The $200M Playbook

I've managed more than $200M in paid media over 20 years - at IAC ($83M a year across 63 accounts and 1,600 campaigns), at Care.com's HomePay business, at WayBetter across six consumer apps, and as Head of Performance and Growth at an agency managing $15M for lead-gen clients.

That much spend, across that many channels and verticals, either turns you into a generalist who knows nothing, or forces you to build an operating system. I built the system. Here it is.

---

01The channel doesn't care about your org chart

Google, Meta, YouTube, and CTV each have their own physics. The mistake most teams make is treating paid media as one discipline owned by one person. It isn't. Google rewards intent and feed quality. Meta rewards creative velocity and signal strength. YouTube rewards narrative and audience architecture. CTV rewards reach math and frequency discipline.

If your paid lead is strong in one and weak in three, your results will look exactly like that. Staff or partner for all four. Don't hire one "paid media manager" and expect a miracle across channels that share almost nothing except a login page.

---

02Prune before you polish

At Care.com I cut HomePay's CAC by 30% year over year. The biggest single driver wasn't a bid strategy change, an audience unlock, or even new creative - it was collapsing 118 active campaigns down to 18. Spend and data were spread across more campaigns than the algorithms could ever learn from, so every auction was starving for signal. Consolidate the structure and the same budget suddenly has enough volume per campaign to actually optimize.

Every scaled account I've ever inherited has some version of this problem. Structure gets bolted onto for years, nothing ever gets retired, and eventually you have a sprawl that's impossible for any algorithm - or human - to read. Before you touch bids, before you brief new creative, before you chase a shiny new channel, prune. You will almost always find performance sitting there, waiting to be freed up.

---

03Build for signal, not for dashboards

Pruning is only half the fix. The other half is making sure the platforms are optimizing toward something that actually matters. At HomePay, the account wasn't just over-structured - it was pointing Google and Meta at the wrong conversion actions, so even a perfectly consolidated account would have been learning from bad data.

Modern ad platforms are machine-learning systems. They are only as good as the signal you feed them. A beautifully wired dashboard that reports a weak conversion event will give you a beautifully wired dashboard of bad decisions.

Before I touch a bid or a budget, I ask one question: what is this account optimizing toward, and is that event actually correlated with revenue?

If an account is optimizing to "Lead" and 60% of leads never convert, the algorithm is actively working against you. Fix the event, fix the feedback loop, then fix the media plan. Not the other way around.

---

04Forecast before you spend. Reforecast after you learn.

At IAC I helped forecast $100M+ in planned expenditures across the portfolio. The job wasn't predicting the future with precision - nobody can. The job was building a model that surfaced the assumptions so leadership could make a decision with their eyes open.

Every new account I take on gets a simple forecast in week one: expected CAC by channel, expected volume at three spend levels, expected LTV, and the three assumptions most likely to be wrong. Then I revisit the model every 30 days and change the numbers. A forecast that never changes is theater. A forecast that changes with data is a compass.

---

05Don't over-index on attribution. Index on incrementality.

I'll be direct: I don't love the modern obsession with attribution plumbing. Tags, pixels, server-side mapping - that's necessary infrastructure, but it's not where the money is made. The money is made in the media buy and the creative. Attribution tells you what the platforms already believe. Incrementality - holdouts, geo tests, on/off tests - tells you what's actually true.

When a channel's platform-reported ROAS and its incrementality test disagree, bet on incrementality. I've turned off whole campaigns that looked great in GA and found my total revenue went up. That's a scalping operation, not a growth driver.

---

06Velocity beats polish

At WayBetter, six consumer apps, 60% YoY growth in paid subs. We didn't win by making better ads than our competitors. We won by shipping more of them, killing the losers faster, and re-deploying spend into what was working inside the same week.

The agencies and in-house teams that lose the creative war aren't losing on taste. They're losing on cycle time. If your team takes three weeks to ship a new ad, you're playing a different game than a team that ships three new ads a week.

---

07Margin discipline is a performance metric

Every dollar you spend managing the account - the tool stack, the extra headcount, the agency fee layered on top - comes out of the same P&L as the media itself. At Umbrella I consolidated a client from disparate platforms into the Google Marketing Platform stack and freed up $250K a year in tool cost while lifting ROI 40%. That $250K was the difference between a profitable account and a break-even one.

Don't just measure CAC. Measure fully-loaded CAC - media spend plus every dollar it took to run the media. Most teams don't. That's how an "efficient" account quietly loses money.

---

The short version

Staff for the channel, not for the title. Prune account sprawl before you chase anything new. Feed the algorithm a real revenue signal. Forecast with honesty and reforecast with data. Trust incrementality over attribution. Ship fast, kill fast, re-deploy fast. Watch your fully-loaded numbers, not the vanity ones.

That's the operating system. I've run it at a $15M agency budget, an $83M portfolio, and everything in between. It scales up and it scales down.