Paid media that compounds, not just spends.
Most paid programs are scoring last-click goals against themselves. We rebuild your measurement first — incrementality testing, MMM, post-purchase surveys — then deploy budget where it actually moves the business. Channels are tools, not strategies. The goal is incremental revenue you can defend in front of a board, not a dashboard that flatters the manager.
Why this discipline earns its budget.
Paid media is the only channel where you can put a dollar in this morning and pull a customer out by lunch. That speed is also its trap. The same dashboards that let you move fast let you fool yourself faster — last-click attribution, platform-reported ROAS, and the inescapable gravity of optimizing the metric the platform happens to give you.
The channel has fundamentally changed three times in the last six years. iOS 14.5 broke deterministic Meta attribution. Performance Max collapsed Google account structure into a black box. Then AI bidding became the only game in town. Every one of those shifts created a window where teams kept running playbooks from the previous era and watched CAC quietly rise for three quarters before anyone noticed.
Done right, paid is the only marketing channel that gives you the data resolution to actually run experiments. Geo-holdouts, conversion lift studies, post-purchase surveys, ghost-bid tests — these are the tools that separate paid teams who know what's working from paid teams who just trust the platform. Most companies are still running paid like it's 2019. The ones that aren't have a structural CAC advantage.
The work, spelled out.
Four phases. No surprises.
Audit
Account teardown, attribution diagnosis, and a channel-by-channel honest read on what's working, what's wasting, and what's missing.
Re-rig
Measurement foundation first. We rebuild your tracking, conversion API, and reporting before we touch a single bid.
Launch
Structured test calendar across channels, creative variants in production, and budget allocation that respects diminishing returns curves.
Scale
We aggressively reinvest in winners, cut losers fast, and run a quarterly incrementality test to prove what's actually driving lift.
Our signature, named.
Our paid program runs on what we call the Incrementality-First Stack: measurement before media, creative before bids, and channel before tactic. Before we spend a dollar on your behalf, we baseline incrementality with a geo-holdout or a conversion-lift study. That number — not platform ROAS — becomes the truth we manage against. Anything that moves that number gets more budget. Anything that doesn't gets cut, even if the platform dashboard says it's working.
On the execution side we run a creative production loop tied to a structured testing matrix — hook variants, format variants, format-rotation cycles — instrumented in Google Sheets and reviewed weekly. We use Northbeam or Triple Whale for blended attribution where DTC, Google Analytics 4 with custom server-side events where ecommerce-light, and a custom Snowflake pipeline for the enterprise accounts. Quarterly we re-run the incrementality test from scratch. The map updates because the territory updates.
Common mistakes, and the truth instead.
- Trusting platform-reported ROAS.
- Meta and Google both over-report conversions by 1.4x to 3x against any sane ground truth. We've measured this across 80+ accounts with geo-holdout testing. If your CFO is making budget decisions off a 7-day click attribution window, the budget decisions are wrong.
- Refusing to use Performance Max because it's a black box.
- It is a black box, and it still drives 30–55% of incremental revenue for most retail and lead-gen accounts when it's structured right. The answer is to constrain it correctly — asset groups, audience signals, brand exclusions — not to avoid it. Refusing PMax in 2026 is refusing to compete.
- Treating LinkedIn as overpriced Meta.
- LinkedIn CPM is 8–12x Meta and the audience targeting is genuinely different. It's not a brand-awareness channel; it's a buying-committee channel. We use it for thought-leader ads to named accounts and almost never for cold acquisition. Wrong tool for the wrong job is most of paid's bad reputation.
- Cutting creative testing to save budget.
- Creative is now 70%+ of paid social performance — bidding, audiences, and placements are mostly commoditized. If you're not shipping 8–15 fresh variants per channel per month, you're not running paid social, you're maintaining it. Performance fatigues on a 4-6 week cycle and the dashboards lag the rot.
- Running paid in isolation from SEO and brand.
- Branded search is just paid harvesting SEO and brand investment. If the brand layer is starving, branded CPCs creep up and assisted-conversion paths shorten. We always look at the blended portrait — paid, organic, direct, and email — before recommending a paid budget.
What people actually ask us about paid media.
No. Engagements are month-to-month after a 90-day onboarding window. If we're not earning our keep, you leave.
Flat retainer based on channel scope and complexity. Percentage-of-spend creates the wrong incentive. We'd rather be paid to do less if less is the right answer.
Yes. We integrate with whatever stack you have — GA4, Segment, Rudderstack, custom event pipelines — and add server-side tracking where browser-side is leaking signal.
Most clients spend between $25K and $1.5M per month across channels. Below that, an in-house generalist is usually the right call.
An in-house manager owns one or two channels deeply and trades coverage for depth. We own the blended picture, run incrementality tests no single channel manager would think to run, and ship creative at a cadence most in-house teams can't match. The right answer is often both — an in-house lead plus us — not either-or.
Agency fees range from $6K to $40K per month depending on channel mix and spend tier. Media is on top, paid directly to the platforms in almost all cases. We will quote a specific number after a 30-minute scoping call, never before — anyone giving a flat number sight-unseen is overcharging or under-delivering.
Google Search and Shopping typically show signal in week two. Performance Max and Meta Advantage+ campaigns need 2–4 weeks of learning before they're representative. LinkedIn and programmatic CTV are 6–10 weeks. Anyone promising immediate scale on day one has not run a learning phase recently.
Yes — static, motion, UGC, and short-form video. We have a 5-person in-house production team plus a vetted bench of creators. Roughly 60% of our paid clients use us for creative; the other 40% have their own studio and we run the production loop with them.
The channels that amplify this one.
↳ Paid Media brief
Ready to talk paid media?
Send us a paragraph about where you are today and what would make the next ninety days a win. We'll reply within one business day.