Excess share of voice: the exchange rate between voice and share

Origin

Jones was answering a practical question: how much should a brand spend to hold its position, and how much more to grow it [1]. His answer was that the relationship is not between spend and sales but between relative spend and relative share. What matters is your slice of the category’s voice, not your budget in absolute terms.

That reframing is the whole contribution. A budget that rises 10 per cent in a category whose total spend rises 20 per cent is a budget cut. Nobody books it as one.

Binet and Field took the idea into the IPA databank and made it operational [2][3]. The databank is the largest body of commercially reported effectiveness cases in the trade, which is both its strength and the source of most criticism of it.

What the evidence shows

Three findings do the work here, and they are worth separating because they get quoted as one.

The exchange rate. Ten points of positive excess share of voice is associated with about half a point of market share growth a year in consumer categories [2]. That is the headline figure, and it is an average. In business-to-business categories the reported figure runs higher, around 0.6 to 0.7 points [3]. Nielsen reached a similar half-point figure working independently across a set of 123 brands.

It works in both directions, and the downside is larger than the upside gets credit for. In the Binet and Field analysis, brands running negative excess share of voice lost share in the large majority of cases [2]. Underspending relative to your position is not a neutral saving. It is buying decline at a discount, and it is the single most expensive thing a finance-led marketing department does.

Creative quality is a multiplier on the rate, not a separate factor. The same excess share of voice converts at very different rates depending on whether anybody remembers the work. Field and Morgan put a number on the penalty for dullness: dull advertising needs roughly 2.6 times the spend to achieve the same effect [4].

Ten points of positive excess share of voice is associated with about half a point of market share growth a year in consumer categories, and the relationship runs in both directions.

Source: After Jones, J. P., Ad Spending: Maintaining Market Share, Harvard Business Review, 1990, and Binet, L. and Field, P., The Long and the Short of It, IPA, 2013. Brave New Media, 2026.

Unit: points of market share growth a year

Show the data

Excess share of voice (points)	Market share growth a year (points)
0	0
+10	0.5
+20	1.0

The compounding is the part that changes decisions. Half a point a year sounds dismissible in a planning meeting, which is exactly why it gets dismissed. Run twenty points of positive excess for three consecutive years and the arithmetic stops being dismissible. The effect is slow, cumulative, and invisible inside any one quarter, which is the same shape as the 95-5 problem and for the same reason.

What happens when a brand stops entirely has been studied directly. Hartnett and colleagues examined brands going dark and found measurable decay, with the rate depending on brand size [5].

The challenge

Four objections are worth taking seriously. The first two are strong.

The databank selects on success. Cases enter the IPA databank because an agency chose to submit them to an effectiveness award. Campaigns that failed are systematically underrepresented. Binet and Field acknowledge this. It does not invalidate the relationship, but it should make anyone cautious about treating the coefficients as precise.

Share of voice was defined when voice meant paid media. The original analyses measured advertising spend in measured media. A category where competitors win attention through owned audiences, search, creators, PR and community has a share of voice that no media auditor is counting. The metric is still meaningful. It is harder to compute honestly than it was, and a number that only counts paid is now a partial number.

It is correlational. Growing brands can afford to spend more. Some of the relationship runs that way, and the direction of causation is argued rather than settled in every published case.

Category matters. The half-point figure is an average across categories with different purchase cycles, different penetration ceilings and different competitive structures. Applying one rate to your own category without checking is the same error as applying one conversion rate.

None of this makes the rule useless. It makes it a planning heuristic with a stated confidence interval rather than a formula, and the article should not present it as more.

Where I see this go wrong

They compute share of voice against their own spend history instead of against the category. Year-on-year budget comparisons feel like the relevant number and are not. If three competitors enter the category, your voice has fallen without your budget moving.

They cut when results are soft. The cut produces negative excess, the negative excess produces decline, and the decline is read as proof that the spending was not working. It is the most reliable self-fulfilling loop in the discipline.

They treat creative quality as a separate conversation from budget. The Field and Morgan multiplier means a dull campaign at a given excess share of voice is a budget decision as much as a creative one. Those two meetings are usually held in different rooms, with different people, three months apart.

The pattern I notice most is a timing one. Share of voice is a category number, so it can only be computed once somebody has looked outward. Most budget rounds look inward: last year, plus or minus. By the time anyone asks what competitors are doing, the number that would have changed the decision is being used to explain the decision.

I spent eight years in television, media and investment before founding Brave New, which is long enough to watch the same argument from the side that sells the media. The sell side has always known the category figure. It is usually the only party in the room who does.

What to do with it

1. Compute share of voice against total category spend, not against your own previous year. If you cannot get category spend, say the number is an estimate and give the range.
2. Count more than paid media, or state plainly that you are not. A paid-only share of voice in a category with large owned audiences is a partial figure and should be labelled as one.
3. Compare the excess figure to your growth target before the budget is set, not after. The rule is a planning input, not a post-rationalisation.
4. Treat a proposed cut as a share forecast. Write down what share you expect to lose, in points, over what period. If nobody will write that down, the cut is not being decided on evidence.
5. Price dullness. If the work is not distinctive, assume you need materially more weight for the same effect, and put that in the budget line rather than hoping.
6. Check the rate for your own category rather than importing the average. Three years of your own voice and share data is enough to see whether the relationship holds for you.
7. Re-read the caveats before quoting the half point in a board paper. The number has a confidence interval and a source bias, and you will be asked.

If you run the company rather than the marketing

This one is about a decision you will make in a bad quarter, probably in the next eighteen months, and probably under pressure from someone whose job is to protect cash.

Marketing spend is the easiest line to cut because the damage does not appear in the same quarter as the saving. That is not an argument against ever cutting it. It is an argument for knowing the price before you do, and the price is measurable in a way most boards never ask for.

Two things to insist on. First, that share of voice is computed against the category rather than against your own last year. A flat budget in a category where three competitors have doubled their spend is a cut, and your finance pack will not show it as one. Second, that any proposed reduction comes with a written share forecast: how many points of market share we expect to lose, over what period. If nobody will put that in writing, the cut is being decided on mood rather than on evidence.

The uncomfortable part is that the loop is self-confirming. Cut, lose share slowly, read the lost share as proof the spending never worked, cut again. I have watched that sequence run to completion more than once, and at no point does anyone in it behave unreasonably.