Does the Ehrenberg-Bass Institute have any reports or research focused on India, and are there any differences in patterns or the application of the Laws of Growth local market?

S.D.
21 March 2025

Is there any research that can help us answer some questions on social media influencers and the impact that they have on the purchase funnel. Should these be used to build awareness or drive consideration?

To answer this question, it is important to keep in mind that the purchase funnel suffers from one key limitation, which is the risk of placing excessive emphasis on the middle and bottom of the funnel (driving consideration and conversion), overlooking the strategic importance of the top of the funnel, where wide-reach marketing tactics should be deployed to attract all types of buyers, including non-buyers – a vital requirement for brand and market growth. Influencers, like social media campaigns in general, tend to appeal to customers at more advanced stages of the purchase funnel, simply because: i) ‘followship’ and social media activity/engagement skew opposite to buying behaviour, with most users interacting with social media content being already existing customers/somewhat loyal to the brand behind that content; and ii) algorithms underpinning the selection of content served to people on social media (including content by influencers) automatically pick-up on existing intent and behaviour (e.g., search intent, content viewed, content interacted with, lookalikes etc.) in alignment with more advanced stages of the purchase funnel. Therefore, although influencers, like any other marketing tactic, should be used to build brand awareness (or any other top of the funnel strategy), they are feasibly more prone to impact the middle and bottom of the purchase funnel. 

Working with influencers calls for two sets of strategic decisions:

• Branding decisions – You need a ‘fit’ between the influencer and the brand. As the influencers marketing is coming of age, this process is resembling more and more a standard endorsement process, often agency-led via a brief. We recommend an evidence-based approach testing influencers as potential Distinctive Asset (hence, to figure out the potential for growing Mental Availability) and evaluating influencers’ potential as sales touch-points for social commerce (Physical Availability).
• Media planning decisions – Like any other media decision, classic advertising principles of reach and frequency lead plans and budgets, as reflected also in influencers’ rate cards. Hence, although it is widely accepted that engagement is quite low across most social media platforms (including when working with influencers) and that the number of followers can be misleading, managers still find themselves needing to evaluate theses typical social media KPIs and resulting trade-offs. Unfortunately, there is a lot of contrasting evidence both in scholarly research and industry reports on this point, with unclear advice such as:

• Choosing influencers based on the sheer ‘size’ or their followers’ base (e.g., preferring celebrities and media influencers) to maximise reach > problematic strategy because influencers of this size have limited to no engagement for their size. 

• Choosing influencers based on the engagement potential and the intensity of interactions with followers (e.g., selecting everyday influencers and content creators) > also a problematic strategy because it implies repeated exposure to the brand but just with a very small (near nil) % of category buyers. 

The evidence used to provide these recommendations is also based on small datasets based on single instances of influencers and social media platforms, which limits generalisability and benchmarking. Between 2018 and 2022 the Ehrenberg-Bass Institute has addressed this issue creating and analysing a unique set of data with the following characteristics: 

• 180 influencers with varying size (i.e., in terms of the number of followers), including an equal number of celebrities/mega influencers as well as micro/everyday influencers.
• 5 social platforms including mature vs. growing platforms and platforms undergoing significant changes at the time of data collection (Facebook, Twitter, YouTube, Instagram and TikTok).
• A sample of 35,000+ posts, with engagement rates randomly derived from 20 posts p/ influencers (industry stats usually use the last 5 posts).
• Double Jeopardy analysis to explore observed and expected trade-off between the share of followers and the engagement rate of each influencer within the platform/sample and within specific areas of expertise such as beauty and fashion, travel and lifestyle etc.
  

The main findings that we extracted lead to the following recommendations:

1. Decisions involving influencers should be platform specific and industry/areas of expertise specific.
2. DIY: For greater generalisability of findings and benchmarking purposes, the number of followers and the engagement rate can be robustly recorded from publicly available data, which can then be analysed using a classic law of growth: the Double Jeopardy.
3. For mature platforms such as Facebook, Twitter and YouTube, there is a reverse Double Jeopardy pattern: influencers with more followers return the lowest engagement of all. Hence, on these platforms, it is best for brands to collaborate with smaller influencers and content creators.
4. For growing platforms such as TikTok, there is a standard Double Jeopardy pattern: influencers with more followers have also greater engagement. Hence, in this instance, brands can choose influencers based of wide-reach potential.
5. Platforms undergoing changes like Instagram reveal no clear or predictable trade-off between the number of followers and engagement rates of influencers.

   Therefore, it makes sense to choose influencers based on their wide-reach potential. 

Recommended read: Pourazad, N., Stocchi, L., & Narsey, S. (2023). A comparison of social media influencers’ KPI patterns across platforms: Exploring differences in followers and engagement on Facebook, Instagram, YouTube, TikTok, and Twitter. Journal of Advertising Research, 63(2), 139-159.

L.S
5 Sept. 2024

Within the context of Specialized Nutrition (especially baby formula), does the concept of Double Jeopardy hold true? This question was triggered by the Double Jeopardy component of “bigger brands get bought more often”. Since the baby will need to eat no matter what we were wondering if this concept remains relevant.

We do see certain repertoire markets to exhibit patterns that are closer to subscription markets (i.e., a closer loyalty towards few brands than what we would observe in other repertoire markets), however, we have seen that Double Jeopardy holds in B2B context, and in this case, baby formula also. Although the factor differences are smaller, we can still see the law applies in this category – e.g. very similar purchase frequency between the largest and the smallest brand (6.8 vs. 6.4) but there are bigger differences between their penetration figures (e.g., 26 vs. 4). We typically see the differences in purchase frequency and penetration to be larger than the examples that I illustrate here.

It is very likely that mothers would get their advice on what formula or nutrition to give to their baby from their paediatricians, thus it is very likely that bigger brands are also recommended by a lot more paediatricians, more frequently. So, in order to grow the baby formula brands, the importance of physical availability and mental availability remains true across paediatricians and parents also.

A.T.
23 November 2021

I have a question about the Double Jeopardy law that I hoped one of you could help with.

We accept that the double jeopardy law applies to consumers’ brand buying behaviours. Does the law also extend to their purchase within categories e.g. alcohol and sub-categories like beer?

In a declining sub-category, will falling penetration (of beer drinkers) see changes to other metrics: e.g. a drop in the frequency of buying the sub-category and an increase in cross sub-category buying / declining loyalty in a similar way it does for a brand?

We have seen penetration of beer drinkers fall over the last few decades, and are in the process of determining the size of the influences. Can we attribute any of the sub-category decline in sales to the double jeopardy law caused by the declining beer penetration or does this only impact brands’ shares within categories? If we can, is there a calculation we can apply to our data?

A good question. Yes, the Double Jeopardy law applies to categories and the sub-categories ‘within’ them; effectively treating the sub-categories as brands. In the broad alcoholic drinks market, product types (sub-categories) that have lower total market share will have lower penetration and somewhat lower purchase rates / loyalty (purchase frequency, share of wallet / throat). If a sub-category grows or declines it will be reflected in higher/lower penetration and associated repeat purchase rates (and all other loyalty metrics) directly in line with the Double Jeopardy law.

So yes, if Beer is declining in sales, then both penetration and purchase rate (e.g. # purchases or consumption occasions etc.) decline, penetration more so than purchase rate.

While it is definitely possible to calculate these sub-category performance metrics for the the sub-category ‘brands’ by fitting the Dirichlet Model to the overall alcohol market, unfortunately it is not a matter of a ‘simple calculation’. But it would be useful to compare over two time periods e.g. 2 years ago vs now, actual beer penetration like % of males 18-60 who drank beer in last month / not, and how many occasions or purchases including zero – so you’d get the full distribution of buyers from non, light, medium, heavy and compare then and now.

What deviations from the Law of Double Jeopardy are you aware of?

We see Double Jeopardy everywhere, from the buying of soap to the watching of soap operas.  The more popular items have higher penetration (more buyers) and slightly higher loyalty metrics (both attitudinal and behavioural).  Because Double Jeopardy is so widespread we call it a scientific law.  But this isn’t to say we don’t know of a few conditions where the law bends:

Some brands have restricted distribution, where they are available they are quite popular but there are large areas where they unusually lack physical availability for a brand of their popularity.  Private labels (retailer’s own brands) are the classic example, because they are only stocked in one retailer’s stores.

Another example are brands that have unusually high differentiation, such that they completely do not appeal to some category buyers.  Pizza with pineapple on it (or anchovies) perhaps.  Extremely low quality or super expensive brands may also deviate.  Or unusually large (or small) pack sizes.  But probably the most common surprise is how such features cause such small departures from Double Jeopardy.

A different sort of deviation is known as “change of pace”.  Here lots of people buy the brand but repeat-purchase is low.  A typical example is seasonal brands (e.g. easter eggs in the chocolate category).

Do you know of a deviation from Double Jeopardy?  Please tell us.  We’ve got bottles of Champagne to give away to those who document something new, currently unknown to marketing science.