Share of Search as a Market Share Metric: Some thoughts on Les Binet’s presentation at EFFWorks
I listened to Les Binet with great interest as he talked about the work he has done with Google on SoS as a useful and predictive metric for market share. Having looked at a variety of social media metrics myself in my years at Ipsos I wondered what new insights I might discover.
Of course there were some things I thought were interesting and some I wondered about. I thought there were some great insights for the categories covered, and I haven’t been exposed to the full learnings or dataset used but I thought I’d put some ideas out there for discussion:
The WWHWW of search. Clearly SoS provides insight in some categories in a “business as usual” scenario. But which categories and why only sometimes? Les had selected auto, mobile phones and energy to demonstrate the metric. No fmcg or services. There’s a reason for that. The first thing to think about is the what, where, how, why and when of search.
· What. What do people search? Sometimes relevant search is based on category searches first: “best mobile phones”; “top 10 small cars” etc. Subsequent brand searches may well flow from category exploration. This is a good opportunity for brands to generate organic search, but those initiatives come with a wide range of effectiveness, especially for different categories. There is also a Masterbrand issue here. Many auto searches are for models, not makes. Les didn’t make it clear if those were rolled up into the totals for any of the categories but they must be included somewhere if they are to be relevant.
· Where. Google is not the only search. Aside from other search engines, depending on the category there are many intermediaries and large sites such as Amazon or dedicated sector platforms like TripAdvisor that capture a large amount of search, at all stages of the purchase cycle. The problem here is that it is different by sector, by market and even by brand. If Google were similarly representative of all search in all sectors and markets, you could live with partial coverage with “known unknowns”, but this is not generally the case.
· How. Partly covered by comments on where above and why and when below.
· Why. People search for many reasons. They may be looking for information, news, following up on a story or event, as the result of other marketing activity or to make a purchase. As Les made clear in his discussion, when the emissions scandal hit, the relationship between SoS and market share broke down. This was a particularly large publicity issue but this type of “noise” affects search all the time. It may be noise that is deliberately generated by the brand by a specific event, but it is often uncontrolled and often negative. These spikes may or may not turn into changes in share down the line by they certainly affect the immediate metric!
· When. Again, a broad topic, but I was very interested in Les’ comment that SoS approximated a measure in between brand salience and brand consideration. And that SoS is a lead indicator of market share. It makes sense for categories with longish purchase cycles but its hard to see that working in fmcg. I expect that it varies depending on the degree to which people seek information before purchase, the size of the investment and the level of repeat purchase that makes up market share. So, yes, search logically comes before purchase, but likely lags SOV or other marketing activity. Does that matter? It might if marketers start using it as a KPI — see below on the dangers of setting targets…
Assuming SoS is a “fast, inexpensive, immediate” metric vs other options
Les didn’t say, but in my experience, and if you look at the comments above the same issue applies in SoS as it does to any other behavioural metric or indeed any other dataset. There is a reason why the comment exists that “data scientists spend 80% of their time cleaning data and the other 20% complaining about cleaning data”. For any brand manager or CMO thinking they can simply take “free stuff” from Google Trends and create a low cost brand tracker, think about how much work needs to go into setting up a reasonable SoS measure, especially if you want to cover brand/Masterbrand or multiple markets. Even once you identify key issues to clean the search data you get, bear in mind that the initial data is not completely transparent and that there are lots of ways that “noise” can mess up the data you receive. There are ways to deal with that, but the point is each type of data has its own vulnerabilities and SoS is no exception. Les did point this out and certainly did not claim that it was a silver bullet, but we all know how these things play in when the C-Suite feels it is the latest idea they need to be on top of.
Which leads to my last thought…
What happens when metrics become targets or KPIs?
So, what will happen if SoS suddenly becomes a Marketing KPI? Les’ analysis shows that ORGANIC search currently acts as a lead indicator for market share. That is partly BECAUSE you are looking at it “in the wild” as it were. Once marketers begin to target search specifically to make their KPIs rather then to generate sales, I can think of all sorts of ways it could go wrong, can’t you? So, like light and quantum physics, observation fundamentally impacts what we are observing… I have discussed before the issue of inputs vs outcomes in marketing metrics and the need to be clear about what you can and cannot control, both for real world outcomes and when setting KPIs. Not to mention that when something out of the ordinary happens the first question is why?
In summary, I thought there were many thoughtful insights about SoS in Les’ presentation and I particularly liked the idea of being able to make a baseline comparison of relative brand strength in categories by comparing long run SOV with SoS under sustained advertising. I also welcome the fact that Les is completely open about where there is more to explore. And I think that is the point. There is no doubt that we need to find a better integration of different data types and sources to develop smarter ways to underpin marketing efforts and we need to keep discussing it to take it further. The challenge is dealing with complexity and thinking about the whole ecosystem vs. the desire for simple metrics that lead to simple actions.