Two clocks, one budget.

Watch what happens when someone who already wants running shoes sees a discount ad tonight: they buy. Tomorrow morning the sale shows up in a dashboard with the ad's name on it. That's activation — finding people whose decision is nearly made and clearing the last obstacle. It's fast, it's measurable, and it decays fast: stop spending and the sales stop within days, because you were never creating buyers, only collecting them.

Now watch the other mechanism. A person who is not in the market for shoes sees a great campaign, feels something for two seconds, scrolls on, forgets it happened. Nothing shows up in any dashboard. But eleven months later, standing in a store with sore knees, one brand comes to mind first — and Part I told you why: the memory link was built long before the need arrived. That's brand building — planting retrieval cues in people who aren't buying yet, so that when they finally are, the auction in their head is already rigged.

Programmer's version: activation is a cache hit — the value was precomputed by someone else (a need, a habit, an earlier campaign) and you're just serving it fast. Brand is building the index — expensive, no user-visible payoff today, and the only reason tomorrow's lookups return you instead of a linear scan of the shelf. A system that only serves cache hits feels efficient right up until the cache empties.

Two mechanisms, two time constants, two kinds of evidence. Most marketing arguments — long copy vs short, TV vs search, "brand is dead" vs "performance is a scam" — are secretly this one argument: which clock are you reading?

In 2013 two researchers, Les Binet and Peter Field, did something unusual for marketing: instead of arguing from taste, they mined a database. The IPA Effectiveness Databank holds hundreds of documented campaigns — objectives, budgets, and crucially, business results over years, not weeks. Binet and Field sorted them by how the budget split between brand building and activation, then asked which splits produced large, lasting business effects: share growth, price resilience, profit.

The answer had a shape. Campaigns that spent everything on activation spiked and flatlined. Campaigns that spent everything on brand built slowly and left money on the table this quarter. The campaigns that won big over three years clustered around a balance — on average near 60% brand, 40% activation — with the brand share creating the demand and the activation share converting it. The two clocks aren't just compatible; they're complementary. Brand makes activation cheaper (people click ads for names they already know), and activation cashes the cheques brand writes.

Run the trade yourself. The simulator below spends a fixed budget for three years at whatever split you choose — watch what each extreme does, and where the compounding kicks in.

If balance wins, why does almost every budget drift toward activation? Not because marketers are foolish — because the measurement is asymmetric. Activation produces receipts on the same timescale as a budget review: click, buy, attribute, screenshot. Brand produces its effects on a timescale no dashboard covers, through people who saw an ad this year and will buy next year — a path no click-tracker can follow. One clock is wired to the reporting system; the other is invisible to it.

Chapter 15 showed the sharp edge of this: last-click attribution doesn't just miss brand effects, it actively overcredits activation, claiming sales that would have happened anyway. Put those together and the weekly dashboard is systematically flattering the short clock twice — counting its wins generously and not counting the long clock at all. A rational executive reading that dashboard cuts brand spend, books the savings, and watches nothing bad happen… for about a year. The costs arrive later, as softer awareness, weaker pricing power, and rising acquisition costs — by which time the dashboard blames the activation team.

Programmer's version: this is optimizing against the metric you can log, not the outcome you want. Latency is easy to instrument; reliability debt is not. Teams that chase only the instrumented number ship faster every sprint while the incident rate quietly compounds. Short-termism isn't a moral failure — it's what happens when one term of the objective function is unlogged.

The toggle below is the whole pathology in one picture: the same two campaigns, scored by the week-one dashboard and then by a three-year incrementality view. Watch the ranking flip.

The long clock has its own planning number, and it's older than 60/40. Take your share of voice (your slice of the category's advertising) and subtract your share of market (your slice of its sales). The difference — excess share of voice, ESOV — predicts where your market share is heading. Out-shout your size and you tend to grow; whisper below it and you tend to shrink. The rule of thumb from decades of ad-spend studies: roughly half a point of share growth per year for every 10 points of ESOV, category and creative quality permitting.

Why would such a blunt ratio work at all? Because share of voice is a proxy for share of memory-building. Chapter 5's mental availability is bought with reach over time; if your competitor is building memory links at twice your rate, your share erodes with a lag, however good your product is. ESOV is the budget-level view of the same physics.

Two things follow, and both are uncomfortable. First: growth has a media price — a challenger planning to grow 4 share points without an ESOV to match is planning a miracle. Second: cutting spend reads as free profit for exactly one budget cycle. Drop below your market share in voice and nothing happens this quarter; the share bleed arrives over the following years, slow and deniable. ESOV is the number that makes the invisible clock legible to a CFO.

The split isn't just financial — the two clocks want different creative. Activation talks to someone mid-decision: it works rationally, with specifics — price, proof, offer, deadline. Chapter 14's Ogilvy discipline and Chapter 15's reason-why copy live here, and so does search, which is why the next chapter is entirely about intent.

Brand work talks to someone who isn't listening and won't need you for months. Rational arguments bounce off that person — they have no decision to apply them to. What sticks is what Part III built: emotion, story, distinctive assets. A feeling attached to a memorable cue (Chapter 6's assets, Chapter 17's characters) survives eleven months of not-caring in a way a feature list never will. The evidence backs the register switch: in long-window analyses, emotional campaigns outperform rational ones on pricing power and share — while rational wins inside the purchase window.

This is why "make the logo bigger and add a discount code to the Super Bowl ad" and "run pure vibes on search" are both category errors. The message has to match the clock of the person receiving it — and the campaigns that endure, from Stella Artois to the DTC brands rediscovering billboards, are the ones that kept both registers running at once.

Now the honest caveats, because "60/40" quoted as scripture is its own failure mode. The number is an average across a particular database, and the spread around it is wide and patterned:

• Category clock speed. Impulse categories with short purchase cycles can run more activation; considered purchases with years between buys need more memory-building, because almost nobody is ever in-market.
• B2B. Long committees, rational cover stories — but the research (Binet & Field's B2B work included) keeps finding the same two-clock structure, with a balance nearer 50/50 and brand mattering more than the culture assumes.
• Stage. A startup with zero awareness has nothing to harvest — early spend is forced toward brand-and-category building. A brand at 60% share has little headroom and harvests efficiently.
• Innovation. New categories (Chapter 11's territory) pay a memory tax: you're building the category's links, not just your brand's, so the long clock needs extra funding.

And keep the critiques in frame: the databank over-represents campaigns someone bothered to write up; award-entry effectiveness papers are not a random sample; and "share of voice" is getting harder to even measure when attention per impression varies a hundredfold. The 2026 refinement isn't a new magic number — it's planning in attentive reach and category-specific priors, then updating with your own experiments (Chapter 15's discipline, aimed at the budget itself).

Programmer's version: 60/40 is a sensible default in the config file — the value you ship before you have production data, not the value you defend after you do. The failure modes are symmetric: cargo-culting the default forever, or deleting it because "our data" — a week of last-click — says the index isn't needed.

Compress the chapter into the artifact you'd use in a planning meeting:

• Split the budget on purpose. Start from a category-informed prior (default 60/40 brand/activation; B2B nearer 50/50; new category, more brand; short-cycle impulse, more activation). An unmanaged split is a decaying one.
• Check your ESOV. Voice below market share is a slow-motion share cut, whatever this quarter's P&L says. Growth targets need an ESOV to fund them.
• Score each clock with its own ruler. Activation: experiments and incrementality (Chapter 15), judged in weeks. Brand: awareness, mental availability, pricing power, base sales — judged in years. Never let one clock's ruler grade the other's work.
• Match message to clock. Emotion and distinctive assets for the 95% not buying; specifics and offers for the 5% who are. Both, always, in different placements.

That's the operating system for Part IV. The next six chapters walk the channels — search, the feed, video, email, creators, retail media — and every one of them is this chapter in disguise: a different bargain between attention now and memory later, at a different price. First stop: the channel that is pure short clock, and what happens to it when the answer engine stops sending clicks. Search.