It’s not wrong to think of developing conversationalapplicationsas the new frontier in interacting with your consumers and customers.
Why? Because it is the new frontier, just like smartphone apps were ten or so years ago.
Know what's not new?
The need to measurethe success of the application you produce. Websites, smartphone apps, emails, social mediachannels, and now voice: their metrics all need to be analyzed to determine their success or failure. Yes, conversational AI is new, but digital analytics is not.
Fortunately, you can follow established best practices— most of which were developed during the creation and implementation of othertechnologies that were once new— to ensure conversational applications are meeting yourorganization’sgoals.
The Four Rules
Many organizations have high hopes for dashboardsand the myriad data points they offer.Butcan you be sure your conversational analytics meet the mark?You can start by following these four rules.
Rule #1:Start with your goals
What do you plan to achieve with your conversational application?What’sits purpose? What customer need will it fulfill?Are you trying to increase customer satisfaction? Increase revenues? Decrease call center volume?
Taking some time to determine what you hope the conversational application willaccomplishcanhelp preventone common pitfall regarding analytics—shooting an arrow and thendrawing a bull’s eye around it.
To follow thisrule:Be clear abouttheproject’s objectives and what outcomes you are lookingto achieve.
Rule #2:Determine how you measure success
Start by understanding what the key performance indicators(KPI) for the conversational application are.Whatdoyouneed to measure to determine if you’re meeting your goals?
Are you developing a consumer services bot that is meant to be sticky and engaging? Then, the KPIs you care about might be engagement and retention. Alternatively, maybe you want to reduce the time a person engages with the chatbot to get an answer. Or, perhaps you would like to minimize repeat transactions with the chatbot.
Once you've determined your goals (rule 1),understand whattheKPIsareto determine if your conversational assistant is meeting those goals.
To follow this rule:Identifythekey questionsregarding the conversational assistant's performance and then determine what data are needed to answer those questions.
Rule 3:Measure engagement
Now that you’ve determined what your KPIs are and what data you need to capture, becertainthat you can collect this data.Often, organizations identify KPIs only to learn that key information isn’t being captured or collected.
To follow this rule:Determine ifthe right datacan be collected. Then collect it.
Rule 4:Review, adapt, improve
Businesses change, requirements change,andhigh-level priorities evolve.That's why it is essential to periodically review the data collected by your conversational application. Do they still measure the goals the business hoped to achieve with this conversational application? Do the KPIs reflect the changes in your business environment and strategic goals?
To follow this rule:Adjustwhatdataarebeing collected to determine if the conversational application is a tactic that continues to help the business meetits newgoals. If not, consider collecting different datato evaluate its usefulness,or evenconsiderrevising the application.
Now that you know the rules, how can Orbita help you follow them?
Orbita Insights is an analytics tool that provides organizations withknowledge ofthe metrics that matter for their conversational applications.The tool allows organizations to define key performance indicators, measure the progress towards defined goals, and facilitate strategies towards optimizing and improving conversational experience metrics.
It is also flexible enough to adapt to changes insuccess metrics as your organization's strategic goals change.
Orbita Analytics is a feature-rich tool.Like all good analytics tools, it provides alibrary of charts, graphs, and reportsthat users can choose from as they review the data collected.
In addition, analysts can buildcustom reportsusing a reusable library of visualization components.For example, they can assess which intents are being used most often.
In this example, you can see that the Help is the top request. Knowing that data,the business hasfurther insight into how the conversational application is being used. From this, decisions can be made on how to improve the conversational application. For example, do things need to be explained better? Is the Help intentassisting people in the conversational experience?
With Orbita Analytics, data analysts canstart probing the data byasking a question, such as “How many questions were not understood?”
The insights gained from these results can help analystsdetermine if the conversational application is handling user requests properly or users are engaging with the conversational application in unexpected ways.