We, as practitioners of professional, commercial photography love to have scapegoats to blame for shrinking markets, eroding margins and disappearing fees. In the recent pass we've blamed: the smartphone and its users, the "soccer" moms with cameras, and, those scurvy dogs, the weekend warriors who would work for free in return for the experience. We've railed at people who are just entering the market for not understanding that it is smarter to charge for the value of a piece of intellectual property to the client instead of gauging prices by how much you might need to make just to buy groceries and keep the heat on.
But in the same way that robotics and automation will ensure the society-wide elimination of repetitive human jobs and fill those positions with machines that don't need breaks, don't make (many) mistakes and don't need retirement accounts, I would conjecture that the erosion in photography markets is a direct result of the granularization of the advertising channels (display media) and the ability of marketers, via the application of psycho-metrics (thank you Isaac Asimov via the Foundation Trilogy...), which allows much more precise message targeting. The value of an advertising image is based on the its effectiveness times its use over large numbers of impressions. To be useful to a very wide audience an image must be more and more "all purpose" which dilutes its impact and efficiency in prompting action (or, in the case of elections; inaction). The more focused an image is toward a defined collection of customer quirks and attributes the most effectively it will reach its demographic target, the fewer impressions it will have, and the CPM will skyrocket.
Going forward you and I will not likely see the same advertisements when we search the web. Our buying habits, incomes, political leanings and our basic personalities (things like our introversion or extroversion) will be analyzed (Thank you! Smart Phone, Instagram, Facebook, Twitter) and data mapped and we'll have our attention focused toward ad messages that will be resonate with our oddly unique points of view, as defined by rigorous data-mining of all our public and private actions and choices. If data from my phone shows that I only drink conventional drip coffee at Starbucks then, over time, I will stop seeing ads for Mocha Frappucinos and my ads will point me toward products and services that algorithms have postulated as within my particular buying preferences.
Your predilection for Pumpkin Spice lattes might swing your ad parade toward various seasonal drinks with high sugar and accentuated taste features. Different images will be used to target the messages more precisely. While more men than women lean republican a much tighter determiner of voting practice might be to look at car preferences instead. The more one prefers American Luxury cars the more precisely likely it is that a psycho-metric marketer can reach out to the same audience that bought a Lincoln in order to sell them a political candidate.
But the bottom line is this, instead of using one overarching image to reflect the paid advertising of a branded item a savvy marketer now requires a much more tightly targeted collection of images in order to address each of a myriad of smaller, more discrete populations. That means the cost of a photoshoot, or licensing package, can't be spread over a large campaign (greater cost spreading drives down the per unit cost of imaging and makes it affordable to pay more). If a marketer requires 60 images instead of 1 in order to better take advantage of a much more rigorously defined collection of target subsets then each of the 60 images has less individual value in isolation than an all purpose image would have had in the days when mass communication ruled.
It also means that there will no longer be uniform stylistic attributes that coalesce ubiquitously in mainstream marketing. Each tiny market segment will respond differently to poses, color choices, graininess, propping, styling, the creative narrative, and even casting. An older coffee drinker may be more effectively subliminally manipulated by an image of a woman of a specific age and income class, in a quiet environment, drinking coffee while reading book (novel). This may trigger his unconscious desire to visit a coffee shop ( the visual stimulant, in addition to his decades long caffeine addiction). He might be motivated the promise of a quiet respite from home or office and the off chance of meeting someone compatible. By the same token a slightly younger coffee drinker might respond to an image of a coffee shop bustling and full of people, who almost all have headphones on, and who are looking at laptop and smartphone screens. The imaging may be driving their desire to go to the coffee shop for greater social connection. One image promises one outcome while the other speaks almost oppositely to a different demographic target. Both are more effective when used concurrently than one solo ad that tries to hit all markets and ultimately fails to spark a tipping point reaction in any market.
But, the images will need to be differentiated in many ways (both content and style) and this is a cost to the advertiser. Since budgets aren't wholly elastic, and ad insertions have to be made in many more channels, and the cost of designing multiple adds is considered, the soft spot for cost optimization points directly at any external, third party cost. And that would be the photography.
Additionally, the drive to reduce the cost of photography per ad also drives the whole scale rout toward using enormous amounts of low cost stock photography. Which can now be easily modified to conform to the parameters expressed by the data-mined and interpreted information about the sub-groups.
In the near future A.I. will work with data mining to discover just which images resonate with you, personally, the most. At that point all the thousands of reference points you've provided, and continue to provide, to the cloud of advertising research will be used to construct CGI ads (which require no actual photographer or actual models) that speak exactly to the visual+emotional constructs you have in your own head. To see them constructed and played back to you means you will feel a deeper emotional connection to the advertising in the belief that you and they are "on the same page," and that they "get" you. And at that point they will certainly have gotten you.
But if you are sitting back smugly in your chair because you "saw the writing on the wall" and dived into video, or some related field, you might want to start studying up on artificial intelligence video editing and artificial intelligence scripting, and technical writing. And consider the implications of face detection, smile detection and automatic camera systems. Far fetched? No! All here right now.