Institute of Practitioners in Advertising (IPA) hosted a webinar titled A New Way to Track Consumer Demand on November 18, 2020.
Parts 1 and 2: were presented by Les Binet, Group Head of Effectiveness, adam&eveDDB
Part 3: was presented by James Hankins, Strategist, Analyst & Planner
Moderator: Janet Hull OBE, Director of Marketing Strategy, IPA
A recording (2 hours 41 minutes) and slides are available on the IPA website.
Acronyms:
- SOS = Share of Search
- ESOS = Extra Share of Search
- SOV = Share of Voice (i.e. ad spend)
- ESOV = Extra Share of Voice
- SOM = Share of Market (aka market share)
The following write-up is from my handwritten scribbles while listening to the webinar.
Part 1: Les Binet. Share of Search as a Brand Metric
Traditionally marketers use surveys. Les in skeptical of surveys. Weak correlation between survey data and sales. David Ogilvy quote: “People don’t think what they feel, don’t say what they think, and don’t do what they say.”
We’re talking about organic search queries. This is not about paid search advertising.
Share of Search = (Searches for Brand X) / (Searches for all brands in category)
Data comes from Google Trends (16 years of data). Indexed, not raw data. But comparable because it is indexed consistently.
Econometric modeling
Compare share of searches (have weekly data) with share of market (have monthly data). But the best relationships are quarterly data.
“pooling” – pooled econometrics
Share of Search is a leading indicator of market share (it correlates in advance of market share) and it may have some predictive power. It is not causal. It tracks interest at the top of the purchase funnel.
It’s a leading indicator with long lead times.
- Cars – up to 12 months
- Mobile phones – up to 6 months
- Energy – up to 3 months
Share of search is a proxy for market share.
For many businesses there is no market data (like Nielsen data). So SOS is a very useful proxy.
Useful as a leading indicator for categories with long lead times where purchases are pre-planned (like cars).
FMCG is less than a quarter of advertising.
BASE LEVEL. The level of search even when people aren’t searching for you much = EQUILIBRIUM LINE.
- If SOS is below equilibrium, SOM tends to fall
- If SOS is above equilibrium , SOM tends to rise
- ESOS = SOS – SOM
- Les pronounces ESOS like “e-sauce”
It worked for 22 of 23 brands in automotive market.
It worked for 10 of 12 brands in mobile phone market.
Energy is more problematic, for reasons forthcoming in the presentation.
SOS is about DEMOCRATIZING market research. All you need is Google Trends, Excel, and a brain.
Some factors that may throw things off…
- SOS is more problematic for small brands because there is more noise in the data.
- BAD PR screws up your search data. For example, with the VW emissions scandal there was a negative correlation for a few months. Also, rising energy prices resulted in more searches but lower sales. Potentially, trying to determine positive or negative search sentiment (through social media listening) could adjust for this.
- Les referred to the Google’s Messy Middle article. (The full report, a 98-page PDF document titled Decoding Decisions: Making Sense of the Messy Middle, can be downloaded.) People start their search at the top of the funnel based on mental availability of brands. But through this exploration and evaluation process, they may change the brands they are considering.
- Conversion rate vs. relative price. SOS divided by share of sales. Potentially enables you to map the demand curve.
- Not practical as a forecasting tool in categories where pricing is changing dynamically, e.g. energy, car insurance, things where people use comparison tools. But still relevant as a measure of share at the top of the funnel.
- Less-Planned Purchases.
Summarizing Part 1:
- SOS is a leading indicator for market share.
- SOS is a useful indicator of latent demand for a brand.
- Measure of interest at the top of the funnel.
- Early warning of changes in demand.
- Some problems, like negative PR.
Part 1 Q&A
In general, market share is by value, but in the car market there is excellent data on volume. So it depends on the category.
SOS is a proxy for mental availability. We know mental availability is a driver of purchases.
Portfolio (master brand advertising) is difficult. Portfolio is also difficult to SOV.
May not work for all FMCG (fast-moving consumer goods). James says it works for cosmetics.
Part 2: Les Binet. SOS as an Advertising Metric
What drives share of search?
- SOV > SOM: brands tend to grow over next 2 years.
- SOV < SOM: brands tend to shrink over next 2 years.
(SOV usually means ad spend. Sometime is means awareness.)
Does SOV also affect Share of Searches? Yes. Strong correlation with autos, mobile phones, energy.
- 10% ESOV increase SOS +0.36% cars
- 10% ESOV increase SOS +0.38% energy
- 10% ESOV increase SOS +0.43 mobile
- Generally about 0.4% increase
Use econometrics. Pooled regression models.
Short-term and long-term effects.
- Short-term: burst, decays within a month. (activation).
- long-term: smaller which decays slowly. Long-tail. (brand building).
Ads are still influencing search ~2 years after exposure.
- 18-month longevity effect… 2.6 long-term multiplier… Energy
- 21-month longevity effect… 2.8 long-term multiplier… Cars
- 39-month longevity effect… 2.9 long-term multiplier… mobile phones
The long-term effects of advertising tend to be 2-3x the short-term effects.
Long-term effects accumulate over time.
Ad bursts with 10% SOV. 5 bursts. Increases base level over time. (Same effect with sustained advertising as with bursts)
- 30% Short-term, 70% long-term…. Energy
- 36% Short-term,64% long-term…. cars
- 39% Short-term,61% long-term…. Mobile phones
Binet and Field 60:40 Rule (actually 62:38)…. So, above numbers are very similar. See Effectiveness in Context by Les Binet and Peter Field.
Long-term increases in mental availability.
People search for things they might buy. (Circle graphic) SOM <–> SOS
Part 2 Summary:
- Most searches are not driven by advertising. There is a base level of demand.
- But advertising can increase demand over base level.
- Advertising has both short-term and long-term effect on search behavior.
- Effects are similar for all categories studied.
- SOV is a key metric.
- 10% SOV increases SOS by 0.4%
{ I was confused about “10% SOV increases SOS by 0.4%.” I thought perhaps there was a typo and that it should say 10% ESOV increases by 0.4%. Or maybe I’m wrong and it means 10% SOV increases SOS modestly (0.4%) above baseline compared with no ad spend?
Adding to my confusion:
part 2, slide 6: “In each category, 10% extra SOV immediately increases SOS by 0.4%.”
part 2, slide 19: “A burst of advertising produces an immediate increase in brand searches. 10% SOV increases SOS by 0.4%.”
I asked Les Binet about this. }
{ Les Binet kindly replies:
“No, the charts are correctly labelled. ESOV is the relevant metric when talking about the long-run equilibrium position, whereas these charts are talking about short term transient effects. However the two are related, and share the same coefficient.” }
{ One more bit of insight from Les Binet on Twitter.
“I would caution against using SOV with high frequency data. It’s only really useful as a longer term metric.” }
How to fit this together. Graphic with lots of arrows…
- Customer Base and “Brand Strength” create a base level contribution to share of search
- Share of Voice has short-term and long-term effects, each of which increases share of search
- Share of search leads to share of sales
- Other Factors: Relative Price and Physical Availability affect Conversion, which also leads to share of sales
- Share of sales loops back to customer base at the top of the graphic.
Conclusion. Search is not the ONLY thing that affects market share. There’s conversion (from search to sales): the messy middle, relative price, physical availability, customer service.
Part 3: James Hankins. He notes that he’s a strategist not a quant.
See James’ blog post The most important metric you’ve never heard of.
James mentions Good Strategy Bad Strategy: The Difference and Why It Matters by Richard Rumelt. . Diagnosis + Guiding Policy + Coherent Actions.
SOS = SOM. But this is not a law. It requires critical thinking about what the data is telling you.
Smoothing: (12-month moving average Brand A) / sum of (12-month moving average Brand A, 12-month moving average Brand B,… 12-month moving average Brand X)
Google Trends only allow you to look at 5 brands at once. So do first 5. Then replace four of them and repeat.
Most people don’t have access to market share data (Kantar, Nielsen).
Other factors besides ad spend: NPD, Cap-ex, PEST factors [political, economic, social, technological]
S-Curve and the Rogers Curve
Q&A
Les: You can define the category any way you want.
Les: To calculate SOV (ad spend) you need access to Nielsen data.
Articles of Interest 