One of the biggest pains conversation designers, and conversational AI professionals in general, go through in their company life is the need to continuously prove their worth. The way AI works isn’t always clear to everyone and the way the technology has been depicted in the media and in popular culture has been known to inspire skepticism and mistrust.
The distance that’s created between the conversational AI team and the rest of their department, not to mention other units, may create negative consequences. That’s on the side of the CAI, as designers and project managers will continuously need to evangelize and go out of their way to prove their worth, but also for the rest of the organization.
For example, it is known that a lot of valuable customer feedback is collected through a chatbot. However, without the proper structures and processes in place, this precious data might never find its way to the person who can actually act upon it.
In this interview, Esha Metiary, Senior Conversation Designer, shares her learnings and expertise on evangelization, conversation design legitimacy, cross-departmental bond building, and effective knowledge transfer.
A Chat with Esha Metiary about Conversation Design evangelization
Why is information sharing so important within an organization that incorporates conversational AI?
As a conversation designer, you need to make sure the content is consistent with all other outputs, e.g., with the website or social media. For this, you’ll need to interact with the people responsible for those pieces of content and align with them.
On the other hand, it’s easy to forget that the bot itself is actually a two-way street for information sharing. We tend to treat it like a channel to put out information and content, rather than a tool to gather input from customers. In my experience, it’s been very important to highlight the value of this opportunity for feedback and to allow it to be part of the design process.
People tend to tell chatbots their whole life story and to go into a lot of detail. Being able to collect that information allows the CAI team to build a better product and the whole organization to understand how customers see things from the outside.
How should the valuable information and feedback you receive from users be dealt with?
The customers are the first to notice when something goes wrong and they will immediately come to customer support, or to the bot, to report that. If there’s a problem with orders, or there is a payment outage, the conversation design team will know, often, even before the involved department has become aware of it. In these situations, we’ll reach out to the department and inform them of what we’re seeing in the data and transcripts.
Does the reverse also happen? That someone from another team reaches out to bring something to your attention?
Many companies today still tend to see the chatbot as just another way to put out their content. For example, it happens pretty often that someone from marketing asks the conversation designer to add content to the chat, as they expect users to enquire about a certain topic.
In this case, the conversation design team needs to check the data and see if it’s true that customers will actually ask about the topic. The interaction with the other department will be aimed at finding common ground between the content they’d like to produce and the questions that the users are genuinely enquiring about.
However, I do see this approach becoming less and less common, as companies adopt more customer-centric approaches towards content.
How are the more customer-centric organizations changing their perspective on content?
I’m seeing a big shift in the ways companies approach content, from process-oriented to customer-centric. So far, it’s been pretty common for companies to put out content focused on the internal processes and framed from the perspective of what the company offers and excels at.
Right now, everyone’s coming to the realization that what customers actually need is an answer to their question. To provide these tailored responses, different departments that might be used to working in isolation from each other are needing to put their heads together and combine all their viewpoints into one, collective view.
How does this translate into the way different departments interact?
Every person that works in content should at least have a seat at the table when it comes to broader communication strategy. This might not always happen at the beginning for the conversation design team, when the role has just been created and the focus for the team is on the output, i.e. creating flows, training the NLP, and so on.
As the designers gain experience, they should be more involved with the people who work directly with the customer, as well as with those responsible for brand identity and marketing. They should regularly set up that table themselves and get everyone to talk. Ultimately, if they’re all working towards the same goals, everyone will be willing to listen and share perspectives with each other.
In what department, does the conversation design team fit best?
Right now, I see conversation design teams either working with the content department, or in customer support. Either of those work, as long as the two are very close.
If I had to choose, I'd probably choose customer support, because they are very familiar with the company's processes, they actually speak to customers all the time, and they know how to operate in all situations to help the customers. They can tailor that solution to the exact needs of the customer and that is what makes them most valuable for conversation design.
What are generally teams that conversation designers should be close with?
This will depend on the domain and use cases conversational AI supports. Generally speaking, it’s good to speak regularly with marketing, to be updated about ongoing campaigns or promotions, and whoever is responsible for the content and deployment of other channels, like the website.
How do you understand what other teams you should be connecting with?
As a conversation design team, it should be clear to you what your organization’s core business is and where it happens. So, if your company is a retailer, your core business will happen where you sell things, but also in your warehouse and delivery facilities. In this case, you’ll need to build relationships with the retail team, with the warehouse, and possibly with the logistics team, as it’ll be important to be updated when, e.g., there are strikes or unusual peaks in users to be expected.
How do you ensure that everyone is aligned and you all have the information you need to operate at your best?
If you’re at a company where the departments are close enough, they’ll keep each other informed of what they're working on, even about developments that others might not be directly interested in. Sometimes there might not be anything immediately valuable to this exchange, but other times you might find an unexpected lesson from the way your colleague has approached a certain situation.
To enable this exchange, you shouldn’t just rely on the odd coffee machine moment that you might or might not end up having. I recommend setting up a weekly, or bi-weekly, meeting where all the content teams come together and talk about things they’re working on.
What’s the most effective way to make as many people in your company aware of your work?
What I first like to do when we're starting out with a new partner is to organize a persona workshop. You grab people from layers all over the company, get them together and put down how the bot’s persona should be. You really want to get people from different areas, e.g., marketing, digital, IT, even the CEO, because they all have different aspects of knowledge about the company and together you’ll be able to create a brilliant persona.
More importantly to your point, this is the moment that the bot starts living in the back of their heads and they will probably tell their colleagues about it. Then, you’ll work on this as a conversation design team and, once it’s ready, you’ll want to do a presentation to show them. It will be like saying: Look guys, this is what I made with your input. This is basically your baby. You can also tell them to invite others and make this presentation public, so you can update everyone on what your next steps are and let them know that you’ll be demoing the first use cases you describe, as soon as they’re built.
Does talking about the things you’re working on also help you further legitimize the role of Conversation Design within the organization?
It’s true that, internally, you have to prove your worth as a conversation designer, but you wouldn’t just jump in and start talking about what you’re working on without any introduction to what your role is about.
First, you’d start by telling others what conversation design is about, paying particular attention to:
- explaining that having a chatbot that people feel like they can connect with is crucial for adoption
- showing how conversation design is, at a very basic level, about understanding the natural instinct all humans (with the exception of some neurodivergent people) have to connect with anyone they speak with.
Then, you can start talking about what conversation design can do, what we are doing and why we are doing it.
Once the conversation design department is maturing, it will come down mostly to a specific set of KPIs. Mainly, you’ll want to be able to prove NPS is increasing and calls to customer support are decreasing, thus lowering costs, helping your customers faster and more effectively.
So, would you say, at some point, data becomes a stronger argument in underlining the importance of conversation design?
Depends on the kind of company you're in. Data is always an argument, but it doesn't really have to be an argument anymore if your organization has been able to go full on customer centric.
If your organization is new to conversational AI, data is your only point of argument there. You want to reduce your customer calls and get people to find their information quicker: these are the basic arguments.
What types of data or information should you be looking at and bringing up with stakeholders?
Checking the actual data from the logs and transcripts to show what customers are saying is very important. You should be doing this systematically, every couple of weeks or more often, depending on the size of your organization and the volumes of interactions.
Customer branding, employee branding, and brand identity should also be curated, to make sure your bot is an extension of your brand message. Especially if you have a brand that is very vocal and feels like they want to spread their message, the argument for having a chatbot that allows you to do exactly that is very strong. Ben and Jerry’s, for me, would be a great example of this type of company.
And if you’re still hungry for knowledge, follow us on LinkedIn for weekly updates on the world of conversational AI, or check out our interview with two AI experts on how to evangelize and educate others about (C)AI.