It’s time for ‘share of search’ to replace ‘share of voice’
Share of voice is a key metric for setting budgets and predicting growth, but digital media make it impossible to calculate accurately. Understanding your share of search queries is a simple and elegant alternative.
James Peckham was a smart guy. He joined AC Nielsen in his early 20s and spent the next 43 years working for the company. He was one of Nielsen’s most important and senior marketing thinkers, working at the company during the key ‘Mad Men’ era. But his claim to fame goes well beyond simply being a big player, at a big company, during a big time in the history of marketing.
Peckham is one of the few marketers to have a theory named after him. Peckham’s Formula posits that when you launch a brand you should set its advertising budget based on your desired share of market. Specifically, your initial share of voice should be 1.5 times the desired market share you want to achieve by the end of the brand’s second year.
Let’s say Marketing Week decides to diversify into ice cream – something I know our editor is constantly considering. He should first estimate the total amount spent by all the ice cream brands on communications. Let’s say it’s £100m in 2020. Next, he has to decide what success looks like for the new tubs of Marketing Week Vanilla. Why don’t we set an ambitious 5% market share target? Finally, he should apply Peckham’s formula and set his ad budget at £100m x 5% x 1.5. About £7m.
The Formula itself is based on some pretty significant empirical analysis. Peckham spent almost 20 years studying the relationship between what grocery brands spent on marketing communications and the subsequent market share they achieved. And, as such, Peckham was the first marketer to connect the dots between a brand’s share of voice and its share of market in a formal, prescriptive way.
The famed advertising thinker John Philip Jones followed Peckham and continued this connection. Jones was less interested in brand launch and more focused on helping existing brands correctly set their advertising budgets. He observed a very clear link between a brand’s market share and its share of voice from across hundreds of case studies. Jones also noted that brands with much a bigger market share could get away with significantly less share of voice, while smaller brands were in the inverse position and had to invest in a greater share of voice than their share of market might initially suggest.
The final major piece in the SOV/SOM puzzle came from Peter Field and Les Binet. The duo might now be famous for The Long and Short of It, but over a decade ago their work was also instrumental in showing an equilibrium between a brand’s share of voice and its share of market. More importantly, it was Field and Binet who also established the idea that a brand achieves excess share of voice (positive ESOV) if its share of voice significantly exceeds its share of market. And that excess was then shown to result in market share growth, gradually over time, if it was maintained.
Watch: Ritson on the effectiveness factor that helped Lidl double its market share
So, to recap. Share of voice is linked to share of market. A relative increase in share of voice leads to a subsequent and corresponding growth in market share. You can predict this growth rate over time. Holy shit, where is my calculator?
The drawbacks of SOV
Before everyone gets too excited, there are of course some pretty meaty caveats to be applied at this point. For starters, most of this work was done in the relatively stable and predictable world of FMCG and might not apply quite as well or as easily to other corners of the marketing universe. It is also important to realise that the actual amount spent on advertising is merely one variable in a phalanx of inputs that lead to overall effectiveness.
Budgets can only tell us so much and they often obscure other crucial strategic and tactical questions. Were the ads any good? Was the media selected appropriately? Was the strategy correct? And – lest we forget – was the brand shit?
But the big practical caveat behind all of this work is access to the communications data. Most companies have a pretty good grasp of their share of market. But working out their relative share of voice is fantastically tricky. If I had a dollar for every client who has asked me ‘how do we work out our share of voice?’ I would be a very rich marketing columnist.
At first sight you might think it would be relatively easy. A marketing director should (one would hope) know how much she is spending on communications across the current year. The snag is working out the total communications spend figure in the market, so your brand’s share of it can be calculated.
Actually, there are three snags.
First, you have to identify who is also actually playing in the category. That is not as easy as it sounds. Categories are – of course – complete bollocks. They might be a staple marketing concept but they do not actually exist for consumers. Roger does not walk out of his office looking for instant noodles. He races out of his office with 10 minutes to get something to bring back to the office that he can eat, fast, before his meeting starts. Chocolate bars compete with noodles, which compete with muesli bars and kebabs.
It is essential marketers get this point. Competition is not up to them; like everything else in marketing, it is ultimately decided on by the consumer. And the competitor that kills you is inevitably the one you did not see coming because you were too busy working on your category to see them attacking you via your blind spot.
Search would appear to be a far more universal assessment tool than the amount spent on advertising. The data for share of search would be just as relevant to sectors where standard advertising is less important.
That means all kinds of different, non-categorical players should be included in the total share of voice calculation. Credit card companies in the US, for example, would be wrong just to include the usual suspects in their voice calculations. Apple needs to be in there because analysts predict it could well account for one in 10 payment transactions by 2025.
Which brings us to the next challenge of SOV. Do you just include the money you estimate Apple is spending promoting its new card or also include the whole branded caboodle and add all the Apple brand spend? Apple Pay is partly about the swanky new metal card but its mostly about the Apple brand, so should all of it be included?
And even if I want to include all of it, how do I get the media data for all the different channels? When Peckham was doing this in the 1970s, you got a syndicated fax each month with a table estimating each brand’s spend across TV, newspapers, outdoor, radio and cinema. Based on that you pretty much had everything you needed. JP Jones mostly used TV ad spend for his data.
Those tidy little estimates ended a decade ago. With digital media accounting for around half the total media spend in most categories, things are very, very opaque these days. That’s partly because some of this media is earned and owned in ways that would not show up in a typical spend estimate. And also because the proprietary nature of the digital giants takes a lot of that syndicated spend data off a one page fax and buries it three feet under the ground, in a safe, behind a big garden wall.
Binet, the lower half of the famous duo, assures me that, despite these difficulties, if you can get your hands on the available data for SOV you will still see a predictive relationship between an excess in it and market share growth. But that still does not obscure the issue of getting the data accurately for the right players in the first place.
The new alternative
The good news is that a solution may be on the horizon. A gaggle of marketers have been talking for a while now about the fascinating concept of ‘share of search’. And, as you think more about that concept, the idea becomes all the more alluring.
The link between share of voice and share of market is the story of joining the top and bottom of the funnel. We estimate how much each brand is pumping into the funnel in advertising dollars and then observe changes in the eventual purchase behaviours and market share shifts down below. What if we shifted our gaze away from the very top of the funnel and looked a little further down? Why not measure how many people are searching online for the different brands in the category and compare that number instead? A share of search rather than a share of voice has a number of big advantages.
‘Funnel juggling’ is the answer to marketing effectiveness
First, because it is further down the funnel, it should be more predictive than share of voice. Even in low-involvement categories like FMCG, a relative comparison of search frequency should provide us with a decent proxy for salience and consideration. Searching for chocolate bars is less common than searching for insurance, but in relative terms the balance of one brand’s share of total search in a category should still be revealing.
If we apply the idea of ESOV to search and call it ESOS – excess share of search – we could incorporate those different category relativities and the impact of brand size too. Your brand has a 15% market share but enjoyed 30% of the total search frequency last month. Congratulations, you have +15 ESOS and we know that translates into X% of market share growth in your category per year.
ESOS would also remove the difficult challenge of working out who, exactly, should be included in the competitor comparisons. Before counting the proportion of search entries made, the associative searches made during the same period could quickly indicate the category entry points that cue a brand search. The other competing brands that are also entered during the search session could then provide a jigsaw puzzle of different competitors that – taken together – present all the brands that should be compared against each other. Search could be used not only to measure a brand’s salience versus other brands but also indicate which other brands should make up that set in the first place.
The frequency of this data could also prove to be a big help to marketers. There are a few brands assessing ESOV on monthly or quarterly cycles. But for the most part SOV data is either not measured at all or used as part of an annual exercise. Imagine getting this data on a weekly basis with your brand and your competitors shown in a single dynamic chart.
And, best of all, search would appear to be a far more universal assessment tool than the amount spent on advertising. The data for share of search would be just as relevant to sectors where standard advertising is less important. B2B markets, for example.
For all these reasons ESOS – excess share of search – looks like becoming one of the most exciting and useful marketing metrics of the near future. Its evolution and availability will depend, of course, on Google and how quickly its executives grasp the treasure trove of data they are now sitting on.
The company has produced some spectacular metrics and tools aimed at the bottom of the funnel. More recently its ‘Messy Middle’ report demonstrated an appreciation for the challenges further up the funnel. It’s now time for Google to press the green button and accelerate their work on ESOS to illuminate the ‘Tricky Top’ of the funnel too.
Marketers want the empirical insights that ESOV promises but rarely delivers. They are increasingly attracted to the ideas of salience and ensuring their brand achieves as much mental availability as possible. They may even be convinced that, of all the hierarchical challenges inherent in brand building, coming to mind might just be the most important one of all. For now, however, most marketers are searching for a solution. And it’s time for Google to respond.
For most of our clients, we have been using this metric for about 6 years. Happy to talk you through it.
Interestingly we have used this metric for SEVEN years. Just call us.
I have been using this metric for EIGHT! years, please fax me if you wish to catch up.
But seriously, I found this article really interesting. Not sure SOS should replace SOV though but its great food for thought. Ideally, it would be great to show how healthy your purchase funnel is going with both SOV and SOS metrics.
The definition of ‘categories’ is a great point to have in the back of your mind too.
I’m now waiting for the person who’s used it for 9 years to butt in 🙂
Sorry. I don’t get this. Isn’t the point of ESOV vs MS that you can establish a relationship between input and outcome? Here you’re comparing two outcomes. Doesn’t make any sense to me…
agreed – unless sos is assumed to be a proxy for sov, or ‘share of mind’? in which case it is very useful as a proxy for non-traditional focused categories, or those who cannot afford to compile sov data.
totally. Although I find it extremely hard to compare apples to apples or even define the category (fruits) – which is a point well taken from Mark’s piece.
It just leaves you with few actionable take-aways…
I’m really struggling to understand how SOS can replace SOV:
1) the same issues of category definition and brand scope (all of apple, only apple pay…) affect SOS calculation just as much as SOV (though I get the point about search data helping to define the scope).
2) SOV is an input. How much we put in. And can be somewhat controlled, albeit with a lag. E.g. if my market share is 20% and my SOV usually 20% but this year dropped to 15%, I can decide to over-invest next year to compensate.
SOS and SOM are both outcomes.
If anything, SOS can be used as a proxy for SOM (market share) for companies that don’t have access to (or can’t afford to pay for) market share data.
But in no way can SOS replace SOV.
@Mark Ritson, it would be great to get your thoughts on this one.
This will strike the fear of God into anyone whose been witness to the sales tactics of Google’s commercial teams.
2 different ways of calculating Share of Search seem to be emerging.
If you look back to articles about Share of Search circa 2012, the calculation seemed to be “number of hits your website receives for keyword X divided by total number of searches for keyword X”.
But with your comment that Share of Search looks at “how many people are searching online for the different brands in the category”, it suggests that the calculation may be something like “number of searches combining brand name X and category Y divided by total number of searches for category Y”.