What service and revenue models for our virtual assistants will offer the right customer experience?
Author: Gavin Maguire
“The next big step will be for the very concept of the ‘device’ to fade away… The computer itself will be an intelligent assistant helping you through your day." - Google CEO, Sundar Pichai
Can’t live without ’em
Soon, most of our interaction with the digital world will be done in the easiest most natural way – by voice. Virtual assistants, such as Amazon’s Alexa and Google Assistant are in a race to provide the most valuable service to us. To do this - and to secure our dependence on them - virtual assistants will become highly capable, near omniscient, and masters of the rapidly-growing internet of things (IoT).
The more we open up to these ‘voice first’ machines the more they will help us. The most useful assistants will combine a deep understanding of the circumstances, attitudes and needs that govern our individual lives with the machines’ access to the world’s information, products and services. This powerful combination has the capacity to make virtual assistants indispensable as they help us to plan and facilitate what we want and need to do – both in our private and professional lives.
The revenue model is part of the experience
As a service designer, I’m looking beyond the current techno-fumbling, loss-leading customer land-grab, to speculate on how things will work when the service is more mature. As we’ve seen with news and music streaming services, the free ride will end.
How will we pay the robots? That depends on the service they provide. I don't wish to make an overt comparison of the experiences offered by current players, but to look – from a customer experience perspective – at the revenue models they might employ to appeal to customers in the future.
Selling things vs. doing things
Their differing heritages (Amazon’s retail and logistics, Google’s data and services) seem at first glance to align them naturally to different future service models. It feels as if Amazon’s Alexa is well-placed to continue connecting us to third party products and services. Third parties are signing up in droves to become Alexa ‘skills’, and Amazon’s store offering speaks for itself. On the other hand, Google’s native ‘intelligence’ and information sources position it well to be able to pull together solutions which respond to our requests. Research from 360i found that Google Home is over six times as likely to successfully answer questions than Amazon Alexa, while Google’s access to the world’s data remains unmatched.
So, which has more promise – the selling things model, or the doing things model?
Of course, this is an oversimplification, and actually just a matter of emphasis; all players will need to offer both selling and doing and are continuing to invest in building these offers. Many of Amazon’s third-party ‘skills’ are services which will solve problems or do things for you. And Google is chasing Amazon’s retail offering by signing with partners including the likes of Walmart.
The revenue model depends on the emphasis of the service offered. For example, in the selling things model, revenue can easily be collected from participating third-party retailers, whereas in the doing things model, it’s somewhat more likely that the customer will have to pay directly for service - one way or another. So, where’s the money going to come from?
Audio ads are inappropriate
Even now, the online ad-supported model is troubled. Most ads are by definition intrusive and distracting. People hate them so much that ad-blocking software is threatening revenue.
Sometimes we tolerate ads, for example on a page where we can try to ignore them by focussing on what something else or on the radio when we’re actually looking for distraction, companionship, or even white noise; which can keep ad time from being ‘wasted time’.
But ads seem particularly inappropriate for virtual assistants. We interact with the assistants when we’re on a mission. Ads would hinder smooth, efficient communication; you might not appreciate a travel agent who kept interrupting your attempt to arrange a flight with recommendations for financial products. Even worse would be one who recommended an airline or destination based on its ad spend rather than your needs. Indeed, Google Home’s experiment with advertising in March proved unpopular and short-lived; people who wanted a news update turned out not to want to have to listen to paid content about a Disney movie.
We’ll be grateful to virtual assistants that proactively interrupt our lives to bring genuinely useful things to our attention, but interruptions by ads would be as popular as today’s "nuisance" cold callers.
Data-mining may not be enough
With their access to everything we do, no finer more effective data mining equipment will be available than virtual assistants. But will the revenue from this data be valuable enough to sustain providers? And more to the point, will it be enough to satisfy them? It seems likely that other methods of monetisation will also be explored.
For the assistants to work at all, we’re going to have to share our data with them. The extent to which data sales can subsidise virtual assistants depends on how willing we are for them to then share our data with others. There’s increased worry about how the unchaperoned adventures of our free-range data will lead to secondary ‘experiences’ such as ads which follow us around – and potentially more sinister outcomes. The jury is still out over whether this heightened concern will lead to significant changes in the kind of data we’re prepared to release.
‘Member retailers’ seem inevitable
As far as shopping goes, the most useful and desired assistant may be the one which gives us easy access to the right shop from the largest pool of retailers. Retailer ‘membership’ and commission models seem to work well. It’s easy to imagine a world where this grows in scale. A world where, as with credit cards or PayPal, many (if not most) retailers pay to participate.
From the customer’s perspective, assistant biases around member and non-member retailers wouldn’t be ideal. A model with (paying) member retailers would need to apply a couple of principles in order to work well for customers (and thereby for virtual assistant providers):
- Inclusive – all businesses and services, regardless of whether they were members or not, would need to be included in the assistant’s consideration and (if appropriate) communication.
- Transparent – the assistant would need to be entirely open about which of the businesses it put forward were paying members and which were not and how its offers in relation to them differed as a result.
For non-chargeable ‘services’ – such as controlling connected home devices through virtual assistants, third-party manufacturers may be asked to pay for the APIs needed to connect their products to our assistants. While these APIs are largely free now, charging for them would allow assistants to pass the right costs to the correct IoT consumers. This would offer a better customer experience than some revenue models, but a provider will only let it happen once they’re the clear marketplace leader – if at all – due to the absolute necessity of signing up as many product-makers as possible.
Tiered service levels are likely
A tiered service level offer (or freemium model) would be a natural fit here. This would let customers choose what they want to pay for, based on how important the service is to their lives. The ‘free’ entry level might offer customers access to full services for member retailers (e.g. full purchase transactions and facilitation), but limited service around non-member retailers (e.g. just providing information). For a subscription or per-transaction fee to the virtual assistant provider, customers could get the full level of service for any business. This model would open the majority of retail to customers, in a commercially-sustainable way for providers.
Subscriptions may apply
Doing things (providing intelligent answers to our questions, making suggestions about the best way to do things, and actioning what we want to do) will be an equally important role as retail.
Doing things may be harder to fund since it will be less directly related to purchases. This function could be subsidised by an assistant’s selling activity. If not, another option is a customer subscription model. After all, people are willing to pay for services they value.
The extent to which retail can subsidise other functions depends on an assistant’s mix of selling things vs. doing things. If there’s a lot of selling with not so much doing, chances are high that a subscription model wouldn’t need to be considered.
Omnipresence amplifies importance of CX
Logic says that the customer experiences with the biggest impact on our lives are the products and services we use and rely on most. Virtual assistants will have a fundamental and ubiquitous role. So, it’s clear how important it will be that they are useful, usable, and desirable.
As business models can have huge implications on customer experience, providers must strive for customer-friendly means of paying the bills. The robot who deserves to win – and probably will – will be the one who can pay itself without compromising or interrupting service delivery, impartiality or integrity.