User journeys on the web turn longer and longer, consideration cycles take loop ways, buyers are constantly looking for bargains, buying alternatives are evaluated constantly, information on alternative products are collected at the point of sale.
Things have turned rather complicated on the web these days.
Users’ information retrieval patterns turn more and more situational. No longer are static entities like the information architecture of a website, the product presentation in a catalog, or the cheapest price amongst a range of vendors the decisive criterion.
Instant availability vs. delivery time frame, seller’s reputation and post-sales services, or, in short: buying convenience, are playing a more and more important role. Communicating core and augmented product utility, abating buyer’s remorse, or, in other words: the promise of reducing the buyer’s perception of risk, are becoming increasingly important criteria.
If we think about it, none of these critical issues are addressed properly by looking at a certain entity on the website. Risk perception is particularly subjective. Satisfying the constantly changing needs that are being brought forth by a user discovering and mastering the uncertainties and indeterminableness along the decision path requires a perspective on the cascade of decisions reducing uncertainty, not on decision points when a prospect turns into a buyer (this happens in the very moment the “purchase” button is pressed. Anything before that is just a prelude).
Looking at click-stream data collected from your website doesn’t illuminate any of these topics sufficiently.
They never did. But for a long while we felt content in believing that aggregated volume data would suffice to confirm a website experience was working.
Understanding and properly attributing internal and external effects in making decisions, and finally taking for granted what Herbert A. Simon coined as “bounded rationality” (in 1957!) is the actual name of the game.
Bounded rationality, in short, describes the constraints under which decisions are being made: limited availability of information, limited capacity to process this information, limited amount of time available for processing this information.
Upon second thought, this puts every single user journey under slightly different starting conditions.
And yet: marketeers are still spending millions luring people in to websites where they offer products with features they hope would speak for themselves, and be attractive for randomly defined target groups according to largely irrelevant segmentation patterns like : IP address = location = language. Using “operating system” or “browser language” for determining the language to address the user in would be a lot wiser (and is, technically spoken, not more difficult) – still I am greeted on websites in languages I don’t understand, as I just happen to use an address within an IP cluster that used to be assigned to a range of devices in Poland until a couple of weeks ago.
The organizations these marketeers work for are buying into BI systems currently, they are integrating their websites with their CRM tools, and are searching for specialists who can make sense of aggregated, yet arbitrary database contents. Organizations are seeking for business data processing specialists fluent in MySQL and Visual basic, to dig out gold nuggets from their data, consequently overlooking that they still are operating garbage in-garbage out systems.
The websites of these organizations are often largely mimicking that of their first-dot-com-crash pendants: built for the purpose of one-stop visits with immediate purchase from a local delivery organization (“Contact us”), featuring imperative calls-to-action (“Buy now!”) and deep navigational structures (“Please select a product group first”. “Please select your country.” “Please select the product you want to make a sales inquiry about.”).
According to a recent study, you are more likely to die in a plane crash than to click on a banner. Still: the criteria for evaluating the on-target acquisition of traffic are mainly volume-driven and reach-based. A click-through rate 200 percent higher than average still makes only 0.3 percent. To get one click, you still need to show your banner to 3000 people. And, of course: one click doesn’t make one purchase. Not even ten would do.
In other words: we are too simplistic with our measurement approaches. And it certainly doesn’t matter whether we optimize our traffic acquisition patterns based on demographics, on household income, or on street address. It’s situational patterns we are looking for. It’s about multichannel, POS presence, clever smart phone spotting, showrooming, split-second decisions, and a great deal of contextual data we don’t even have an idea about yet.
THIS is the precise promise of big data: Utilizing insight sources yet unaccounted for. That may mean a lot of different things: deriving knowledge patterns from data about unemployment and debt rate, and transforming it into a buying propensity index for consumer durables. Identifying good locations for second-hand shops with kids’ clothing from gentrification data within a quarter, or correlating rising search volumes for “influenza” in certain regions with data about the regional spreading of a flu epidemic.
Contrast that with the prevailing assumptions still widespread in digital marketing: people will continue to buy products from us because they love our brand, a click on a promotion banner is intentional, a customer is interested in buying directly from us because we offer great service with our products, a “buy now” button is increasing the user’s propensity to buy, and having a presence on Facebook will increase our brand preference.
I am not saying that none of this is true. I am simply saying: we don’t know properly. So let’s find out.