or cut specifically for the platform’s native format.
- Lower CPMs, faster fatigue. TikTok delivered CPMs roughly 40% lower than Meta in this category, but creative fatigue hit faster — which reinforced the need for the creative velocity engine.
The creative velocity engine
This is the load-bearing piece of the engagement. In 2026, audience targeting is commoditized across every major paid platform. Creative volume and quality are the only variables left that move performance materially. We treated it that way.
The cadence:
- 12–15 new ads per week across Meta and TikTok, mix of in-house production and creator-sourced UGC.
- Hook testing framework. Every ad concept tested with 3–4 hook variants (first 3 seconds). Hook determines 80% of view-through and click-through. Hooks that win get the body variations; hooks that lose get killed immediately.
- Statistical significance thresholds. Ads need 300+ impressions and at least 48 hours before being called. No gut-feel kills, no premature scaling.
- Iteration cadence. Winners got 3–5 iterations within the same week: new hooks, new CTAs, new aspect ratios, new overlays. Losers were documented for pattern analysis and archived.
- UGC and creator pipeline. Rolling roster of 6–8 creators producing 3–4 assets each per month. Briefs were specific (hook, key message, CTA, visual style) but not scripted — authentic always beats polished.
By month six the brand had a library of 280+ tested ads, with 34 documented winners that were rotating in production at any given time.
Landing page CRO
Scaling spend without fixing the destination is malpractice. The brand’s homepage was an ad-traffic graveyard.
- Three campaign-specific landing pages built on Replo, each message-matched to the top-spending ad concept for that campaign.
- Mobile-first design. 78% of paid social traffic was mobile. Every element designed thumb-first: sticky add-to-cart, swipeable social proof, compressed hero.
- Message match. Ad copy, headline, hero image, and social proof on each landing page reflected the specific creative that drove the traffic. No generic catch-all pages.
- Add-to-cart rate from paid traffic moved from 4.1% to 6.8% over the engagement. That single shift is responsible for a meaningful portion of the ROAS expansion.
Scale with creative-first scaling rules
Scaling paid social without a rule set is how brands blow up their CPA in a single week. We ran every budget increase through four non-negotiable gates.
- 1Only scale ad sets where the top-performing creative had at least 3 fresh variants tested behind it.
- 2Increase budgets in 20% increments, no more than every 72 hours, to protect the learning phase.
- 3Kill any ad set where ROAS dropped below the campaign floor for 5 consecutive days, regardless of historical performance.
- 4New product launches got their own consolidated test budget — never mixed into a scaled campaign.
Spend ramped from $32K/month at month 0 to $58K (month 2) to $94K (month 4) to $128K (month 6). ROAS held at the campaign level and expanded at the blended level because TikTok came online efficiently and creative kept improving.
The math
| Metric | Baseline | Month 6 | Δ |
|---|---|---|---|
| Paid social spend | $32K/mo | $128K/mo | 4x |
| Reported ROAS | 2.7x | 5.2x | +93% |
| True ROAS (CAPI-corrected baseline) | ~3.4x | 5.2x | +53% |
| Paid social revenue | ~$86K/mo | ~$666K/mo | 7.7x |
| Average order value | $108 | $112 | +4% |
| Add-to-cart rate (paid traffic) | 4.1% | 6.8% | +66% |
| Active ad concepts in rotation | 8 | 34 | +325% |
| TikTok share of paid social revenue | 0% | 31% | New channel |
| Blended CAC (across all paid) | $74 | $52 | −30% |
On why ROAS expanded rather than compressed at scale.
This is the question every experienced operator will ask. Three drivers, in order of impact:
- 1The starting baseline was under-reported. Roughly 70% of the 2.7x → 5.2x gap is real performance work; roughly 30% is what was always there but invisible due to broken CAPI. We’re explicit about this with clients. Anyone who isn’t, isn’t to be trusted.
- 2Creative volume found genuine new efficiency. 8 active concepts to 34, with documented winners scaled and refreshed. The algorithm performs better when you give it more good options to choose from.
- 3TikTok came online at a structurally lower CPM. TikTok added incremental volume at a CPM roughly 40% lower than Meta in this category. Blending that in pulls the aggregate ROAS up even before considering creative quality.
On the AOV lift. Small but real. Better audience signal through Advantage+ surfaced higher-intent buyers, which biased toward bundle and full-price purchases versus the discount-driven traffic the previous setup was pulling in.
What we’d flag to anyone reading this
This worked because four things were already true. They aren’t always.
Gross margin was strong.
This brand sat at 64% gross margin. Paid social scaling math only works above ~55% gross margin in most DTC categories. Below that, even 5x ROAS struggles to produce healthy contribution. We turn down DTC engagements where the unit economics can't support paid scaling.
The product was visually compelling.
Outdoor lifestyle is creative-friendly. Mountains, water, gear, dogs, fire pits — there's a near-infinite supply of hook variations. Categories with weaker visual storytelling (B2B software, financial products, generic supplements) need a different playbook and a different agency posture.
The brand committed to creative production tempo.
10–15 new assets per week was non-negotiable. Brands that want paid social results without funding creative production are asking for results that don't exist anymore. We had this conversation on day one and it was the most important conversation in the engagement.
Operations could absorb 4x demand.
Shipping, inventory, CX. If any of these break under a 4x demand surge, the paid program kills the brand by working too well. We coordinated with the client's ops lead before pulling the trigger on each scaling step.
And a fifth thing worth naming: iOS and CAPI fix the dashboard, not the customer journey.
A meaningful share of conversions that look “incremental” from CAPI improvements were always going to happen anyway. We treat the CAPI-corrected baseline as the honest comparison number, not the pre-fix reported baseline. The 30% blended CAC drop is the bottom-line check that the work is real.
Engagement details
- Team on the account
- 1 strategist, 1 Meta specialist, 1 TikTok specialist, 1 creative lead, 0.5 production coordinator
- Tool stack
- Meta Ads Manager, TikTok Ads Manager, Stape (server-side GTM), Shopify, Replo (landing pages), Motion or Atria (creative analytics), GA4, Looker Studio
- Creative production cadence
- 12–15 new ads/week across Meta and TikTok, mix of in-house production and creator-sourced UGC
- Reporting cadence
- Weekly creative performance review with the client’s marketing lead, monthly executive performance review, quarterly QBR
- Contract structure
- 3-month minimum, month-to-month after that
Ready for the same teardown on your paid social?
We’ll audit your Meta and TikTok accounts and show you what’s leaking before you commit to anything.