Behind the Desk of a Conversation Designer: A No-Nonsense Peek at the Job Role

Conversation Design is a complex domain to explain during extended family gatherings and hurried elevator pitches—not for a lack of words, though.

• I design virtual assistants.

• I teach them what to say.

• I draft their stories and their paths.

• I plan what they can do and say in the short period that a brand’s consumers interact with them.

All great, but there’s much more to the role.

Here’s the official definition we’ll break down in this article —
A Conversation Designer is responsible for designing the user experience of a virtual assistant. They ensure that the virtual assistant is conversationally engaging, impactful for the end-user, and matches the voice of the brand. They translate the brand's business requirements into natural dialogue flows backed up by UX research and good design practices.


Whew. Now that the definition is out of the way let’s begin. I’ll share the stories of 4 roles that I’ve seen and undertaken during my career as a Conversation Designer.

🖋 A Conversation Designer is a Copywriter

Like most Conversational AI companies, Haptik, too, started with a nebulous definition of “Conversation Designer”.

As the first “Conversational Writer” on Haptik’s payroll, my initial days designing virtual assistants(or chatbots as we called them) consisted of proofreading and correcting the user journeys we developed on a spreadsheet.

I would also sometimes take the liberty of suggesting some changes(gasp) to this journey(as defined by the client or the Program Manager) to meet the end-user’s needs. However, the execution was left entirely to the engineer and the PM.

The result? Chatbots with good content in chunks but an inconsistent end-user experience overall. This was because the Conversation Designer was only contributing to bits of isolated copy, not the entire user journey.

Why being a copywriter is essential: You need your bot to seem well-spoken and speak in grammatically correct sentences. You also need your bot to have this elusive “personality” that everybody speaks of. However, personality does not shine through individual chat messages.

This brings us to role number two.

🤩 A Conversation Designer is a UX Designer

• Should I use a form here? Or a carousel?

• Is ticket booking in chat a good idea?

• How much text is too much text in a message?

These are all questions I’m asked(or ask myself) daily.

Did you know that users often tend to scroll back in a conversation and select another option instead of continuing linearly?

Right from the time a user initiates a conversation to when they get an answer to when they bid the IVA adieu, many UX decisions are to be made.

Results:

Like any other product, Good conversational UX is the sum of its parts.

You ideally want your experience to be better or more effective than its current existing counterparts. If it's isn’t, why bother building it?

Without user-focused thinking, you create an experience that meets the BRD or business expectations but doesn't truly solve a problem for the end-user.

Like humans, the first impression is usually the best impression when it comes to virtual assistants.

Plus, the success rate of selling to an existing customer is 60-70%, while the success rate of selling to a new customer is only 5-20%! (Market Metrics)

Why being a user experience designer is essential:

Thinking like a UX designer helps you create a cohesive, user-friendly experience. 

You also ensure that your virtual assistant does not violate the fundamental UX laws, like Grice's Maxims with the Co-operative Principle or Jakob Neilsen’s Heuristic principles when applied to chat.

You think like a user - their thought process, problems, and the best way to give them a solution. 

Example: A virtual assistant that tells you how to purchase a SIM card

If the UX focus is speed, then you'd opt to answer in a single quick message.

If the UX focus is clarity and engagement, then you'd break the answer down into several steps, maybe identify the user's needs and then recommend a SIM card to them.

What changed? The user expectation.

🤖 A Conversation Designer is an AI Trainer/Domain Expert

Story:
So if a Conversation Designer is responsible for designing the virtual assistant, who will teach it?
This is an age-old question at any Conversational AI company.
And if you’ve read any Machine Learning textbook, you’ve probably also heard that ML algorithms need LARGE amounts of data.

As the person who understands the bot’s capabilities the most, it falls to the Conversation Designer to provide a framework for the bot training. They need to identify which user utterances lead to which journeys and help the developer understand this correlation.

Results:
Having a Conversation Designer onboard during training ensures that your virtual assistant interprets 'I want to learn about mutual funds' and 'I want to buy mutual funds' differently.

Lazy bot training would mean that slight semantic differences that make a world of difference to users would be treated the same by the bot.


Why being an AI-trainer/Domain expert is essential:
You need your bot to answer questions the right way.
You also need to define intents the right way.
You also need to anticipate what a user would typically ask, given the domain of the business.

If it's a bot for a telecom company, and you're asking for a document, you can bet that a user will cross-question why their PAN card is needed for verification.

Define these during the Conversation Design phase to improve the accuracy of your IVA and eliminate out of scope queries. Happy users, happy company, after all.

✨ A Conversation Designer is a Product Designer

Story:
All too many times, internal company stakeholders are not fully aware of what new technology can do for their company.
And considering that Intelligent Virtual Assistants are relatively new, every Digital Initiatives team that has “Deploy a bot” in their OKR makes the mistake of starting from scratch. This leads to what we call Internet Expertise AKA Head in the Cloud syndrome — where no vendor will ever be able to meet the highly bloated picture that your product team has in its head.

Result:
A Virtual Assistant is a complex software product.
Every brand is unique, and so should be their virtual assistant.
Treating any company’s virtual assistant like a cookie-cutter, the grab-and-go solution would be doing a disservice to the brand.

Solution:
Keep your brand’s voice, along with the final business goals in mind while designing your virtual assistant.

As the resident Conversation Designer at Haptik, I'm responsible for ensuring that our virtual assistants can solve problems in the most efficient and human way possible.

Even when a business team forgets to keep what possible in mind while looking at this new product, the Conversation Designer understands what the bot building platform can practically achieve and sets targets accordingly.

And like any good product manager, keeping an eye on the data and user behavior patterns yields gold. You can get genuine reviews of your products, understand user needs, and so much more through a quick perusal of your analytics information.

Tracking metrics like query completion, automation levels tell you exactly how much or how little you're helping customers. The proof is always in the conversation rate here.


And that’s what a day at my desk looks like!

Coincidentally this meets all four corners of the Ikigai framework as well. I’m doing
✅ something I’m good at (writing),
✅ something I love(UX design),
✅ something I can be paid for (training data creation)
✅ and something the world needs(creating technology that can handle how the next billion users converse)

TL; DR. Conversation Design is a pretty great field to explore if you love UX, writing, logical thinking, and Product Design. If you think you need one, talk to us, or start thinking of hiring one already.