Chatbot design: how to be disruptive in work without getting fired

“Stakeholders are the invisible third rail of user requirements gathering and successful IT adoption in the enterprise. Forget about them and you may be badly shocked!”

Ultan Ó Broin (@ultan) looks at suitable chatbot use cases for the cloud enterprise and then how to approach designing them right.

Image licensed from iStock

Chatbots: Your New #BFF in Work

OK, let’s nail this one.

Chatbots are not about automating humans out of the #lovemyjob world of job satisfaction.

But, they are definitely about improving the user experience (UX) in work across the entire journey of getting a task done:

• Chatbots are about taking the drudgery out of jobs, about offering real people doing real jobs in real places a natural way to do that stuff easily, so they can then focus on real work.
• The chatbot approach of talking tasks to completion means users can work “in the moment”, insulated from complexity, and they can think about going home early, about spending more time with their kids, whatever. Unlike a lot of work applications, chatbot users don’t dread logging in again the next day,
• Chatbots mean IT departments can focus on serious enterprise issues. HR can relax about hiring and training. C-level execs can dial down the pressure about productivity levels and maximizing participation in their SaaS investment.

Everybody ❤️ donuts. Donut bot on Slack takes advantage of casual collisions in the workplace so that people work better, together.

Time for a chat?

Identifying chatbot uses in the enterprise are a classic “it depends” UX situation, but start with tasks that:

• Involve employees spending a disproportionate amount of time and effort on simple things they could talk about doing in seconds.
• Need special apps or devices obtained through an official procurement process.
• Are best done on the go, right now, even inside the office.
• Are time-critical but often deferred, yet doesn’t require much data entry.
• Can be completed by providing readily at-hand information in any order.
• Are no-brainer self-service models anyway.

Some enterprise chatbot opportunities to get your innovative juices flowing might be when:

• 80% of your IT service calls are easy password or PIN change requests.
• Sales reps need to order new marketing merchandise every time before a road trip.
• Contractors must submit expenses to meet an accounts payable deadline for payment.
• PMs want to proactively schedule meetings with remote team members to take a quick project decision.
• HR onboarding wants to hook up interns with other team members in a friendly way.
• Managers have to send candidate interview feedback so the hiring process can move forward.
• Light bulb innovation programme moments that need to be recorded while still fresh in the mind.
• Field technicians need to record time on the job, when they started, finished, took breaks, and so on.

Critically in the cloud enterprise environment, a task and ask might be standalone sure, but they can also be part of a wider SaaS process where the user may switch to another device or location to finish a task or augment that application data later.

To-do bot for Slack. For creating to-do lists within your chatbot conversation. Don’t overlook simple use cases!

Remember, even the simplest use cases (such as creating a list of things to do within a conversation) can be powerful productivity enablers that add up to major wins for the enterprise.

Being Disruptive in Work

A good chatbot opportunity disrupts an existing process that might previously have been done using a graphical user interface (GUI) for application data or no IT at all!

For example, if your company relies on paper timesheets scanned in by contractors and emailed to the accounts department to calculate project-based compensation or the retail team has to line up in the back office to punch their time into a kiosk UI before and after work, then a chatbot might be worth a shot.

Long, data-intensive, manual entry tasks or having to read large chunks of online content impossible to glance at on a small device are not recommended. We all know that lengthy form-filling is a major conversation stopper, chatbot, or otherwise.

Essentially, asking first why a chatbot is the best solution trumps the“me-too” approach of simply lifting and shifting old ERP green screens into a messenger or bot UI experience. That latter approach won’t succeed. You need to design your chatbot UX.

So, really understand why a chatbot is being hired for the job before you address the mechanics of how. The Jobs-to-be-Done approach used by Intercom and others lets you do just that before you commit to building a solution for your enterprise. So, if you find that Microsoft Excel spreadsheet is just fine for the job at hand, then stick with it and explore other use cases for “botification”.

Chatting up accountants

Enterprise UX design is not just about gathering requirements from the end-user.

A broader stakeholder-inclusive context of use is critical: where end-users work, with whom and with what inside and outside of the organization, who they interact with, what goes on online and offline, what are the cultural, legal, or reporting conventions, who is making the project decision and approving the budget, what chatbot platform features are available, and more. That whole contextual journey to getting the job done must be considered.

Yes, for your chatbot design and development, techies may have to sit down with accountants, clerks, road warrior reps, and more. All will have to play nice during the design process of a chatbot. Agree how to test designs early before development starts and when the “But all I wanted was . . .” surprises come.

Remember this: Stakeholders are the invisible third rail of user requirements gathering and successful IT adoption in the enterprise. Forget about them and you may be badly shocked!

Be safe, be seen, be heard

In the enterprise world, there are other IT considerations you need to plan for, and chatbots are no exception:

• Provide for business efficacy and security around how that business data is transmitted, stored, and retrieved. For example, before signing off on a contract or funds transfer, an electronic signature or authorization code might be required. For the financial period closes, audit trails and other reporting requirements must be met. Even at a coffee shop transaction level, enterprise data security and policies apply: nobody should be standing in line in Starbucks yelling the latest sales or hiring opportunity details into their smartphone. Typing that text provides for information discretion — the beauty of chatbot modality!
• Ensure the performance of the chatbot experience matches the context of use. Can your data be committed quickly to the cloud? Can it be retrieved? How fast are the graphics rendered? How fast is the AI and NLP processing for those “What if?” questions?
• Meet accessibility requirements. In the U.S, IT accessibility is driven by what we call Section 508 requirements, but there are other requirements worldwide such as those from the W3C. Just think of accessibility as simply “the right thing to do”. Without accessibility, equality in the workplace cannot be provided for, and you may encounter legal problems.
• Localization and internationalization considerations are vital for growing markets and for accommodating how people really work differently around the world. This means understanding cultural nuances and local work patterns, but also allowing for the easy translation of UI prompts and images, for text expansion, character set and writing direction, built-in global date and time formats, global currency support and multilingual capability, and so on. Language is a form of user experience that is critical to any conversational UI such as chatbots, reflecting how we live and work. Facebook Messenger, for example, is available in multiple languages well beyond the basic FIGS (French, Italian, German, and Spanish). It’s critical that the platform you choose for your chatbot also has natural language processing (NLP) support for your users’ work intents and utterances, or at least plans to soon.
• Build for SaaS extensibility and integration in the cloud. That enterprise chatbot must be plugged in seamlessly to how people work with other applications and processes. For example, the Meekan scheduling robot in Slack enables teams to quickly schedule meetings that are synced with their regular work calendars.

Start talking sense

Although chatbot adoption is a consumer-driven phenomenon, those enterprise requirements must be met when designing chatbots for work. With the right design methodologies and stakeholders, however, that conversational-driven interplay of people and processes will make organizational sense and you can meet the challenges of this exciting and fast-moving chatbot opportunity.

A conversation about the enterprise chatbot context of use is welcome in the comments!

Ultan Ó Broin (@ultan) is an independent user experience professional, based in Silicon Valley and EMEA locations. Also a member of the editorial board of MultiLingual. Read the review of the 2017 CoverCon in Dublin for more: ConverCon: Talking and Walking the Conversational Interface Story in Dublin

All screen images by Ultan Ó Broin.