BLOG: Chatbot KPIs to Calculate your Chatbot ROI
Like all investments, you’ll be concerned with what your chatbot actually brings in for your business, specifically, the chatbot ROI.
Like all investments, you’ll be concerned with what your chatbot actually brings in for your business. Is it just patrolling your website after-hours, getting the occasional bot-snack out of the vending machine and waiting for the humans to take over in the morning?
To understand your chatbot’s performance, you’ll have to identify its key performance indicators (KPIs). These help ensure that your objectives are being met and that the investment has a desirable, measurable impact on your business. Luckily, your chatbot provides plenty of analytics for you to study, revealing not only information about its performance but also about your customers’ behaviour, new trends in the questions being asked, and a tonne of other useful data that can help you identify opportunities or potential threats to your organisation.
It’s important to remember that your chatbot’s metrics are specifically related to, well, chatbots. This means that the KPIs are significantly different from web-focused metrics. In reality, what you’re actually studying are conversational analytics that show how successful your bots' interactions are with humans and whether they’re getting the job done without any negative impact on your brand.
What are conversational analytics?
Conversational analytics utilise AI to study human written or verbal speech. It’s a natural language processing (NLP) solution that enables your bot (and the computer behind it) to understand and sort the data provided so that it gathers insights and makes meaningful responses. Data is gathered from calls, emails, chats and all other customer interactions. It is then organised with structured metadata in order to make it comprehensible.
Why are chatbot KPIs so important?
Getting your chatbot to the launch phase sure is exciting. You’ve learned a lot, experimented with different scripts, personalities, and even chatbot names. He or she is now ready to sit on your social media channels or website and be your eyes, ears, and mouth when it comes to first-level customer service interactions. But that’s not quite where your role stops. The quality of the customer service you provide will continue to be a key differentiator for your business – whether a bot or a human delivers parts of that service. This means that continuous monitoring and analysis of the bot is essential to understand how it’s being utilised and where it can be optimised. Here’s why your chatbot KPIs are a priority:
They enable data-driven decisions
Analytics and KPIs help you make improvements or add new scrips and upgrades to your automated service that will be genuinely impactful and measurable.
Provide insight into your customers
Chatbots give you an unfiltered glimpse of your customers journey. You can identify the preferred paths a customer takes to find a resolution and the patterns and trends they exhibit – all visible on a centralised dashboard.
Give your first-party data
Your chatbot is a revolutionary listening tool for your business. Conversational AI collects a treasure trove of data that’s all yours to study and review so that you can improve your service.
KPIs that help you understand your chatbots efficiency
Activity volume
This metric refers to the number of interactions the chatbot has handled – from first hello, to the more complex part of the conversation where the bot gets down to the problem-solving. This KPI will help you understand how frequently the chatbot is utilised and by what portion of your audience.
Retention rate
Your chatbot may have a high volume of users when it’s first launched – perhaps you’ve promoted your new customer service feature across your social media, and people are keen to try it out. But how will you know if the chatbot has been widely accepted as a go-to solution for future queries? Studying the retention rate allows you to see the number of users who repeatedly returned to your chatbot for support over a specified period.
User distribution per hour
Knowing when customers seek out your services is essential to ensure you provide adequate, robust support. The user distribution per hour metric helps you see at what times of the day your chatbot is most sought-after. You may see a pattern that certain types of queries come through late at night and identify a need to offer a limited call centre service to connect people to human call agents, for example.
Interaction rate
This metric tells us about the user’s engagement during the conversation. How many messages were exchanged, and did the user remain responsive during the interaction? Understanding interactions helps you to assess the quality of scripts and the overall usefulness of your chatbot.
Questions per conversation
Yes, we want the chatbot to be engaging, but it’s important also to recognise the value of your customer’s time. How many questions did the user have to ask before receiving the right responses or being provided with the correct information?
Non-response rate
This happens when your bot finds itself lost for words. This is usually because it may not have the right type of content programmed into its typical responses, or it may have struggled to interpret what was being asked. Either way, this is a close one to watch since no one wants to get stuck in awkward silence with a bot.
Goal completion rate
What percentage of the customer engagement through a chatbot was successful? Goal completion rate helps you understand how often the chatbot successfully processed the data it was provided by the customer and gave actionable/satisfactory information.
What do chatbot KPIs reveal about their ROI?
Of course, the crucial question is, how does all of this translate into revenue for your business? There are several factors you could look at to understand this. Essentially, you can apply a straightforward calculation:
Chatbot ROI = Profit from chatbot - (Installation/Build charge + Maintenance/Optimisation charge)
Chatbots can facilitate profitable interactions and cost-savings
There are a number of things to look at here. Firstly, if a chatbot is handling basic queries that previously used to go to a human, you’re automatically lowering the average price per contact by minimising the number of touchpoints. A customer can self-serve on a query, freeing your agents to do other, revenue-generating tasks, like outreach or providing the type of customer support that ensures retention and upsells. Your chatbot may also reveal to you that your operational costs can be significantly lowered. If the vast majority of calls coming through are to gain assistance on straightforward issues that don’t actually require human intervention, you may be able to introduce more flexibility in your call centre’s working hours and make significant savings on human resources.
Balancing installation costs with value for money
Building your bot and installing it on your platforms will require a process of integration as well as training and adoption by agents. Involving decision-makers and users in the early stages of your planning process will ensure that the bot is customised according to genuine customer support cases, enabling you to leverage the agent’s experience to ensure that the bot is as effective as possible. Deployment can be pretty expensive, but there are chatbot-building software services (like We Build Bots) that really put the power of chatbot design into your hands. You can quickly and easily programme your bot without any complicated coding or developer skills. Ultimately, you’ll have to factor in the type of chatbot-building process you’ve selected as well the time it takes to implement and integrate it into business operations when calculating its ROI.
Maintaining and optimising your chatbot
Your chatbot isn’t a set-and-forget type of guy or gal. You’ll have to frequently check in to see how it’s performing under pressure and if there are any improvements or optimisations that can be made. The cost of optimisation or maintenance will be determined by whether you’ve fully outsourced the build and management to a third-party chatbot specialist or used a platform like Intelagent to build and manage your bot independently.
The balance
If your bot successfully handles 50% of your queries, you’ve potentially gained 50% back in agent time and operating costs. You’ve also created a satisfied customer who’s potentially very pleased with the efficiency of your business and willing to return, time and time again.
