Facebook’s Moonshot: Reinventing Retail Beneath Amazon’s Feet
A speculative analysis of one of the world’s largest corporate giants
Facebook’s outward purpose may be the inspiring catch cry to ‘bring the world closer together.’ But its internal mission is almost certainly something a little more commercially pointed and likely aimed squarely at the burgeoning activities of its chief rival Amazon.
With over two billion users, Facebook may have one of the largest active user bases on the planet. But Facebook’s revenue still lags well behind Amazon, whose revenue in Q4 2018 surpassed Facebook’s entire 2018 earnings. As Amazon’s reach continues to extend well beyond the realm of retail, Zuckerberg has no choice but to accelerate Facebook’s push into retail and attempt to close the revenue gap.
According to ecommerce expert Brittain Ladd, Facebook’s best bet is to buy an off-the-shelf solution, such as an eBay or Rakuten. But, Zuckerberg recognises the existential threat Amazon poses to his business and knows that simply manoeuvring Facebook into Amazon’s shadow will not be sufficient to guarantee its future. His only option is to go on the attack. To try to outmanoeuvre, destabilise, and ultimately out-retail Amazon.
Facebook’s Strategy to Out-Retail Amazon
Facebook’s strategy to out-retail Amazon is already well underway. It’s being built on the shoulders of the one towering advantage it has over its rival: the personalised user interface.
With the help of its AI-powered prediction engine, Facebook harnesses its vast, active userbase and endless reserves of human data to make billions of self-improving content recommendations a second. This process generates a unique and highly personalised interface for each user, which generates high levels of stickiness and allows it to farm and sell off human attention at unprecedented rates.
For some time now, Facebook has been tinkering and experimenting with the nature of the personalised user interface to more directly influence and facilitate the purchase of products from within the newsfeed. Facebook’s clear end-goal is to convert the vast stocks of attention into endless streams of ‘clippable’ transactions.
This sounds relatively straightforward, but Facebook has long grappled with the fundamental challenge of shifting the balance in the newsfeed from ‘serving users more of what it knows they want’ (content recommendations) to ‘making users want what it serves’ (product recommendations). The latter involves a higher form of influence that can create deleterious interference in the personalised user interface. This makes everything far less sticky and drives down the all-important ‘time on site’ metric, which its revenue-generating customers find so alluring.
Adding to the challenge, Facebook’s advances in AI-powered predictive personalisation appear to have made it harder for product recommendations to flow seamlessly into the newsfeed. The pinpoint accuracy of its content recommendations has upped users’ sensitivity to recommendations that stray too far from their sphere of immediate need. Given that product recommendations are a lot more about what the seller wants, even the most relevant product recommendation, when fed into the most personalised space on the planet, is at high risk of being parsed as an intrusion.
Facebook’s recent and very public volte-face to refocus the newsfeed around ‘meaningful interactions’ highlights just how sensitive its users have become to heightened commercial interference. It also points to just how difficult it will be for Facebook to dial up the visibility and frequency of product recommendations, all without undermining its existing revenue model.
Moving forward, Facebook must ensure product recommendations are stitched into the user interface in a way that consistently and reliably improves, rather than detracts from, the quality of the personalised experience. The market has to make the meeting place a more rewarding place to be.
With advancements in neural networks and deep learning, Facebook is edging ever closer to this goal. Facebook’s early AI engine, ‘FBLearner Flow,’ enables the current system to orchestrate the deployment of billions of highly personalised signals.
This already feels less like intrusive advertising and increasingly more like highly useful notifications. But, it’s the more recent rollout of its ‘Deep Learning Recommendation Model’ (DLRM) that looks capable of generating the kind of improvements that could lift its product recommendations out of the ‘occasionally useful’ box and into the ‘always indispensable’ zone.
This order of improvement is vital if Facebook hopes to dislodge Amazon’s stranglehold on ecommerce. Product recommendations that are ‘always indispensable’ no longer look like advertising, or even behave like a recommendation. Instead, they become something far more powerful and game-changing. They become the basis for a ‘guaranteed transaction event.’
Once Facebook is able to accurately predict, at scale, where in the market ‘guaranteed transaction events’ exist, Zuckerberg will have successfully reinvented the ecommerce game beneath Amazon’s feet.
Facebook’s Intelligent Market System
Facebook’s vision for the future of ecommerce isn’t a single, fixed destination that waits for users to visit, knows what they need, and knows what products are available to satisfy that need. Its vision is for an ‘intelligent market system’ that perpetually monitors, analyses, and anticipates its users’ fluctuating needs.
The system will check these needs against its inventory of product solutions to serve up hyper-personalised recommendations. These recommendations will be so indispensable that they not only create a more rewarding user experience, but they also result in a guaranteed ‘clippable’ transaction.
This vision of a highly responsive, efficient, and automated market system may just be within Facebook’s grasp. But, its early struggles suggest that even the most advanced AI-powered prediction engine might not be enough to overcome the one thing standing in the way of generating ‘always indispensable’ product recommendations. And that’s the market’s ability to generate enough indispensable product solutions in the first place.
To elevate its recommendations to the zone of indispensability, Facebook will have to do something no other organisation in the history of commerce has likely ever conceived of, let alone attempted. Facebook will have to rewire the underlying dynamics of the market system.
How Facebook’s Ad Impressions Work
As is the case for most companies, Facebook operates at the mercy of market forces. Most notably, it lacks the ability to choose what products and ad impressions its commercial customers want pushed into the paths of its users. That’s why Facebook refuses to simply dump ad impressions evenly across its userbase.
In an effort to protect the sanctity of the personalised user interface, it firstly serves ad impressions to users it predicts will make more useful recommendations. Of course, useful recommendations not only minimise interference, but they also lead to a higher probability of transaction.
However, with such a heavy concentration of spending from the market’s biggest businesses, each with a relatively narrow mix of highly commoditised product solutions, its prediction engine quickly runs out of best-fit users. The prediction engine is then forced to spill intrusive ad impressions out into the paths of a large majority of its users.
This scale of ‘negative prediction error’ is unavoidable given the nature of its business model. But the problem for Facebook is that it’s also eating away at its engagement model and slowing down its efforts to reinvent ecommerce.
Facebook mitigates against this issue with the use of hyper-personalised messaging and other forms of cognitive trickery that can manipulate a user’s perception of a product recommendation, In some cases, this forces users into unnecessary purchases.
However, Facebook knows better than most that if the product fails to align to its users’ authentic needs, then no amount of message personalisation can make up for a lack of product personalisation. These recommendations will eventually be detected as intrusive, leading to widespread interference across billions of personalised user interfaces.
Attempting to bend the will of its users to process the market’s existing crop of product solutions as seemingly indispensable is not sustainable. It’s also certainly no way to go about transforming the ecommerce game. This leaves Facebook with one last, but logical solution. It must shore up and deploy a new source of indispensable product solutions for its users.
Facebook’s New Product Solutions
Unfortunately, today’s global market system has become a wholly unreliable source of indispensable product solutions. In fact, it has become pretty inefficient at generating products that align with the best interests and needs of individuals or society at large.
This is because the market is largely dominated by corporations that have become desensitised to consumer needs and are hyper-sensitive to the needs of shareholders. Driven by the profit motive, they’re structured to efficiently mass produce products and suppress consumer needs, rather than to rapidly innovate products at scale in line with the emerging and authentic needs of the market.
The good news for Facebook is that a brave new paradigm is pushing the market system to the brink of a radical phase shift. The paradigm is structured around widespread socioeconomic and technological disruption, but it’s spearheaded by a rising tide of small, agile enterprises. These enterprises are using newfound access to advanced technology, data tools, and networked platforms to learn, collaborate, and rapidly innovate new product solutions.
Unencumbered by the rigid, siloed, and centralised decision making culture that weighs down many large corporate organisations, these small enterprises are flexible, inventive, and, crucially, able and incentivised to solve unmet needs as they emerge at the edges of market disruption.
By operating at the edges of disruption, where new seams of human need are exposed, small enterprises are naturally more likely to generate product solutions that are intrinsically indispensable, albeit to a smaller segment of the market. As the pace of disruption accelerates and new market edges appear, the opportunity for this community of tech-empowered small enterprise to make a more meaningful and impactful contribution to the market system is only set to grow.
It’s in this long tail of empowered small enterprise that Facebook has identified a rich and potentially limitless source of product solutions capable of satisfying the fluctuating and highly diverse needs of its two billion users. With over 60 million business pages on its platform, Facebook is already the natural home to this long tail of small enterprise. That said, currently only a small percentage of these 60 million businesses would likely be categorised as the agile and innovative enterprises driving the paradigm shift.
From this bottom-up perspective, Facebook’s investment in deep learning and neural networks isn’t just about directly improving the quality of its predictive personalisation capabilities. It’s also about using AI to create the conditions for the long-tail of small enterprises to do it for them. Instead of trying to bend the will of its users to buy the existing crop of commoditised products, Facebook’s audacious plan could well be to force the hand of the market to start generating a vast and more diverse range of indispensable product solutions.
It’s often overlooked, but the market’s original purpose was not to simply set prices and maintain a faithful supply of the existing crop of products. It was supposed to act as a mechanism to promote and facilitate the inflow of innovative new products, capable of addressing emerging human needs and solving societal problems.
Over time, the dominance of large corporate interests has badly distorted its purpose. These large corporations have engineered a closed and restrictive market system that has created the conditions for limited market adaptation and high levels of failure for innovative small enterprises.
Currently, even if an inventive small enterprise manages to establish itself in the market, it quickly finds itself in a frantic fight for survival. This isn’t because its products aren’t useful or needed. It’s because small enterprises often struggles to identify and manufacture adequate customer demand (revenue) quickly enough to sustain the business and stand up to the inevitable resistance from larger, incumbent competitors.
How Facebook’s Intelligent Market System Will Help Small Enterprises
The market system may currently work against small enterprises, but Facebook is in a prime position to help reverse the polarity. The emergence of Facebook’s ‘intelligent market system’ would help to tear down barriers for small enterprise. With its vast network of users and self-improving, AI-powered predictive engine, Facebook could help a small enterprise identity, reach, and secure the totality of its buyer market almost immediately, and at a relatively low cost.
In fact, Facebook’s recommendation algorithm is already positively geared to favour this new breed of small enterprise and its product solutions forged out of clearly identifiable unmet needs. In much the same way that Google’s PageRank algorithm favours websites that have a high user relevance, FBLearner Flow and now the DLRM prioritise product recommendations that are predisposed to the needs of an individual user and will likely lead to a higher probability of transaction.
Favouring small enterprises will reduce the failure rate for new entrants and further help to sustain the activities of Facebook’s 60 million business customers. Plus, a new wave of innovative small enterprises will be encouraged to innovate more boldly and push their new products into the market.
Increased numbers of small enterprises on Facebook create a promising new source of indispensable products. Plus, by exposing users to a far greater array of product recommendations, these enterprises also provide Facebook’s AI with a rich, diverse, and valuable source of data (most notably purchase propensity data). This will allow the AI to radically learn and self-improve its capacity to identify ‘guaranteed transaction events.’
Now, add into the mix the pipeline of transaction data that could arrive with the widespread adoption of Facebook’s cryptocurrency, Libra. With that, Facebook’s AI would be able to take its prediction capabilities to unimaginable heights.
Crucially, all of the future improvements made to Facebook’s predictive personalisation capabilities won’t just enable the flow of more valuable recommendations out to users. They will also allow Facebook to start sending indispensable recommendations out to the community of small enterprises. The more Facebook is able to illuminate and predict the collective needs of its users and their likelihood to buy certain products, the more it can use this intelligence to help improve small enterprises’ decision-making and outcomes.
At its simplest level, as the primary interface into the sprawling ‘intelligent market system,’ any small enterprise would be able to test the level of demand for new products or innovations to existing products, all before committing serious investment.
In the same way eBay recommends the optimum selling price for a product, Facebook would be able to predict the precise level of demand for a product, even based on simple prototypes or even high-level product descriptions.
Predicting Facebook’ s Future
In the future, there may come a time when Facebook chooses not to wait for the community to initiate the development and deployment of new products into the market. That might prove to be too slow and inefficient. Instead, it would monitor the evolving need-scapes of the market and send signals out to the community when it detects the emergence of pockets of new or unmet need. In response, swarms of small, agile, and nimble enterprises would coalesce around this intelligence to learn, test, and rapidly generate the innovative solutions required.
This two-way flow of data and information is the true realisation of Facebook’s ‘intelligent market system.’ It’s the key to generating the new source of indispensable products needed to transform its own fortunes.
Facebook’s goal is not necessarily to help some of its 60 million small enterprises unlock the kind of exponential growth venture capitalists get excited about. Instead, Facebook’s goal is to enable an exponential number of small enterprises to stay small, nimble, and local. This would ensure the ‘intelligent market system’ behaves less like a closed and inefficient machine and more like an open, highly adaptive, and interdependent organism.
Facebook would not only be the primary interface with the ‘intelligent market system’ — it would eventually become its fully automated central nervous system. Every second, Facebook’s AI would deploy billions of signals out to different parts of the organism (enterprise or consumer) to influence decisions and determine the level of energy and resources that should be allocated to different parts of the market system to ensure it remains sustainable.
More hopefully, as the ‘intelligent market system’ starts to behave more like a complex adaptive system, it would also be incentivised to think and act in the best interests of the whole system. For example, the elimination of huge market inefficiencies might be passed down to all market users in the form of more affordable products or lower cost of entry for enterprise. At the same time, levels of consumption and production that put the sustainability of the whole market system or environment in jeopardy would be discouraged and de-incentivised.
If Facebook’s mission to create an ‘intelligent market system’ is successful, then a number of things would happen to the fortunes of Facebook and its competitors. Firstly, Facebook would operate an entirely new kind of retail revenue model. It would no longer sell access to its users, or even simply take a clip of every sale. Instead, it would sell customers access to guaranteed transactions.
Facebook would be in the business of selling market certainty. This kind of relationship with the world wouldn’t just elevate Facebook’s profits past its rival Amazon. It would likely take its profits past the combined value of all competitors.
But, in many ways, profit would no longer be the only basis for Facebook’s power and influence. If it does indeed become the nervous system for an ‘intelligent market system,’ Facebook would transcend corporate entity or global government. It would become the underlying operating system for our entire civilisation. Facebook would become the ‘visible hand’ controlling the fate and future of humanity as we know it.
Much of this is article is of course pure speculation — a mere prediction of what could be. Like many predictions, it may well prove to be way off the mark. But, maybe we have all the data we need to make an accurate enough prediction? We know a lot about Zuckerberg’s psychology, his behaviour, and his ambition — namely, that it knows no bounds. We know a lot about the business strategies of his counterparts, and we understand the macro disruptions taking place in the market at a system level.
So while this future may sound implausible and far-fetched, given the unlikely turn of events this digital revolution has already served up, this prediction might just prove to be an indispensable guide to Zuckerberg’s grand master plan.