One of the key findings from our Media Consumer Survey this year was the increasing lack of interest (and outright hostility towards) social media advertising. This is not really much of a surprise. While people have become more sophisticated in their use of social, splitting general narcissism (Facebook), professional narcissism (Linkedin) and photographic narcissism (Instagram) from sidesplitting banter (Whatsapp), advertisers have persisted with the blunt instrument of display advertising (see below). For the record, I have a long term girlfriend and don't currently operate a retail store...
...but perhaps Facebook is telling me something?
In any case, only 9 percent of people responding to Media Consumer Survey told us that they ever clicked adverts on social media. Furthermore, our focus groups suggested that the main reason that people are migrating to platforms like Snapchat (7 second narcissism) is because they can't be spied on or marketed to by advertisers.
This presents a problem. Many businesses I work with invest heavily in social marketing, often under the auspices of 'community building'. The return on investment of these activities is strongly negative - ROI is often in the single digit percentages. And the investments are increasing, driven by general migration to digital spend.
I don't expect that to stop. I also think that social media is a phenomenal advertising channel. But it requires a fundamentally different approach: a truly digital one.
The work we did with Facebook is one such method and formalises work we've done with other businesses on more bespoke data sets. What we did was twist the segmentation logic applied to digital spend, which is a development of the blunt demographic-by-channel approach taken on traditional channels (C2DE women watch ITV in the day, for example).
Our view was that on social media people are coming to share life events. Most of those are small - "I'm in the park... SELFIE!" - but some are indicative of longer term trends in lifestyle. And it's when big changes in lifestyle happen that people are most likely to change their service providers, favourite consumer and fashion brands, buy a car etc...
We borrowed the excellent work done in the US by Target as a basis, focusing in on mothers and looking at churn behaviour for mobile operators. What you find is that there are particular churn peaks that happen along the journey from finding out you're pregnant to having the child and then being a parent. And they vary by the age of the mother, whether this is the first, second or third child and so on.
What you build up is a model that segments mothers based on a number of factors and enables a radically more efficient targeting of adverts at them only at the key moments when they are likely to be amenable to changing their contract... which for over 60 percent of them is a once every seven year thing. Better still, you quickly come out with information about the type of tariff and handset that is most likely to appeal. For example, it turns out that mothers post significantly more pictures that non-mothers of the same age and demographic, consuming more data and presumably placing greater emphasis on picture quality.
A simple model enables you to create a model of roughly 1,000 segments that covers all mothers at different life stages. Into that model you then build A-B testing for each segment based on a relatively small set of offers, tuning the action-per-segment in order to optimise ROI and getting away from blanket blunt targeting that leads current (remnant, in this case I assume) advertisers to hit a 34 year old guy with a girlfriend and a corporate job with two completely meaningless dating/ payment platform ads. For the record, the more closed platform equivalents I've been part of in other digital businesses are roughly 10x more productive in terms of ROI than that achieved with 10 segment operations.
All of this is not just possible, it's actually really easy. We built the model in a matter of days. Facebook can create segments this granular, as can other platforms. The psychology is harder. It means abandoning a segmentation approach which has functioned well in traditional since the 1950's and marketing teams that are set up without the truly objective end-to-end performance metrics that enable the returns of new approaches to be effectively measured. That, we find is much harder. But it's definitely worth it.
You can read the Deloitte + Facebook report on the 1,000 segment operator here
And 2015's Media Consumer here
In any case, only 9 percent of people responding to Media Consumer Survey told us that they ever clicked adverts on social media. Furthermore, our focus groups suggested that the main reason that people are migrating to platforms like Snapchat (7 second narcissism) is because they can't be spied on or marketed to by advertisers.
This presents a problem. Many businesses I work with invest heavily in social marketing, often under the auspices of 'community building'. The return on investment of these activities is strongly negative - ROI is often in the single digit percentages. And the investments are increasing, driven by general migration to digital spend.
I don't expect that to stop. I also think that social media is a phenomenal advertising channel. But it requires a fundamentally different approach: a truly digital one.
The work we did with Facebook is one such method and formalises work we've done with other businesses on more bespoke data sets. What we did was twist the segmentation logic applied to digital spend, which is a development of the blunt demographic-by-channel approach taken on traditional channels (C2DE women watch ITV in the day, for example).
Our view was that on social media people are coming to share life events. Most of those are small - "I'm in the park... SELFIE!" - but some are indicative of longer term trends in lifestyle. And it's when big changes in lifestyle happen that people are most likely to change their service providers, favourite consumer and fashion brands, buy a car etc...
We borrowed the excellent work done in the US by Target as a basis, focusing in on mothers and looking at churn behaviour for mobile operators. What you find is that there are particular churn peaks that happen along the journey from finding out you're pregnant to having the child and then being a parent. And they vary by the age of the mother, whether this is the first, second or third child and so on.
What you build up is a model that segments mothers based on a number of factors and enables a radically more efficient targeting of adverts at them only at the key moments when they are likely to be amenable to changing their contract... which for over 60 percent of them is a once every seven year thing. Better still, you quickly come out with information about the type of tariff and handset that is most likely to appeal. For example, it turns out that mothers post significantly more pictures that non-mothers of the same age and demographic, consuming more data and presumably placing greater emphasis on picture quality.
A simple model enables you to create a model of roughly 1,000 segments that covers all mothers at different life stages. Into that model you then build A-B testing for each segment based on a relatively small set of offers, tuning the action-per-segment in order to optimise ROI and getting away from blanket blunt targeting that leads current (remnant, in this case I assume) advertisers to hit a 34 year old guy with a girlfriend and a corporate job with two completely meaningless dating/ payment platform ads. For the record, the more closed platform equivalents I've been part of in other digital businesses are roughly 10x more productive in terms of ROI than that achieved with 10 segment operations.
All of this is not just possible, it's actually really easy. We built the model in a matter of days. Facebook can create segments this granular, as can other platforms. The psychology is harder. It means abandoning a segmentation approach which has functioned well in traditional since the 1950's and marketing teams that are set up without the truly objective end-to-end performance metrics that enable the returns of new approaches to be effectively measured. That, we find is much harder. But it's definitely worth it.
You can read the Deloitte + Facebook report on the 1,000 segment operator here
And 2015's Media Consumer here
Comments
Post a Comment