We have seen before on Influence & marketing (1) that influence is more complex than it seems and deserves a much more detailed approach in order to distinguish influence from all of the other influence-like behaviors like for example homophily.
Now, let’s apply this to one of my huge passions: Marketing. After all, marketing and psychology are much alike. Is influence really what is affecting a purchase decision? We should differentiate between behavioral changes and behavioral tendencies.
How much influence is there in a purchase decision? That’s a tough question. For starters, we should know the effect of recommendation above the initial probability of purchasing. In other words, what is the added value of influence on my previous probability of purchase decision? will it have any effect?
This problematic can be solved by the use of a person’s social data, but where can we find all the social data about someone, where he indicates all his passions and interests? Would he be willing to share correct information? Well enough rhetoric questioning: You can use social media.
By analyzing the data obtained from social media, we can separate (or distinguish a little better) influence from other confounding factors such as homophily and measure what we really want to measure (it may sound obvious saying “measure what we really want to measure” but if you had a slight idea of the amount of statistical validity problems, you would recognize it is not a trivial matter).
Now imagine being the head of marketing strategy. Data scientists present you information that says when someone adopts your product, many of his friends will start using it too.
If 70% of the correlation has to do with influence, you will probably allocate a lot of the budget to peer-to-peer and viral, offering friends and family promotions and discounts. It appears the best strategy.
But instead if it is 70% related with homophily (or any other confounding factor), you would have spent a lot of the budget in useless promotions instead of realizing that peer to peer campaign is unlikely to work. Your inaccurate measurements are making you literally throw away money and devaluating your product by applying discounts when it is not needed. You should have been segmenting the market and targeting likely adopters through a different strategy like traditional ads.
Accept that it is not always influence that makes friends buy a new product: it may be caused by being in the same group, like early adopters who share a passion for the new technology. They already made their purchase decision so it is not an influence factor, it is probably all the other myriad confounding factors of influence instead.
I talked about influence in marketing for products, but influence goes way beyond. If you could really analyze what drives behavioral changes in your market the possibilities are endless: a new iPod-like disruptive product, political mobilization or even health care programs.
Remember if it seems like being influence, it probably is not only influence. Making this confusion can turn out to be a very bad marketing strategy and a very expensive mistake.