ATL frameworks & methodology: how CMOs build above-the-line demand that actually converts

Next year's plan comes down to one cell long before it comes down to a channel list: the share of working media that sits above the line. Move that number ten points in either direction and you have either bought awareness your sales operation cannot convert, or starved a performance machine that has nothing left to harvest. Three calculations settle it: the share of voiceshare of voiceYour brand's share of total advertising or conversation volume in your category, measured against competitors over a defined period.View full definition → you need to hold or grow share, the shape of your reachreachThe number of unique people exposed to your message in a given period. Unlike impressions, reach counts each person once, no matter how often they see it.View full definition → curve, and the split between long-term brand work and short-term activation. Everything else in an ATL plan is execution against those three numbers. This lesson assumes the channel definitions and audience currencies the foundations lesson sets out, and spends the money instead.

Step 1: set the share of voice target

Share of voice (SOV) is your share of category advertising spendadvertising spendAny media you pay for: display ads, search ads, social ads, and sponsorships. You buy access to someone else's audience on a per-click, per-impression, or flat-fee basis.View full definition →. Share of marketShare of marketServiceable Obtainable Market: the share of your SAM you can realistically capture given current resources, channels, and competitive position.View full definition → (SOM) is your share of category sales. The gap between them, excess share of voice (ESOV), is the single best predictor of whether you gain or lose share over a year. Simon Broadbent's work in the 1980s established the relationship; the IPA Databank analyses by Les Binet and Peter Field have re-estimated it several times since. The historic rule of thumb was roughly 0.7 points of annual share growth per 10 points of ESOV. Their 2017 re-run put it closer to 0.5 points, because media inflation and fragmentation have made a point of voice more expensive to buy.

Run the arithmetic before the creative brief. Category advertising spend is £200m. You hold 8% of the market. Spending £16m puts you at parity: you defend, you do not grow. To target a full point of share you need ESOV of about 20, so SOV of 28%, so £56m. Most CMOs discover at this point that their growth target and their budget are not the same plan. The honest responses are to narrow the battleground (one region, one segment, one occasion where £16m does buy 28% of voice), to accept a slower share curve, or to raise the ask with the maths attached.

Two failure modes live in the denominator. First, if you count only TV in your category spend base you will overstate your SOV by a wide margin in any category where competitors run heavy retail media, online video or sponsorship. Second, a funded new entrant can double category spend in a quarter and silently halve your SOV while your absolute budget is unchanged. Track share of searchshare of searchThe proportion of searches in your category that mention your brand, a leading indicator of market share.View full definition → as a cheap monthly proxy between annual studies: Binet's work on share of search shows it moving ahead of share of market by several months in categories like automotive.

Mars is a useful case for what compounds. The Snickers hunger platform has run since 2010 across dozens of markets with the same joke structure and the same brand codes. Consistency of that length means each year's GRPs deposit into the same memory structure rather than rebuilding one, so the effective voice is higher than any single year's spend line suggests. A brand that re-platforms every three years pays for the same ground twice.

Step 2: build the reach curve before you argue about frequency

A reach curve is concave. Your first tranche of ratings buys 1+ cover cheaply; every subsequent tranche buys less of it, because you keep hitting people you have already reached. Somewhere around 70% cover of your target the cost per additional point of reach starts climbing steeply, and beyond about 85% it usually stops being rational to buy from the same channel at all.

That is the planning decision, and it is a channel decision rather than a weight decision. The last fifteen points of cover come from adding a different channel (radio in drive time, roadside OOH, broadcaster VOD, cinema), not from more of the one you started with. It matters most for younger targets, where linear TV alone plateaus well below the cover levels it delivered a decade ago, and a TV-only plan will quietly become a plan against over-45s no matter what the brief said.

On frequency, the Ehrenberg-Bass position is that reaching more people once or twice near the purchase occasion beats reaching fewer people seven times. Erwin Ephron's recency planning made the scheduling consequence explicit: in a category where someone buys every week, continuous low weight across 40 weeks beats three heavy bursts, because someone is always in market. The exceptions are real and worth naming. A launch with an unfamiliar claim needs frequency to teach it. A category with one buying season does not benefit from January weight: John Lewis concentrates its heaviest ATL into a six-week window because that is when the purchase occasion exists, and continuity outside it would fund memory that decays before it can be used. Nike flights against tournaments for the same reason.

The number to interrogate is the frequency distribution, not average frequency. An average frequency of 4.0 often conceals a plan where a third of the reached audience saw the ad once and the heaviest-viewing tenth saw it fifteen times. Buying cheap late-night inventory to top up a GRP total is the usual cause: it adds ratings, adds almost no cover, and inflates wastage against the group least likely to be a light buyer.

Step 3: split long and short

Binet and Field's headline finding from the IPA Databank is the 60/40 split: around 60% of budget on brand-building work aimed at long-term effects, 40% on activation aimed at this quarter. It is an average, not a law. Short-cycle categories such as retail sit nearer 50/50; categories with long consideration periods and high price points tilt further toward brand.

The timing asymmetry is what makes this hard to hold. Activation effects appear within days and decay within weeks, so they show up inside the reporting period. Brand effects build over six months and beyond, so in-year attributionattributionA framework for assigning credit to the touchpoints that contributed to a conversion, so you can measure which channels and interactions actually drive results.View full definition → systematically under-credits them. Left alone, a business that judges media on quarterly revenue will ratchet toward activation every planning cycle, one defensible decision at a time, until the ESOV number in step 1 has quietly collapsed. Write the split into the budget as a constraint at the top, before channel plans are built, or it will be eroded from the bottom.

Step 4: creative is the multiplier on all three numbers

The IPA Effectiveness Databank, covering more than 1,400 campaigns over 40 years, attributes a large share of the variance in campaign profitability to creative quality rather than media weight. That makes pre-testing an economic decision, not a taste one: a weak execution requires materially more ESOV to move the same share.

System1 Group, which sells creative pre-testing, scores ads for emotional response on a 1 to 5.9 star scale and publishes the correlation between high star scores and long-term share growth. John Lewis has tested its Christmas work with adam&eveDDB for years; "The Beginner" in 2022 scored at the top of that scale. The planning consequence: if your execution tests at two stars, moving budget into better creative development will return more than any optimisation of the media schedule you were about to sign.

Putting the three numbers together

Sequence matters. Fix the ESOV target first, because it sets the size of the budget. Then build the reach curve to find the point where cover stops being worth buying from your lead channel, which sets the channel mix. Then apply the long/short split, which sets what the money says. Only then does anyone open a schedule.

Two inputs belong in that sequence and usually are not. The first is the branded search response: weeks with TV weight reliably lift branded search volume within days, which means your search team needs the TV flight calendar in advance or it will under-bid on terms your ATL budget already paid to create. The second is decay. When finance models a budget cut, they model it as an in-year saving, and in-year it usually is, because brand memory decays slowly enough that the first two quarters look free. Model the ESOV consequence over three years or the cut will look costless every time it is proposed.

CMO action items

• Get a current category spend estimate covering all paid channelspaid channelsVisitors arriving via paid ads or sponsored placements, where you pay a platform to display your message rather than earning visits organically.View full definition →, not just TV, and calculate your ESOV. If it is negative and your plan promises share growth, one of the two is wrong.
• Ask your agency for the reach curve, not the GRP total, and the frequency distribution behind the average. Mark the cover level where cost per point starts accelerating and refuse to buy past it from the same channel.
• Set the long/short split at budget level before channel plans exist, and record the category evidence you used to pick it.
• Pre-test creative before committing media. A two-star execution at 28% SOV loses to a four-star execution at 20%.

Common mistakes that kill results

Mistake 1: Optimising ATL for clicks and direct response. When teams force response mechanics onto reach channels (short URLs, promo codes, QR codes on a 30-second spot) they trade attention away from the emotional encoding that makes the work stick. Measuring these channels on last-click attribution will always make them look broken, which is a measurement failure, not a channel failure.

Mistake 2: Isolating ATL from the rest of the demand plan in both planning and measurement. A brand team owning ATL and a performance team owning digital, with no shared KPIs and no shared calendar, produces demand that nobody captures. The ESOV, the reach curve and the long/short split are one plan. Splitting the ownership splits the result.