The guy who coined the term “conversational commerce” was wrong about conversational commerce
In January of 2016, I declared that it would be the year of “conversational commerce” — a new paradigm in which automated assistants and bots in voice and messaging channels would radically alter how people use computers! Data from that year that showed a migration to one-to-one messaging from one-to-many, passively consumed social feeds has just been reinforced in the 2018 Digital News Report from the Reuters Institute for the Study of Journalism. With this shift, a new front has opened in the battle to dominate distribution through conventional app stores thanks to conversational interfaces like voice assistants and bots.
To seize on this moment, most of the biggest tech companies opened up APIs to help businesses and brands connect with their customers over messaging and voice contexts, and a flurry of experimentation ensued — much of which quietly persists today. Armed with sequential chat-bubble form inputs delivered by bots addled with a dusting of NLP and backed up with a phalanx of humans-in-the-loop, brands, businesses and developers have explored a wide range of approaches to capitalize on these new channels.
Hindsight is 20/20, but who’s gunna wait until it’s 2020?
What I got right — and still believe — was that “over an increasing period of time, computer-driven bots will become more human-feeling, to the point where the user can’t detect the difference, and will interact with either human agent or computer bot in roughly the same interaction paradigm.” …And that this new behavior will lead to exciting commercial opportunities. See also: Google Duplex.
I underestimated the size of the dollar signs that would pop out of people’s eyeballs when they read “commerce” (did you ever see The Mask?), obscuring from view my very intentional “conversational” modifier. Looking back, I’m not surprised, given the industry average ~3% click-thru rate for marketing emails, and what with GDPRmageddon, what are these folks to do? Even though Marshall McLuhan died before email gained popular use, he did warn us that businesses and brands would rush to flood new and unpolluted channels with largely the same drivel which they’d already bought and paid for. Rest in peace, Marshall. Maybe we’ll learn someday.
Where I went wrong was in failing to define how to conduct ethically automated conversations that drive commercial outcomes effectively. While I provided evidence that a new mode of computing interaction was upon us and would result in new economic activity (not unlike the explosion of economic activity spurred by the US interstate highway system in the 1950s !) I failed to follow Jane Jacobs’s model and delve into the socio-economic behaviors and etiquette that would be prerequisites for success.
To cut myself some slack, it was (and still is) early, but now I’d like to address that omission with a new idea I’ve dubbed Relationship Design.
Before I go there, I’d like to introduce a new podcast called Machine Yearning!
Last fall I met up with Shane Mac from Assist to record the inaugural episode of Machine Yearning, a brand new podcast “where you can hang with smart people who are guiding a fundamental reordering of how we interact with technology”. We get into some pretty wild stuff, and discuss Relationship Design for the first time. It’s a good backgrounder for what’s to follow! Check it out:
Chris Messina: Full Interview - Machine Yearning
Chris Messina has been a pioneer in practically everything that has happened in technology over the last 15 years. He…
Adding your review would help more people discover this great new series!
Ok so: Relationship Design?
Yes, relationship design.
Relationship Design gives brands curiosity-tentacles and elephant-memory to support consensual empathy and intimacy with people
I know I know, bear with me.
Brands and companies on occasion have and deserve bad reputations. We scoff and bemoan their behavior because, historically, they’ve been both inscrutable to us and act as they were programmed to, i.e. to maximize shareholder value. But it’s necessary to put corporations as they’ve existed into proper context, because we find ourselves at a unique inflection point where everything is about to change, in a major way.
Throughout antiquity, there have been but a few step function improvements in cognitive ability stemming from increases in brain size. These changes resulted in enhanced perception, discernment, decision-making, and collaboration, among other skills. It has been speculated that Homo erectus’s mastery over fire enabled him to cook meat, thereby improving the efficiency of his caloric intake, leading to surplus energy which could be spent on more complex intellectual tasks. Research shows that the brain size of Homo Sapiens 300,000 years ago matched that of present-day humans, but that the brain shape evolved more gradually until 100,000 to 35,000 years ago.
So basically nothing’s changed for us in 35,000 to 100,000 years. I dunno about you, but AFAIK the written record of humanity is only about 5,000 years old—so there’s a lot of time in there that I know nothing about.
Meanwhile, the corporation came about in the 17th century. If you want to be extra smudgy with your math, “the alleged oldest commercial corporation in the world… obtained a charter from King Magnus Eriksson in 1347.” So at best, the corporation is about 671 years new. Next to the evolutional heritage of humans, octopi, elephants, or especially cute puppies and kitties, corporations’re mere protozoa!
So ask yourself: has the modern corporation really reached its peak evolutionary potential? Doesn’t it seem somewhat clumsy and maladroit compared to even the simplest of conscious creatures?
If so, then you too may see the opening for Relationship Design.
Ok, so what the fuck is Relationship Design?
Let’s do some wordplay, shall we?
A person is an individual human (typically with the ability of discernment).
People is many persons.
A brand is container for a set of messages that gain emotional currency within a collective consciousness.
Empathy is the experience of downloading someone else’s consciousness and inhabiting it, temporarily.
Intimacy is a measure of familiarity which, as it increases, may result in willingness to get close or to open up — i.e. to become vulnerable.
Conversations are interactions performed with language.
A single conversation is a loop in which meaning between two or more parties is negotiated.
Negotiating meaning is an attempt to transpose a thought from one mind to another.
Meaning is the subjective significance applied to experience.
Shared meaning is understanding.
Understanding drives intimacy by increasing the fidelity of empathic experiences.
A relationship is a ledger of interactions.
A relationship grows and deepens as diverse and meaningful interactions occur.
A relationship can be reified through external validation (i.e. being recognized and envied by other people) or realized through the manifestation of artifacts.
An artifact is an outcome of a creative conversation. Another word for artifact is “product”.
Therefore, products result from meaningful conversations.
A healthy relationship is one in which the interactions and products serve the needs of all parties, equitably.
A succeeding conversational brand engages in negotiated meaning loops (“conversations”) while expanding and fostering a network of healthy relationships between people and products.
Relationship Design, then, is about meaningful interactions, and how we create, observe, measure, edit, and improve the way conversational brands behave and relate to us in the era of the one-to-one, talking internet.
Relationship Design is how, over time, we course-correct corporations.
What does a relationship designer do?
One human consequence of the “Uber for X” trend is an increased sensitivity to intent awareness (i.e. “What do I want?”) which can be rapidly converted into task fulfillment (i.e. “Who can get it get for me as fast as possible with minimal cost?” (where cost is a combined measure of time, attention, money, and effort)). In other words, when someone senses a need — no matter how trivial — they will seek to fulfill or offload it, while expending the minimum amount of attention and energy necessary—lest they get interrupted and forget what they wanted (walk down the boulevard of broken intents of the browser tabs you’ve opened and never returned to for evidence of this). The impact of increasingly mercurial attention spans on brands is to force them to operate within increasingly narrow windows of opportunity and to become much more sensitive to the whims of consumers. Compounding the problem, digital gravity wells like fake news surreptitiously rob from the coffers of human attentional resources, whittling away at the remaining sliver even further.
To combat this challenge, relationship designers create a “service field” in which people can express their needs in various ways, which brands and businesses capture, and then turn around to opportunistically fulfill them through a series of discontinuous multimodal conversations. A key output from this mesh is to collect, collate, and triage “errors” — an incredibly special kind of signal unique to digital conversations. More on this in a subsequent post, but check out this post on Random Access Navigation to get acquainted.
Some of these conversations will be automated, and others will be handled by humans; some will be hybrid. The set, setting, and capabilities of the parties involved should inform the appropriate mix.
I will delve into these topics in subsequent posts — specifically on “Curious Tentacles and Elephant Memory”. Subscribe to my email newsletter and I’ll let you know when it’s live!
The purpose of this post, after a long hiatus, is to articulate at a high level, the broader shift I anticipate in consumer and brand behavior that precedes success in conversational commerce initiatives. It’s admittedly light on specifics because it’s still so early. It is motivated in part by the dearth of successful deployments of bots that pursued the mission of conversational commerce prematurely—without giving sufficient consideration to the norms and mediums and environmental context that gave rise to conversational channels in the first place. I left this analysis out of my first posts on the subject (though grazed it lightly).
As this is a relatively new area, and we’re still figuring things out, I welcome your input to make my perspective more concrete.
Early examples of Relationship Design in the wild
So, even if this post isn’t going to tell you how to do Relationship Design or become a Relationship Designer, I hope to plant some seeds which will grow.
As is typically the case, I’m not conjuring up these ideas out of whole cloth. There’s plenty of evidence and material to point to that represents what I’m talking about, but you have to be able to connect the dots and slice reality with a different blade.
Here are three examples to get us started:
Alexa and the Amazon Prime Universe
The clearest and most obvious designed relationship is from Amazon, with its enormous head start becoming the world’s most customer-centric company given that it was founded in 1994. In recent years, it has done more to advance the experience of living in a proactive “service field” than nearly any other company (though I may be biased because I live in it).
The Alexa strategy is the beachhead of their effort to make access to Amazon ubiquitous and friction-free. In addition to their own pedestrian-looking voice access terminals (Echo, et al), they’re moving quickly to secure third party hardware partners, and unlike Google, don’t require those partners to adopt a black box platform (at least as I understand it) to integrate their assistant tech. Anywhere you go, Alexa will be just an utterance away.
With their acquisition of Whole Foods, their experimentation with Amazon Go (a self-checkout convenience store), forays into physical book stores, and deployment of hosted delivery lockers, Amazon’s tentacles (wait for it) into the physical world are altering the kinds and frequencies of conversations Amazon can have with its customers.
I mean, check out the entryway to my Whole Foods… have I not just walked into an Amazon distribution center? (Where are the humans??)
In addition, with a huge push into video and original programming, access to an enormous library of free or discounted digital media, free or discounted delivery, credit card deals, wardrobe-as-a-service, etc — Amazon keeps increasing the value of a Prime membership because they know that a Prime member spends $600 more annually than a non-Prime shopper, and Echo shoppers spend another $400 more!
If that isn’t a leading example of the spoils of conversational commerce done right, I don’t know what is.
The Stitch Fix is in
At the 2018 Code Conference, Stitch Fix CEO Katrina Lake said many useful things about her unicorn personal shopping service which recently went public. Without saying so specifically, she pointed out how her company is organized to have ongoing conversations with Stitch Fix customers to capture their feedback, improve the product, and deliver an anticipatory experience:
At the core of what we’re trying to do is to take this element of personal shopping that used to be available to a very high-end customer, of somebody who knows you really well who can make selections on your behalf and have you try things on, and make that accessible… [It] is like … going into a restaurant, and the restaurant doesn’t give you a menu and just brings you food.
She also talked about the “service field” that Stitch Fix has created, to inspire conversations across a plurality of channels:
Adding more channels was really important. Having diversity of channels was really important. To your point, we don’t wanna be completely all-in on Facebook or all-in on Google, we wanted to have TV, and direct mail, and lots of channels that were effective. So today we are at a place where, year to date, we’re spending about 8 percent of our revenue on marketing. For many e-commerce companies that’s a pretty modest amount, and it is helping us to drive this 20 to 25 percent year-over-year growth that I described. So we feel that we have been doing a great job of bringing in a more diverse set of channels.
Finally, as an example of designing to correct “errors” or specific customer “failure” cases, she explained why Stitch Fix now offers a la carte items like socks or undergarments by pointing to a customer that’s spent over $10,000 with the service:
…we have the majority of her wallet share. And yet she’s still having to go elsewhere to buy things like socks and undergarments and bralettes and whatever else. So with extras, it was really about how can we be that full-service solution especially for those clients where we already have a large amount of wallet share.
From a technology and operations perspective, it is also super interesting. Because it’s not that selling socks is gonna be a huge market opportunity, but it certainly keeps people out of stores. It helps to make sure that we can be a full-service solution. But building that capability is interesting because now it allows us this testing bed to do all kinds of new stuff. Because to your point, we’d really only had our first core model. And now allowing customers to choose, being able to have a marketplace of sorts where we could test and try things that aren’t in the categories that we’re doing today. From an operational perspective now our warehouses know how to ship fixes with 10 items in them, and 6 items in them. There’s a lot of infrastructure stuff that we had to do in order to do that that actually opens up a lot of capabilities that allow for more flexibility in Stitch Fix in the future.
Through this focus on expanding the variety, length, and quality of conversations with Stitch Fix customers, you can begin to see how they’re designing a very specific, deep, and desirable relationship.
Spotify’s ever-improving Discover Weekly
Spotify is an engine of meaning creation and enabler of downstream conversations thanks to the shareability of items from its vast music, podcast, and video library.
It may be unintuitive to think of Spotify as a relationship designer, but many of its best features arise from the observable contexts it has created in which people start conversations around music, many of which are off-platform or embedded (see the recent Share to Instagram Stories feature or their advice to artists on social media). The service also gets better as more people use it, by providing light feedback (up- and downvoting tracks) on generated stations or by creating playlists, a topic Kathryn Hume explores in Episode 2 of Machine Yearning.
Furthermore, Spotify has launched Line In to allow trusted users to update metadata like Artist Links, giving customers a way to meaningfully contribute to the service they love.
And in a time when albums are getting shorter and becoming more like software projects that are updated after they’re released, artists too are looking for ways to design relationships and inspire conversations with their fans, as Kanye did with the cover of YE or Weezer did when it released a cover of Africa by Toto after a Twitter campaign run by 15-year-old Mary Klym. The fateful tweet that inspired the track ended up as the cover of the single—which in turn just landed on the Billboard 100 chart.
This, in my book, is exactly the kind of success from relationship-driven “conversational commerce” that I expect to see more of in the future.