- Starting With Game Design
- Typical Microtransaction Design
- Alexa In-Skill Purchasing
- Pricing And Selling Experience Design
- In Closing
So last week’s investigation was focusing on monetisation - looking at the viability of making money from Alexa skills as well as ideally identifying successful monetisation models to adopt. I’d intended on looking at and adopting successful models from the broader gaming industry, and things didn’t quite work out that way. But it’s all good news!
Instead, what I got was a primer on the process of good microtransaction design, an understanding of what capabilities and constraints exist in the Alexa space, and an understanding of design considerations to take into account when ‘selling’ to an Alexa skill user. You can find each section in the table of contents for this post. Now, all of this comes with a caveat that this is one week’s investigation. Maybe because more than anything else, survival and success can hang on monetisation strategy, I feel the imposter syndrome on this subject acutely. I feel like I’ve got a good starting point here, but this is a topic that folks build whole careers and consultancies on.
Starting With Game Design
Please consider the term ‘game’ and ‘experience’ interchangeable throughout this post. I’m going to default to using ‘game’ mostly here. Before monetisation can be added to anything, a stable foundation needs to be in place. The core game loop needs to be understood. What systems are in place for player progression? What do the game systems look like? What are the innate reward systems for a player?
These need to be understood first so that monetisation systems don’t fight with or replace the core experience and progression. Also, monetisation systems need to interact with the opportunities provided by core game systems. You can’t sell hats if players can’t equip hats.
Typical Microtransaction Design
Speaking from the perspective of traditional digital games, the common wisdom seems to be that monetisation means microtransactions., especially in the mobile space where developers are less likely to build a business model on digital sales (less likely, but not impossible!) Design then starts from understanding user motivation - what do people want from things? Typical motivations when it comes to the needs games fulfil are mastery, autonomy and relatedness.
(Incidentally, a lot of this foundation design is cribbed from this excellent GDC talk by Crystin Cox)
- When it comes to Western audiences, mastery is culturally sensitive - we have an aversion to ‘pay to win’ games as a rule. But monetisation can certainly mitigate difficulty - hinting and helping, for example.
- Autonomy as a motivator seems one of the richer potentials for monetisation with Alexa - trading time for money in progression systems, or providing access to more content.
- Relatedness presents more of a challenge with Alexa skills, despite being one of the richer areas for traditional digital games. Player expression through cosmetics gives a rich potential for creating relatedness in online communities, but as a primarily audio medium, how to achieve this in Alexa requires more investigation. There’s likely some potential here when surfacing information in custom web apps or using community hubs like Discord to communicate.
Finding an optimum price point for microtransaction content is another key challenge in traditional game design, and a whole art form unto itself. Games operate as monopoly economies, and perhaps surprisingly have relatively inelastic demand. Players will, when given a range of purchase options via an in-game offer, purchase whatever the cheapest option is, regardless of its actual price.
A key consideration with pricing is also balancing monetisation and ethics. The optimum price point for maximising revenue for a game may be much higher than the optimum social point - if developers are attempting to draw and foster a player community, higher price points will likely create backlash, as well as driving away less affluent players, particularly if premium content creates a seemingly insurmountable competitive edge.
Putting it more bluntly, the question becomes, “How evil do you want to be?”
Alexa In-Skill Purchasing
It’s worth discussing the options that Alexa provides for monetisation next, and how those systems work. Alexa offers two models for monetisation - selling real-world goods via Amazon Pay, and in-skill purchasing of skill-related digital goods. We’ll focus on in-skill purchasing here.
Alexa offers three different kinds of In-Skill Purchase (ISP) with their own price ranges.
One Time Purchase - $0.99 - $99.99
A one-time purchase is a purchase made once within a skill that then provides an entitlement to the user forever, unlocking some sort of content or functionality.
Subscription - $0.99 - $99.99
A subscription is a regularly renewed (typically monthly) purchase that provides access to regularly updated content.
Consumable - $0.99 - $9.99
A consumable is a purchase that can be used once within a skill (for example, allowing a user to save their progress in the Yes Sire skill, or a hint in a trivia game)
Each purchase offered in an Alexa skill is one of these three categories, and must be configured as part of the skill configuration provided for certification. So while the exact content provided in a pack of new trivia questions may vary, the price point and description of a purchase is set as part of skill submission.
The act of purchasing itself is handled by Alexa - the skill ‘hands off’ to Alexa which provides detailed pricing information (a skill must never mention pricing information as part of its direct messaging to users, as Alexa may apply user discounts at runtime for situations like servicing Amazon Prime customers)
Each skill must have some sort of free content for the user to experience first before being asked to purchase (we’ll come back to this in a second)
This creates some definite challenges for developers. Price points must be set ahead of time, and cannot easily respond to user behaviour over time without the skill being re-certified. This complicates or makes impossible activities like A/B testing price points or moving prices along sliding scales to determine optimum price points and user behaviour. All users must be offered a standard experience and price point. Which brings us to…
Pricing And Selling Experience Design
The first consideration for pricing is what are users paying in competitor skills and in similar markets? For example, if a skill is selling audio stories, developers would need to consider their customers purchasing power in alternative markets like Audible.
Secondly, from a sustainability perspective, price points need to consider the cost in time, effort and resources to create each piece of content. Then factor in:
- size of skill user base
- likely conversion rate for users to paying customers
- cost of additional activities like marketing spend
- desired profit margin
- retaining 70% of incoming revenue (30% is taken by Amazon)
This creates an individual (and likely complicated) formula, to start to determine a suitable price point. It’s worth noting that I still have some unknowns here - user acquisition remains a mystery, although success stories are definitely out there. The design of the selling experience presents unique challenges as well. Unlike a screen-based experience, there is a single channel of communication with the user, making undesired sales pitches far more frustrating.
An excellent series of Alexa videos on ISP design contains the recommendation to model selling within Alexa on face to face selling techniques. Understand the user, understand their needs, and sell to them then.
Not only should a skill present the user with delight first, and allow them to fall in love with the experience, but a user should also be familiar enough with a skill’s systems to understand what they are being sold before a given offer is made.
It is critical then to understand user behaviour - how often does a typical user open your skill? How long do they play for on average? How long could they play for? Finding the sweet spot of when to make targeted offers is absolutely when techniques like A/B testing can then be introduced to find optimum points for offering upsells. In order to understand user behaviour, it appears that a need for some sort of user tracking library suggests itself. We’ve looked at logging in the past for post-mortem analysis and reporting, but in order for a skill to make educated decisions on when to make an offer and when to pause on upsells, this information would need to be available to the skill at runtime.
As with everything, user experience underpins all of this. Offers should be targeted and brief, purchases celebrated, and refusals cheerily acknowledged. Skills should make it easy for users to understand what they’ve previously purchased, and purchase on their own terms.
There’s a lot here that feels like it’s still up for grabs, which is exciting. There are no acknowledged ‘best practices’ that I’ve bumped into, and even Amazon themselves ran a hackathon late last year to help discover ‘the future of ISP’. Most players in the market appear to be keeping numbers close to their chest, so hard numbers are hard to find.
Alexa has an install base of 100M+ devices world-wide, and a current market of around 80K skills, the majority still using text-to-speech.
That feels like an exciting combination of opportunity and challenge.