What Agencies Need To Know About Bots And AI
Here’s what digital agencies need to keep in mind as they begin working with new bot platforms and strategies with their clients.
Call it the season of bots.
Over the past six months or so, bots have dominated much of the tech conversation. If you’re in a digital agency, chances are you’ve already had clients interested in developing a strategy around them. It could be you’re already working to sell bot opportunities into clients.
This is a big shift for marketers, who have traditionally had an antagonistic relationship with bots. After all, bots have been the bane of data-driven marketing, skewing social and advertising numbers just as analytics were becoming truly useful and insightful.
However, as companies like Facebook and Microsoft and Google have rushed to open up bot-development platforms, they’ve created a new channel for marketing. We are still in the early days. That said, bots will have a major role in marketing over the next couple of years, but we must also recognize they are really just the gateway to bigger AI strategies that companies will need to develop.
Here’s what digital agencies need to keep in mind as they begin working with new bot platforms and strategies with their clients.
Bots For Bots’ Sake?
Now that bots are dominating the conversation, and clever bot deployments have received more and moreattention, brands are looking to get in on the action. For agencies, their first move should be to help these companies see the bigger picture and understand exactly why they want to pursue developing bots.
Being able to force brands to address their underlying strategy for exactly why they want to deploy bots will prevent a lot of the missteps in development that come from companies simply pursuing bots for bots’ sake. There is an impulse to follow the latest craze to either keep up with competitors or be seen as on trend, but this can backfire without first understanding the bigger picture of why a company wants to deploy bots. The capabilities of bots do not change the underlying principles of marketing, advertising, and branding. Bots are just another medium for a company to present their brand story.
As trusted consultants and partners, agencies have to push for this vision from the brands they work with. What are the business goals behind the bots? They can return real ROI for companies, but it must be clear exactly how bots will operate across channels, and companies must have a solid understanding of their customers and the expectations. (More on this below.)
Agencies must also help brands decide how to future-proof their bots, especially as both expectations and technology rapidly evolve. What channels should their bot strategy focus on, and how can they operate asynchronously across multichannel environments?
All of these questions get to a bigger concern that agencies need to force their clients to articulate: What is their overall AI strategy for the company? Bots are one endpoint of that strategy, but you don’t start at the end and work back to strategy.
Instead, companies need to understand how AI and bots will work across departments, from marketing and customer service to sales and even internal communications. All of these pieces of the organization need to be aligned strategically before the technology can be implemented usefully. Developing this strategy requires a much deeper conversation, but it’s one that digital agencies should force brands to have before just chasing after the latest trend by developing bots.
Communication And Context
Right now we’re seeing a lot of task-oriented bot deployments; bots excel at being able to perform simple and straightforward actions on command. Order a pizza? Done. Pay a bill? Done. Summon an Uber? No problem.
So long as the action (and communication) is fairly straightforward, bots can provide a much more convenient channel to fulfilling these tasks. In the past, advertising has been a one-sided, push conversation, but now consumers have the ability to engage two-way conversations with a brand. Social has likewise opened up that expectation for consumers, but struggles to scale. Bots offer a new opportunity for two-sided engagement, but as most bot developers know, it gets much trickier when customers start asking questions and expecting a more conversational interaction.
Handling more complex inquiries requires a great deal of context and domain-level expertise. A bot for an airline must operate fundamentally differently from a banking bot. There is an entirely different language (in the broadest sense) that must be understood to carry out the appropriate tasks. Developing these contextual and domain-expert models can be much more complex than many developers initially realize.
Brands need to be keenly aware of the required complexity of developing an effective bot-communication channel. If execution is poor, companies are at real risk of losing customers. With bots, companies are effectively giving over one of their most important points of contact with customers to automation, which can prove incredibly valuable if done correctly—and disastrous if done poorly.
Natural language processing models have improved greatly in recent years, but developing your own brand and business-specific languages are necessary for bots to be truly effective. Brands must consider many elements of what their bots need to know, but here are three main questions that will get them started:
1. What are their customers’ expectations for the interaction? While bot development has exploded this year, many developers have not advanced the experience beyond consumers’ existing points of reference for brand engagement, such as search, FAQs, IVRs, social, etc. Agencies have an opportunity to think and design in the most natural paradigm, a two-way dialogue, which requires a thorough understanding of how consumers ask for information and interact with brands. Designing conversational interactions based on that understanding will intuitively educate and evolve user expectations.
2. How will the bot act? This question is, at its core, a branding question and one that digital agencies are especially effective at addressing, but it’s also a technological question. All bots ultimately have personas even if they are straightforward and obvious. Developing that persona goes far beyond simply a voice or vibe exercise: It must account for how a bot represents a brand, how it resolves issues and interactions, and especially, how it fails.
3. What happens when the bot fails? This question is key because all bots will eventually reach the limit of what they can do for a customer; in fact, customers usually immediately attempt to “break” bots when they begin interacting with them. Developing a strategy for how the bot responds to failure is not only a question of the brand’s personality but also a bigger issue with regard to the customer journey. There needs to be an elegant fall-back and handoff that can actually help the customer, whether that means handing off the problem to a human agent or directing the customer to the proper resources.
Training And Human Oversight
At a recent panel discussion around agencies’ use of bots, Tom Goodwin, VP and head of innovation at Zenith Media, made a point that all brands and agencies should heed: “What concerns me most about AI is that we hope that people are going to come to our websites, we hope that people are going to show interest in our products and the idea that we have this rare moment where someone shows this massive interest in a product and then we outsource that interaction to a bot is crazy.”
Goodwin expressed one of the biggest concerns that agencies need to address for their customers, and it should be at the forefront of any initial conversation with a brand about its bot strategy.
Brands (and their agencies) can’t consider bots simply to be plug-and-play technologies that they can deploy and let run. The best-developed bots will be able to naturally learn, adapt, and even personalize interactions, but that won’t happen in a vacuum. Company products and services shift, customer questions change, and even the technology itself evolves. At every step of the way, bots offer a scalable and valuable solution, but they are built and maintained with a combination of human and machine expertise.
The comparison is often made between bots and AI and various stages of human education and learning, but the implied metaphor is misleading. There are things that bots will be able to do much better than humans, and there are human capabilities that bots will not be able to reproduce.
Agencies need to start their conversations with brands by ensuring that their clients understand the investment they are making. To be effective and competitive, brands need to realize that bots are just the first step to a much bigger shift toward AI that is already underway, and agencies need to help them develop that clearer vision of AI strategy in the long term.