In our last instalment of the Bots series, we will explore how brands can get prepared for this new form of marketing and communication channel with consumers.
Beyond the exponential number of smart phones and messaging apps in circulation in the world, one of the most surprising figures I’ve encountered recently is that Amazon Alexa went from 0 to more than 10,000 voice-controlled skills learned in a year and a half. Whilst this doesn’t come close to the millions of apps available to consumers in any app store, this is sending a clear signal to developers and brands: consumers are ready for the next stage of marketing interaction, are we?
Consumers already have the choice: 5 major virtual assistants currently battle it out (Siri, Alexa, Cortana, Google Assistant and to a degree Facebook M), and an estimated 100 to 200,000 chatbots are already in circulation.
But with all the puzzling buzzwords flying around (machine learning, deep learning, neural networks, natural language processing, etc) and, let’s face it, the hype around bots right now, what does this all mean for agencies and brands and should we start tackling the topic?
What does the change mean?
There are, right now, 3 main ways for brands to start engaging with bots. Depending on needs, objectives and appetite to risk, different approaches will be required:
1/ Start experimenting with virtual assistants:
The brand wants to start capitalising on the emerging visual assistant environment and needs to start with:
- A brand skill in Alexa (Amazon)
- A brand action in Google Home
- A brand skill in Cortana (Microsoft opened to developers only a few weeks ago its Cortana skills kit)
The key benefit here is that the brand can take advantage of an already existing large user base. It can be integrated, with prudence, to existing paid media or existing marketing efforts. These are usually built by developers trained on software development kits for each virtual assistant. Though cloning a brand skill from one environment to another should in the long term be simple, the early days may be difficult.
2/ Start experimenting with chatbots that function based on rules
This bot will be very limited. It will only respond (text or voice) to very specific commands. If you say the wrong thing (or slightly differently to how it was taught to it), it just won’t know what you mean, ultimately it will only be as smart as it is programmed to be. Therefore, it is important to focus on one (or very few) task(s) with a controlled set of rules for each task, and teach the program to handle them well.
The benefits here are, unfortunately, only the fact that it is the path of least resistance. It may be the easiest way to build a first chatbot but it may also be the riskier bet by offering a slightly frustrating experience to the consumers by serving the infamous and ever annoying:
In short: this option offers low cost and low complexity but also low potential return on investment and potentially high risk from a consumer experience.
3/ Start experimenting with chatbots that function using machine learning
This bot has a degree of artificial intelligence, in other words, users don’t have to be ridiculously specific when they are talking to it. It understands language, not just commands. It can also get better at understanding context of commands.
This bot continuously gets smarter as it learns from conversations it has with people.
Unlike the former type of chatbot, this one is much harder to develop and requires much more upfront work in understanding what type of requests can be formulated and how to answer them. This will have an incidence on coding time and skills required both from the developer and the marketing team involved.
What are the implications for advertisers?
Bots are created with a purpose.
Chatbots will be successful, at least at first, in use cases where interactions are simple, fast and easily automated. Brands should therefore focus on what makes the most sense for both their business and the convenience of their consumers:
- Providing local services and information
- Delivering simple and effective e-commerce functions: recommendations based on historical purchases, repeat purchases
- Customer services: from order and shipping tracking, to quick diagnostics for faulty products
- Conversational on certain relevant topics: news, entertainment, etc…
- Internet of things: connecting devices such as thermostats, living room assistants (such as Alexa), pulling in your connected car in the driveway or in/ from the parking lot,
- Voice functions when it is inconvenient for users to use their hands (any case of multitasking)
- Personalised content: for content providers or brands
The illusion of simplicity
I find this quote from Shane Mac, CEO of Assist, a company specialised in building bots for brands quite interesting:
“Beware though, bots have the illusion of simplicity on the front end but there are many hurdles to overcome to create a great experience. […] Analytics, flow optimization, keeping up with ever changing platforms that have no standard. For deeper integrations and real commerce like Assist powers, you have error checking, integrations to APIs, routing and escalation to live human support, understanding NLP, no back buttons, no home button, etc etc. We have to unlearn everything we learned the past 20 years to create an amazing experience in this new browser.”
But does that mean we must become artificial intelligence (AI) expert? Not necessarily. So many advancements have occurred in the past few years alone in this space that it has become possible to incorporate some level of AI in products. Brands just need to make sure they do not over promise on the bots’ applications and capabilities.
From this point on, one could even envision a future soon where bots could by-pass the need for apps… who knows?