Frameworks & methodology: GRP, TRP & offline metrics

Your agency comes back with two schedules for the same £600,000 of airtime. Plan A delivers 65 percent reach of your target at an average frequency of 4.3. Plan B delivers 45 percent at 6.2. Both bill identically, both land at roughly 280 TRPs, and the equation that produced them cannot tell you which to buy. What follows is the arithmetic behind that choice, and the conversion that turns the answer into a number your digital team can put alongside their own.

Core concept: solving the planning equation

You already have the relationship between rating points, 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 → and frequency from the foundations lesson. Planning is what happens when you solve it for the unknown you actually control.

Start from cost. Cost per point (CPP) is your budget divided by rating points delivered against the target. At £600,000 for 280 TRPs, you are paying about £2,143 per point. Fix any two of budget, CPP and TRPs and the third falls out. What does not fall out is the split between reach and frequency, and that is the only part of the equation with strategic content.

The reason the split matters is that reach is bought on a curve and frequency is not. The first 100 TRPs against a broad target typically buy something close to 45 to 55 percent 1+ reach. The second 100 add perhaps 12 to 15 points. The third add single digits, and everything after that is almost pure frequency against people you have already reached. Get your agency to show you the actual curve from their planning tool, calibrated on panel data for your specific target, rather than accepting a rule of thumb. The shape of that curve is the single most useful number in the whole exercise, because it tells you where your incremental pound stops buying new people.

Key sub-concept 1: reach vs. frequency trade-offs

Average frequency is a mean over a badly skewed distribution. A plan reporting 4.3 average exposures does not give most of the target 4 exposures: the heaviest viewing quintile can absorb more than half of all impressionsimpressionsThe total number of times an ad or piece of content is displayed, regardless of clicks. Each display counts as one impression, even to the same person.View full definition → delivered, while a large slice of your light-viewing target sees the ad once or never. When you plan to a "3+ reach" target rather than to total rating points, you are forcing that distribution into the open.

Procter and Gamble pulled more than $100 million out of digital spend in a single quarter of 2017 and reported no drop in business growth. Marc Pritchard's account of it turned on exactly this arithmetic: much of the money was buying repeat exposures against audiences already saturated, so the impressions were real and the incremental reach was near zero. The planning fix is unglamorous. Cap frequency, widen the target definition, and re-run the reach curve before adding weight.

Key sub-concept 2: converting cpp into CPMCPMCost Per Mille: the cost to deliver 1,000 ad impressions. A pricing and benchmarking metric for awareness campaigns where reach matters more than clicks.View full definition →

CPP and CPM are the same trade priced against different denominators, and moving between them is one line of arithmetic:

CPM = CPP x 100,000 / target universe

With a CPP of £2,143 and a target universe you have defined as 12 million adults, your effective CPM is about £17.86. Change the universe definition and the CPM moves without a single spot changing hands, which is why the universe number in your plan deserves as much scrutiny as the rate card. Narrowing a target from all adults to a 12 million subset does not make the buy more efficient; it makes the same impressions cost more per thousand relevant people, and that is the honest number.

CPP varies by daypart, by month and by demand cycle. Q4 points cost materially more than Q1 points in most markets. Know your category's benchmark before you negotiate, because your agency does.

Key sub-concept 3: what panel size does to your numbers

Nielsen, which sells the currency it measures, runs a national US panel in the region of 40,000 households. That is ample for a broad demo and thin for a narrow one. A programme delivering a 0.4 rating against adults 18 to 49 may rest on a handful of panel homes, and the confidence interval around it is wide enough to swallow the difference between two schedules you are choosing between. Two practical consequences: do not optimise a plan on rating differences smaller than the sampling error, and be sceptical of post-campaign delivery reports on niche targets that claim precision to one decimal place.

Nielsen ONE, launched in 2022, folds set-top and streaming return-path data in alongside the panel to produce deduplicated cross-platform reach. That changes the denominator you are planning against, sometimes mid-flight, so any multi-quarter comparison needs to state which measurement basis it uses.

Key sub-concept 4: turning a TV plan into digital-comparable impressions

The conversion is simple and the caveats are not:

Impressions = TRPs x target universe / 100

280 TRPs against 12 million adults is 33.6 million impressions. Hand that number to a digital planner and they will compare it with their own, and the comparison will be wrong in three specific ways.

First, the counting rules differ. A TV impression derives from average-minute audience among people credited as viewing; a digital video impression under MRC standards needs two seconds with half the pixels in view. Second, panel measurement counts people, while much digital and connected TV delivery counts devices, so co-viewing is inside the TV number and missing from the other one. Third, the two currencies have different audit status, which is the board-level trade the playbook lesson takes up rather than a planning question.

The workable answer is to publish both numbers with their definitions attached, and to state a haircut you apply to unaudited impressions before adding them to anything. Pick the haircut deliberately, document it, and keep it constant across quarters so your trend line means something.

Real-world cases

Case 1: P&G's frequency reset. After the 2017 cut, the company kept pushing on the same lever, reducing excess frequency and broadening reach targets rather than buying more weight. The instructive part for planners is the direction of travel: spend came out, delivered impressions fell, and reach held. That only happens when the original plan was sitting deep in the flat part of the reach curve.

Case 2: Sky AdSmart. Sky, which sells the addressable inventory in question, launched AdSmart in 2014 to serve different ads to different households watching the same programme. Delivery arrives as households and impressions, not as ratings against a panel universe, so a campaign mixing linear spots with addressable delivery is running two currencies at once. You cannot add them. You can convert the addressable side to a rating-point equivalent only by declaring the same universe used on the linear side, and if the addressable target is narrower (which is the point of buying it), that equivalence is a fiction. Report the two streams separately and combine only at the level of incremental reach.

Case 3: The Nielsen accreditation suspension. In September 2021 the Media Rating Council suspended accreditation of Nielsen's national and local television ratings after panel maintenance problems during the pandemic led to under-counting. Buyers with audience guarantees written in rating points found the basis of those guarantees in dispute. The planning lesson is that a currency is an institution, not a constant, and any contract that pays out on delivered points needs a clause covering what happens when the measurement itself is restated.

CMO action items

• Require weekly pacing reports showing reach, average frequency and 3+ reach against the target, not total points delivered. Total points hide the shape of the plan, which is the only thing you were choosing between.
• Ask for the reach curve, not the reach number. Where does the curve flatten for your target, and how many points are you buying past that point?
• Track 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 → as your rating points over total category rating points each quarter, and compare it with your market sharemarket shareThe percentage of total industry sales your company captures in a given period. It measures competitive position relative to rivals in a defined market.View full definition →. A gap held for two consecutive quarters without a deliberate reason is a plan drifting, not a strategy.
• Fix one written conversion method between rating points and impressions, including the haircut on unaudited inventory, and make every plan use it.

Common mistakes that kill results

Mistake 1: Planning against a universe you did not choose. Agencies default to broad demographic universes because they produce flattering CPPs. If your buyer is a 12 million adult subset and the plan is priced against 45 million, the efficiency on the page is arithmetic, not audience. Restate every plan on your own universe before comparing options.

Mistake 2: Treating average frequency as the frequency. A schedule showing 6.2 average exposures across 45 percent reach is delivering double digits to your heaviest viewers and one exposure to a large tail. Response curves flatten well before most plans acknowledge it, and at the top of the distribution additional exposures buy irritation rather than salience. Concentrating 400 TRPs into a fortnight against a small target is the fastest way to find that out.

Mistake 3: Comparing converted impressions without stating the counting rule. Two numbers labelled "impressions" that were produced by a people-based panel and a device-based ad server are not the same quantity, and averaging them produces a figure with no referent. Publish the definition next to the number every time, or the comparison will be made for you by whoever benefits from it.