Share of Search - a new way to track brands & advertising
ฝัง
- เผยแพร่เมื่อ 28 พ.ค. 2024
- For the last few years, we've been experimenting with Share of Search as a tracking metric. It's fast, cheap and predictive. Not only that, it can measure both short and long term ad effects. This video gives a brief overview of the technique and what it tells us about how marketing works.
This is bloody amazing. I’m working with a retail brand that has a competitive set that is mostly privately held, so there’s no publicly available spend / data on advertising. Plus, they have never tracked performance. This is so so good to retrospectively view brand activity / mental availability.
Thanks heaps Les. Amazing.
Thanks Les for this research.
Very interesting, very useful. Thank you!!
Excellent lecture!
Dear Mr. Les Binet,
Thanks for this video, very clear & appreciated.
I would like to validate the understanding of your theory on the equilibrium level of SOV vs SOS (24:18 min).
Can we say that the brands are permanently out of balance, and that the equilibrium line is only a concept towards which the marks tend but are never actually there?
Furthermore, if a brand is out of equilibrium, above or below the straight line, will it converge towards the straight line in a perpendicular manner? the most direct way to reach the straight line.
Many thanks in advance if could share you thoughts on this.
Thanks Les for sharing this great thinking and video! Two questions: 1) In your formula, is "searches for brand" defined as only searches for the brand name itself (e.g. "Ford"), or all searches that include the brand name (e.g. "Ford Explorer," "Ford SUVs," "2022 Ford truck lineup," etc.)? And 2) is this search data relative (e.g. Google Trends) or absolute (e.g. Google Keyword Planner)? What tool/source did you use to generate the search data? Thank you so much!
It’s all searches that include “Ford”, eg Ford Explorer. Relative data from Google Trends.
How would you measure share of voice? Great video.
Share of adspend, eg via Nielsen. Not perfect, by any means!
Would you adapt this analysis for a product with a high consideration and long (time) sales cycle like a mortgage? Would the time gaps make identifying forecasts or equilibrium difficult?
I think it would work well for something like that. It works for cars, where the purchase cycle is several years and consideration is high. Low consideration / high frequency purchases are probably more of a challenge.
This looks interesting! Would you care to comment as to why you chose to keep the (extreme) outliers in the data in this case, at least to the naked eye, they look to affect the relationship a lot.
Which chart do you mean?
Love your work on this, Les. Couple of questions:
- How do you handle a brand that's already in the vernacular (e.g. nutmeg - the roboinvesting service)
- What if your business is competing against a subset of a brand (e.g. let's say you're a credit card that gets you the best foreign currency rates, brands like Monzo also do this...but Monzo does so much more [and surely just picking up on "Monzo travel card" searches won't be representative]?)
Appreciate you won't have all the answers, but would love to get your thoughts
Using categories can help - Eg Nutmeg within Finance. And these issues help explain why the relationship between SoS and SoM is different for each brand.
@@lesbinet7977 Categories is a good shout!
0:00 @@lesbinet7977
Thanks Mr. Binet for sharing your insights. Could take it up immediately on Google Keywood Tool and also refered to your work with Prof. Mark Ritson and Mr. Peter Field on B2B Marketing Growth Formula. A query, if you don't mind and my be stupid query also, when doing share of search in a category the brand may have been searched for Careers and Support. When share of searches is calculated, is the careers and support influence removed to take only pure product category?
No. We’re interested in the degree to which the brand is in people’s heads for any reason. It’s a kind of proxy for mental availability.
How do you know did you know that the baselevel was 35% of SoS for Apple when no advertising was done? How do you rule out that no advertising was done by apple or the other brands in the same period.
It’s a mathematical projection of what we expect share of search WOULD be with no ads, based on a regression model.
I need practical help - I can find in Google Trends only relative figures (%, peak as 100%), but not absolute number of searches for particular words. Could anyone help me? I have almost no experience with Google Trends.
You can use Google Keyword planner tool for estimated numbers. For more accuracy, Ahrefs/Semrush.
You don’t need absolutely numbers, just a measure of relative interest.
All the categories are ones that need a bit of research. Do people search much for FMCG goods? Can’t imagine i’d ever search for Ketchup, or toilet paper..
Agreed.
The relationship is likely to be weaker and noisier for FMCG, with price movements in particular making it harder to detect. But some researchers have still found a relationship - James Hankins found it worked well for tea, for instance.