I was recently asked how do I think about a website. This was a simple question and after years of optimizing professional and personal websites, I thought about the website in terms of a discovery, and transaction platform optimized to reduce friction at each step of finding or transacting, pre or post most desired action (say a conversion of any type for simplicity). I was dissatisfied with my answer and kept thinking about it. I had only looked at websites from perspective of discovery pages (each page and action for different stage of discovery from awareness, intent to conversion and nurture) or in terms of top, middle and bottom, with the same of type of pages bucketed in a framework of a marketing funnel. The purchase funnel is a good framework, to quantify ecommerce in terms of funnel actions by page and it works very well, specially in advertising, audience targeting and optimizing pages to intentional goals.
With all the website events now available and all the elements on the site tune able at lightening speeds on client side with single page applications or server side rendering, I would think of a website as a machine, that takes inputs and generates outputs like a page. The inputs of the machine will be the context and actions with in a time scope that bundles those actions. The context can be defined by different signals which can be user specific, device specific, channel used, to any other parameters that define the intent or a goal, packaged within the URL and privacy protected client side data. The actions can be as micro as a single position of a mouse scroll. The machine can then decide, on the fly, the best way for it to run and create output. Imagine that depending on the page you come to, the machine is trained to a specific goal for that page, and now think that it constantly learns from the context and actions. To further build on the thought, imagine that all the creative elements from copy to images are created through machine learning and trained to the ultimate goal that the page is trained for, whether the goal is best consumption of the page measured by time on page, or a desired action, and even the goal for that matter can be eventually discovered by same process of testing and learning, automatically. This determination of the goal itself is the ultimate aim. Because at the end of the day, the purpose of the machines' outputs is to provide the users with the inputs that make the achievement of users' goal simpler, or clarifies their purpose or need, making it easier for them to achieve fulfilment. Depending upon the scale and scope of incoming signals the machine can run multiple outputs at scale and learn, optimizing to a target. The similarity traits of the user need only be specified through privacy protected parameters in the URL and combined with first party data, that will give the machine material to self optimize its output to the best probability of a desired next action, whether off page or on page, in future or immediately. You might think that I am talking about personalization at hyper scale, experience management platforms or content optimization and you are right. Maybe I am dabbling with semantics here but to me concept of a fully configurable machine sounds like something that will have all this type of engine built within the page itself and the output of the machine optimized for best possible outcomes, constantly learning, more like a native functionality instead of borrowed engine. The success of such a machine is heavily dependent on the quantity of inputs, diversity and quality of inputs. Maybe this is a conceptual difference and not a paradigm shift.
Regardless of the machines' internal working, the future is here where right framework defined for end result in mind, in time, with enough inputs, will allow the machine to learn the best way to serve its output, the page, to nudge the visitor in the right direction.
