Tom Hewitson is a Webby-award-winning conversation designer, Alexa Champion and founder of Conversational AI studio labworks.io - one of the largest publishers of Alexa Skills in the world. In this article, he outlines four steps every skills developer should conduct while coming up with an idea for a new Alexa Skill.
Over the last 6 years I've published something like 80 skills across 15 countries - including some breakout hits such as True or False?, Would You Rather?, Trivia Hero and Nightlight.
As part of my Alexa Skills development journey, I've developed a method for evaluating skill ideas and deciding what to build next. In this article, I’ll outline three tips to help other developers find the right use case for their skills (pun intended).
So, let's get into it:
This is controversial but I’d argue that, when you’re first starting a project, market research might be more useful than user research. If someone else has already built whatever it is you were thinking of, and didone a good job, it’s going to be much harder to be successful - you’ll have a much better chance if you can find less contested ground.
It is also worth asking yourself the question, “If no-one else has built this, then why not?” It could be that you’re the first person to have come up with the idea, but it’s more likely that there’s a problem you haven’t yet uncovered. Don’t waste months building something that will never work (something I’ve done plenty of times)!
You can start your market research by checking out the Alexa Skills Store to see which skills are performing well. You can find these by looking for those that have a large numbers of reviews, as that’s usually a good proxy for how many users a skill has. This will give you a good idea of what types of skills are popular. You may even be able to spot an opportunity where a skill is getting lots of usage but users aren’t completely satisfied.
The same approach can be used on the various mobile app stores. Look for apps that perform well there that don’t have skill equivalents. This can be a bit harder as user expectations around apps are quite different from skills and not every experience will translate well to voice — however, this research can still provide a good source of inspiration.
Once you’ve created your list of competitor apps and skills, take the time to play with them, read the reviews and understand what users like and don’t like about them.
Look for common themes or characteristics that make these apps and skills stand out such as:
Look for opportunities where you can solve a user’s problem better, faster or cheaper than the existing solution. If you can find a way to do all three then you’re onto a winner! An easy way to do this is to provide just the part of the experience that users find valuable without the bits they find annoying. However, you can also develop a unique take on the solution which changes the way the user thinks about the problem.
In an ideal world your competitor research will turn up a list of opportunities that you could work on and you’ll be able to choose which one you like best. This is especially helpful as it will stop you getting too fixated on your first idea, and will make it easier to switch if you run into problems once you start work.
Now that you’ve got a list of potential opportunities you’ll want to make sure you’ve fully understood them so you can prioritize them. User research is a great way of doing this.
While it is possible to do lots of user research, and build a skill for a group of users that doesn’t include yourself, it is so much easier at the start if you build something that you would use too. In fact, I would go as far as to say that if you can’t imagine using your own skill at least weekly then perhaps it’s not the right idea for you, even if it could be great for someone else.
Assuming that you would want to use the skill you’re thinking of, the next step is to get out of your own head and try to verify if other people feel the same. A really easy way to do this is to run an online survey (there are plenty of companies out there that have a pre-existing audience of qualified survey respondents with an Alexa device).
For an initial survey you should look to collect around 100 responses. However, it’s completely fine to ask the respondents about multiple ideas in the same survey — in fact, asking them to rank which ones they prefer can be a very effective way of getting a clear answer on what is the best problem to tackle first. From this first round of responses you should have a clear signal on how people feel about the skill you’re proposing. As with many things in life, if the answer isn’t a definite yes, it’s a no!
The other benefit of running the survey is to get the email addresses of people who are interested in your ideas. You can invite these users to more in-depth conversations via video conferencing. Don’t skip these conversations, they are where the best and most important insights are discovered that allow you to turn your ideas into real products. Again, it is completely fine to ask people about multiple ideas and which ones they prefer the most.
While carrying out these conversations try to ask questions about how people use Alexa and how your skill ideas would fit into their lives. I also like to ask them for ideas about what they wish Alexa could do to feed into future ideation sessions and as a way of understanding the broader appeal of Alexa outside of the skills I make.
You’ll know that you’ve completed your ‘discovery’ user research when you can confidently select 2 or 3 of the opportunities from your market research that you know exactly what users want and that they would use a skill doing that if you built it.
Now that you’ve established your user needs it’s time to figure out what’s going to be involved in building your ideas. This is why you should try to keep 2 or 3 ideas in the running as it often turns out that some ideas are significantly quicker and easier to build than others.
To estimate how much work a skill is going to take to build, try to break it down into as granular component steps as possible. At first glance an idea can seem pretty simple but once you start thinking about all the different bits of data, visuals, and prompts involved, it can turn out to be much more complex. Unfortunately, this is an ability that you only really get with experience - the more Alexa Skills you build the better you’ll get at estimating how much work it will take to build the next one.
Something that I do is to build a small technical proof of concept of what I believe to be the most complicated part of the project to see how long it takes. I usually set myself a time limit of a couple of weeks to get something working, and if I can’t do it in that time then I’ll move onto one of the other ideas instead.
Once you’ve established something is technically possible, and have a reasonable sense of all the steps involved in building it’s you can move on to estimating costs, revenue and profits on your project.
I’m assuming that if you’re reading this article it’s because you want to make money building Alexa Skills - I know I do!
If you’ve conducted your user research well, you should hopefully have confidence that if you built the right Skill that the people you’ve spoken to would pay for it. From the people you surveyed, you will hopefully have a sense of how many of them would be willing to pay and at what price. Depending on the size of your market you might want to run a larger survey to make sure you have enough data to base your forecast on.
From this you should be able to forecast for each of your ideas how much you could potentially earn and how much it will cost you to develop. Once you’ve got this it’s a simple task of selecting the project that will be most profitable (or a combination of the revenue and what’s most enjoyable to work on depending on your disposition).
A word of caution: avoid projects where there isn’t a significant profit as it is almost certain that you will have been over optimistic in both estimating the potential revenue and the potential cost (we all do it, it’s human nature!) The best way of combating this is to make sure your project would still turn a profit even if it was far less successful than you’re predicting. For example, you can estimate that your skill gets a tenth of the users you predicted and takes twice as long.
Most of all, don’t over-invest in any single skill. The best way to increase your chances of success is to make sure you don’t have all your eggs in one basket. That way, even if one of the skills you work on ends up losing money hopefully some of your others will make a profit that balances it out.
At the end of the day, building Alexa Skills is a business like any other and coming up with the right skill idea is essential to making that a success. By using market user and technical research, plus forecasting, you can prevent yourself from making any missteps and increase your odds of building a Skill that will be a success.