Users no longer want to interact with others in the same ways they used to. Instead, users value the convenience of a conversational, more human, interaction. This shift in human behaviour has given rise to next big thing: chatbots. Here are the best analytics tools to measure them.
Why invest in chatbots?
The benefits of chatbots are clear. You can save on customer support costs, respond faster, engage with users more frequently and outside of hours, get feedback quickly and gain an insight into your users.
Have you ever overheard just one side of a phone conversation and been totally confused? With chatbots, your users are literally telling you what they want.
Messaging apps now have more monthly active users than social networks (BI Intelligence, 2017). You might think this is proof enough that chatbots are going to revolutionise customer-business relationships, but a survey from Warwick Analytics earlier this year suggests “59% of those businesses who have a chatbot were either unsatisfied with their chatbot or thought there was room for improvement.”
So what gives?
- Yes. Humans are confusing. You won’t understand them at first, but eventually they’ll love you.
Businesses are struggling with chatbots because they assume too much. They assume that chatbots are plug and play; they think they can bolt them onto their existing digital platforms and expect to get instant benefits. If you’ve ever visited a website and had your existing intent interrupted by a pop up asking “How can we help you today?” UGH.
You, along with many others, will share my pain. A bad chatbot is worse than no chatbot.
Can you measure how useful a chatbot is?
53% of the consumers surveyed (54% US and 52% UK) found chatbots to be “not effective” or only “somewhat effective”. – chatbots.org, 2017
Don’t be left thinking, “Was it something I said?”
If you want to do chatbots the right way, and get the same benefits mentioned previously, you need to start measuring what’s going on. You need Analytics.
Analytics and chatbots are two sides of the same coin. Just like with your website, you can’t speculate that your users know what they want or that they know how to get there. You need to validate and make sure you’re serving customers with certainty. I’m going to show you some of the available services that will help you.
So can I just use Google Analytics?
Google Analytics is an incredibly versatile suite and, with enough knowledge and time, you can make it accept data from most platforms. The problem is that the Google Analytics tool is built and designed to gather data on websites, not chatbots.
Most of the metrics used to measure websites (such as sessions, session duration and bounce rate) can be relevant to both platforms, but some measurements aren’t. Page views mean nothing in relation to a chatbot and how is Google Analytics going to mark an unhandled message, a 404?. This is why you need a different service built specifically for chatbots.
How can I link my chatbot to an analytics service?
Some chatbot services such as Chatfuel and Dialogue Flow offer basic built-in analytics. However, currently no external service has created an easy solution for linking up their analytics data to chatbot creation services. It all currently involves playing around with API keys or code snippets.
Coding? Linking APIs? Sounds terrifying…
The idea of learning how to use another service or playing around with APIs can be scary and time-consuming. But if you aren’t committed to the whole process of validating and optimising your chatbot – an ongoing process that will never end – then don’t consider a chatbot. I repeat, a bad chatbot is worse than no chatbot.
That doesn’t mean it needs to be perfect first time! I guarantee it won’t be (and if anyone tries to sell you that dream I’d love to meet them). But with proper analytics, testing and learning, perfection will get closer every time.
So, without further ado, here are the chatbot analytics services you need to use:
Chatbase is both familiar and effective. That’s probably because it’s made by the same people, Google.
- Screenshot from Chatbase for chatbot analytics.
It’s Google Analytics for chatbots. That means it benefits from the years of experience Google has developing their existing suite. As well as Google’s incredibly powerful algorithms, it includes an AI that will automatically look for ‘problematic patterns’ and suggest improvements to improve accuracy, enhance user experience and increase conversions.
- Chatbase validating user behaviour and suggesting optimisations for chatbots.
You can also link your chatbots from any platform to Chatbase, and even distinguish between each of them when reviewing the data. Setting Chatbase up is reasonably simple, although like all these services it will involve sharing a generated API key with your chosen chatbot platform so they can communicate with each other.
Dashbot promises to help you increase user engagement, acquisition, and monetization through ‘actionable bot analytics’. This means that Dashbot will sift through every single message sent through your chatbot and create meaningful insights that can be used for optimisation.
Dashbot also supports Sentiment Analysis: a clever report that looks at the words each user uses and their behaviour to create an overall sentiment for each session.
- Dashbot gives a more human analysis of conversations with sentiment analysis.
Similar to Chatbase, Dashbot supports a range of platforms from Facebook Messenger to Google Assistant and Alexa. It will also let you compare engagement between the services and see where your users are engaging most.
- Validate by easily seeing how your chatbot is performing across multiple platforms.
Botanalytics, like the previous mentions, supports multiple platforms including Facebook, Slack and Kik. Botanalytics will help you understand your engagement rates and what kind of engagement your users are having with your app.
By telling Botanalytics the stages of an engagement funnel it will even keep track and report back to you on how many users are making it to each stage. This is helpful for discovering potential pain points where users stop engaging.
For example if your funnel is ‘initiating an order’, ‘entering payment details’ and then ‘placing an order’ (this would be the whole funnel), Botanalytics could recognise that only 13% of users are placing the order and from there you could work on improving that stage.
- Overview of funnels and how far your users get through a funnel.
Botanalytics also serves up more insights around your users; giving you both quantitative overviews, such as how many messages are sent, as well as allowing you to track specific conversations, and see the exact messages each users send and how the bot responds. Setting up BotAnalytics is more hands on than the other services but you’re rewarded with a clean interface and easy to understand metrics.
Woah! That’s a lot of information to take in…
Phew – a lot of information there so let me summarise what you should takeaway from this. Chatbots let you have a closer relationship with users by letting you respond quicker and extend that relationship out of working hours.
The real value of chatbots is that it lets you understand what your users really want. Setting up chatbot analytics is all part of this process. The services I’ve mentioned help you validate what your users want, meaning you can work towards providing a more engaging and smoother experience for them. By optimising your bot with analytics you will increase your customer relationships, retention, conversion and more importantly, understand them better.
Or in short:
Analytics + bots = <3
If you need help validating your idea, developing your use cases or creating a prototype, you’re in luck because that’s what we do – so feel free to get in touch.