How to supercharge your Alexa skill
Creating an Alexa Skill or Google Action has varying degrees of complexity. Complete a spreadsheet, click a few buttons and hey presto you have your first voice app. However, you’ll end up with the “default” skill experience, there are thousands of Google Action Quizzes which anyone can create with a spreadsheet, all very similar to each other.
So, how do you supercharge your Alexa Skill or Google Action to make it stand out? Here are a few of the techniques and features we use when building voice apps;
Use voice-over artists, rather than the synthesised voices
It can be good for voice apps with lots of content or dynamic elements to employ purely Alexa or Google synthesised voices. However, the default voice from Alexa and Google can lack character at times. In order to combat this, you can use SSML (Speech Synthesis Markup Language), tap into Alexa’s new emotions and speaking styles to add some character to responses, or look into Alexa Polly, which turns text into more lifelike speech, for instance allowing you to vary gender and tone.
- Using a mix of voice-over artist with Alexa / Google synthesised
You might want to consider employing a mix between voice-over artists and synthesised voices in certain scenarios, for instance, for a game. For all static content use your voice-over artist and then ‘hand over’ to Alexa for any dynamic elements, allowing the ‘handover’ to act as part of the story in the app.
- Using a voice-over artist for 100%
If your content is static and you’re not going to repeat anything the user says, then use a voice-over artist. Build your skill with text first for testing, before spending budget and effort in the recording studio.
Screen devices
There are the three main reasons to create graphics to supplement your Alexa Skill;
- There are millions of Alexa devices which employ a tv screen, for example, the Fire Stick, the Fire TV cube and the new set of TVs that have Alexa built into them
- The Alexa Show and Google Home Hub devices have a screen
- Google Assistant is on all Android phones version 5.0 above
To enable images and video interaction into your voice skill, you can use APL (Alexa Presentation Language) for Alexa. This area of the industry is moving quickly, with Alexa Web API for Games stepping the game up a notch by allowing users to create visually rich content that runs off of traditional web tech. Google are also moving that way with their Interactive Canvas setup.
The negative of screen devices, is that it adds more design, development and testing time to your project.
If you have a skill already, look at your analytics which can track which types of devices have screens. In the meantime, use some of the template APL and image backgrounds on Google to give some elements of customisation.
Use data to customise the experience
There are several data points from the customer's device you can use to make your conversations dynamic and personalised.
- Username: you can then address your user “Hi, John”, “well done John, you answered.”
- Email: when appropriate you can email the user.
- Address / location: use the location to change tone, slang or data e.g. “your nearest store is...”
- Distance and temperature: do your users prefer Fahrenheit or Celsius? Change your content according to their preferences.
For the above, check documentation on what requires permissions to ask the user, make sure it’s a onetime ask so you remember the settings for the users next visit.
Combine time of day and the user’s history to vary copy
In the same way you would say good morning, evening and good night to people in everyday conversations, your voice app should follow this custom. You should then combine this with the history of the user – are they opening the app for the first time, are they a frequent user or a lapsed user.
Combining these two things for example, with just using eight variants, means you gain twelve variations on how your voice app opens, as seen below.
| New | Regular | Daily User | Lapsed | |
| Morning | Good morning, welcome to the quiz | Morning, let’s go, here is question one | Morning, good to see you again | Morning, nice to see you again, it’s been a while, here’s a reminder of the rules |
| Afternoon | Good afternoon, welcome to the quiz | Afternoon to you, let’s go, here is question one | Afternoon, nice to see you back again | Afternoon, welcome back, do you need a reminder of the rules? |
| Evening | Good evening, welcome to the quiz | Evening, let’s go, here is question one | Evening, it’s great to you have you back | Good evening, do you require a reminder of the rules? |
All of above can be done using code, it doesn’t require asking the user any questions. For example, asking them if it's morning or afternoon or evening. Further expansion on this is possible using days of the week, taking important days for example public holidays, Valentine's Day Christmas etc. can all be utilised to vary up the conversation.
The elements listed above should be seen as additional features to the core functionality of your voice apps. Such additional items can fall flat if the foundations of your app are not tight - design, handling errors and rigorous testing’s are essential. However, once you have those elements in place you can then look to start to supercharge your skill.
John Campbell, managing director and co-founder, Rabbit & Pork
Content by The Drum Network member:
