iOS broke your attribution. Most agencies still pretend it didn't.
Pixel-only attribution dropped 40% of conversions to dark mode. CAPI deployments are half-finished. Creative cadences run on vibes. We rebuild the measurement, run a real creative testing pipeline, and report on incremental — not last-click theatre.
Paid social is interruption-based advertising on social platforms — Meta (Facebook + Instagram), TikTok, LinkedIn, Pinterest, Reddit, Snapchat. It is creative-led: the creative is the strategy, and the algorithm rewards iteration speed. The discipline lives in three layers — measurement (server-side Conversions API to recover the 30–45% of conversions iOS broke), creative cadence (5–25 hypothesis-driven variants per month with kill dates), and audience architecture (Advantage+ with structured signals, LAL stacks against high-LTV cohorts, ABM layers for B2B). Reporting honest paid-social performance requires reconciling platform-attributed ROAS to blended MER; platform numbers alone are a fiction.
What's broken in most paid-social accounts
30–45% of conversions never make it back to Meta. The optimiser is making decisions on a partial dataset and underperforming as a result. Most agencies haven't deployed a clean Conversions API integration since 2022.
Three ads running for 90 days at declining ROAS. No creative test pipeline. No briefs tied to message-market hypotheses. Production is a bottleneck nobody owns.
Audience targeting at default. Placements at default. Bidding at default. The brand is treated like every other DTC brand and the algorithm responds accordingly.
Static creative repurposed from feed ads. No native vertical UGC. No TikTok Shop integration. The creative IS the strategy on TikTok and most accounts ignore that.
SaaS accounts running broad audiences with consumer-grade creative. No ABM layer. No retargeting tied to sales sequences. LinkedIn budget burning at $180 CPL with no MQL definition.
What we run
Meta (Facebook + Instagram, Advantage+ Shopping, Reels)
TikTok Ads + TikTok Shop integration
LinkedIn Campaign Manager (B2B, ABM, lead gen)
Pinterest Business Ads
Reddit Ads (community-led DTC + dev SaaS)
Snapchat (where relevant)
Pixel + CAPI server-side via GTM
Creative test register + UGC creator pipeline
How we audit & operate
Tracking stack rebuild
Pixel + CAPI deployed server-side. Event match quality scored. Deduplication verified. iOS 14+ post-back configuration audited. Most accounts gain back 25–40% of conversions in week one.
Creative test pipeline
5 message hypotheses per 30-day window. Brief → production → launch → kill date. Static, video, UGC. Variants A/B'd at message and visual level — not the same hook in three colours.
Audience architecture
Advantage+ with structured audience signals. Cold/warm/retargeting funnels segmented. LAL stacks tested against actual high-LTV cohorts pulled from Shopify or your CRM.
Reporting on net revenue
Triple Whale or Northbeam attribution overlaid with platform-reported. MMM where spend justifies it. Reports show platform ROAS AND blended MER side by side.
What lands in your inbox.
- CAPI deployment + tracking validation report
- Creative test register (rolling 90-day pipeline)
- Audience architecture + naming convention
- Weekly performance scorecard with net revenue + MER
- Monthly creative review + briefing for next cycle
- Quarterly attribution model recalibration
Where this worked.
Email rev 8% → 22% of total + paid CAC −31%
- CAPI deployment recovered 38% lost conversions
- 12 UGC creators tested in 60 days, 3 promoted to evergreen
- TikTok Shop integration drove $180K incremental in Q1
MQL volume +2.4× without raising spend
- LinkedIn ABM layer added at $40K/mo
- Demo-request LP message-match rebuilt
- Cost-per-MQL fell from $480 to $202
Questions we actually get asked.
Do you produce the creative?
Yes. We run an internal creative pipeline plus a vetted UGC creator network. We brief, source, edit, and ship — including post-purchase iteration off performance signal.
How many creatives do you test per month?
10–25 net-new variants per active account, depending on retainer tier. Each launches with a hypothesis and a kill date. Winners promoted, losers killed within 14 days.
Do you work with brands that have channel conflicts?
We don't take competing brands in the same vertical/category at the same time. If you're a $25M home-goods brand, we won't take your direct competitor — full stop. Public conflict policy.
What about iOS attribution loss?
We rebuild it server-side with CAPI, then validate event match quality. Most accounts recover 25–40% of conversions Meta was missing. We also model incremental lift via geo-tests where the budget supports it.