TV, retail, advertising and cascading collapses

Two months ago I gave a presentation talking about the fundamental structural trends in tech - on the one hand, we talk about what we can build on the billions-scale platform that the smartphone gives us now, and who can compete there with GAFA, and on the other, we wonder what the next decade-scale, billions-scale platforms or trends might be - machine learning, autonomous cars and so on. 

However, there are also some important current trends that don’t necessarily fit into those over-arching narratives, but might have almost as much impact in the next, say, five years. I’ve written some long pieces about what might happen to TV, and to discovery in general, but I haven’t written a single unified theory of the future of retail (and I’m not sure anyone could), nor advertising. Yet there is a set of accelerating and interlocking changes happening in TV, advertising and retail that could lead to some interesting discontinuous and cascading effects. So, in no particular order...

TV viewing is finally starting to unlock and move away from linear and cable bundles, especially in the USA. That ought in due course to have some effect on TV ad inventory and rates: some viewing will go to places without ads, or where ads are sold very differently, overall viewing might fall (though this seems unlikely), and viewing will be distributed in different ways - probably, as tends to happen with digital, the curve will get much steeper. The hits are bigger and the mid-market stuff that’s only viewed because of where it is on the linear schedule will suffer. So, TV ad inventory will probably look different. 

As ecommerce keeps growing, at some point we will start to see certain retailers disappear - it’s common to say there are strong parallels with newspapers, in that they have a fixed cost base, falling revenue, and the wrong assets & skills. When internet reading or internet buying was 5%, it felt as though it might be additive to newspapers or retails - at 10 or 20%, as it is now, it becomes an existential problem. That is, at a certain point they stop being able to cut costs at the margin and start closing stores, or radically changing format etc. So, rhetorically (or apocalyptically) speaking, when Sears and Macy’s go bust, how many malls do they take with them, and how many other retailers that might have been doing fine on their own will go or lose a lot of their footprint because of that? And, where were those retailers advertising? What was their TV budget? How much of this is self-reinforcing - the more you buy online, the more you buy online? Conversely, did Aeropostale’s customer base go online to buy all the same kinds of clothes when the stores went bust, or buy different clothes, or buy different things? That is, do retail failures caused (partly) by ecommerce cause further ecommerce adoption and further failures?

As buying goes online (regardless of how much the previous points happen at all), buying patterns change. Really, whenever you change the channel, buying patterns change - people do not buy the same in malls as in department stores, nor in department stores as in small shops, and so people do not buy exactly the same things online as they do in supermarkets or malls. The channel shapes what is bought. For FMCG in particular the models of shelf space, eye level, end-caps and marketing support etc don’t exist online - there are new mechanics for driving purchasing online, whether that’s on Amazon or Instacart/Ocado etc. This ought probably to mean fewer brands, and different ones, which again flows to advertising and TV. Do you spend the same on fewer brands to stand out online, or spend more, or less? When you say ‘hey Alexa, I need more soap’, what brand do you get sent and why? Meanwhile, the same applies somewhat even at the other end of the spectrum - sales distribution on Net A Porter must surely look different to that in physical retail.  

Then there’s machine learning, and especially image recognition. Ecommerce has always been good at retail as logistics but bad at retail as discovery. Now suppose I can connect a retailer to my FB or instagram feed and it can look at the clothes I’m wearing in photos and make suggestions? Suppose you go on eBay and buy the last ten years’ of Elle Decoration and drop it into Google Brain, and then wave your phone at your living room and ask what cushions or lamps you would like? Those magazines (or anything else laboriously and manually ‘curated’ by experts) are effectively now all structured data, or will be - you can extract the objects from the images and names from the captions and work out what data about taste and association is encoded in every editorial shoot. Five years ago this would have been science fiction, but now we know exactly how to go about building it. Suppose I put a bunch of HD cameras in the right parts of Berlin and Brooklyn and track what people are wearing, entirely automatically, and then see what of that shows up in middle America in a year, and then apply that pattern matching to what people are wearing in Berlin now? Again, no longer science fiction. Computers are going to be able to analyze images and video the way they’ve been able to analysis text and numbers for 50 years, and indeed much better, with much more sophisticated pattern analysis. So - what does that do to how ecommerce can make suggestions? And what does that do to retail and advertising?

There’s a famous Jeff Bezos quote that ‘your margin is my opportunity’ - right now Amazon is building a billion dollar ad business in its own search results, but I suspect he also looks at the $500bn that’s spent every year on advertising and the further $500bn that’s spent on marketing and sees money that should be going to lower prices and same-day or 1-hour delivery. P&G spent 11% of revenue on advertising last year and plenty more on marketing. What will that look like in 10 years, where will it be spending it and how will people be buying?