Morning all! Chief Fretting Officer Kirk here with another fret. How do we keep the lights on?
A core value at Keen is radical transparency. You can find examples of this all over our blog (e.g. how did we fundraise and how did an early employee negotiate comp). However, all of my professional training up until Keen has been more along the lines of radical opacity. It wasn’t that there was ever anything to hide, but rather that we strove for simplicity. “Does this directly matter to the customer? No? Then STFU.”
Anywho, we’re at an inflection point in our business, and we want to walk the walk of transparency and seek feedback from the community. The question: how should we price this service?
Quick primer on our business for new readers: We make a kickass set of APIs that let mobile app developers collect, analyze, and visualize data.
For instance, if you had a social-local-mobile shopping app (SoLoMo for you cool kids), you’d probably want to track every time a user opens the app, connects to Facebook, likes/comments/shares an item, adds an item to their shopping cart, removes an item, and completes a checkout. You can easily grab this information using collection APIs.
Once you’re doing that, you can make a little analytics dashboard for your product manager to agonize over. I bet she’d like to know how many people opened the app each day over the course of the last week? No problem using one of our Analysis APIs (specifically the Series API). You’ll be a hero.
Wanna check it out? Hop on the beta list. Alright, back to talking Benjamins.
Here’s what we know:
- It won’t be free, but there will always be a free tier to let developers play around.
- We want to build useful technology for developers. It’s actually pretty easy to imagine a way in which we could offer the tech for free and monetize through other channels (i.e. consolidated usage information, vertical-mobile intelligence, etc.). It just creates an incentive conflict. We think you should pay us so that we have a direct incentive to make the product awesome for you.
- If we really believe that we’re playing in big data (we do!), then we can’t roll-out a pricing system that discourages you from sending big data. Less data means poorer outcomes for you. Poorer outcomes for you means you’ll leave Keen. If you leave Keen, I have to move back in with my mom in rural Illinois.
- For big, enterprise customers, we’re happy to consider one-off pricing where it makes sense. It’s likely that the BigCo customer will have extremely specific needs and constraints. They might start-off with published pricing, but as the relationships grows and changes, we’ll engage with greater specificity.
- Contrary to conventional wisdom about bulk, marginal costs actually increase as you scale with data. At least until Costco starts an analytics API company, the marginal cost of more and more data will probably be higher.
Here’s what we think:
- Predictability will be paramount. Whatever solution we implement must make it simple for you or your Chief Fretting Officer to understand how much our service will cost.
- Our customers (you!) will be folks with apps that have steady, predictable traffic (i.e weather app, generic e-commerce) as well as time-boxed, explosive growth, rapid decline cycles (i.e. Super Bowl related app, tradeshow apps).
When we put all of that together, we brainstormed a handful of frameworks with which we could price.
Model ourselves after energy providers. Everything is metered and there could be some base charges. Often referred to as cost-plus pricing.
- Absolutely transparent. You used this many units this month and the cost is $5 per units.
- We can probably be cash flow positive on every customer, rather than having some customers effectively subsidizing others.
- Feels as though it reduces us to a commodity. Lean hogs for November delivery: $72.725 per pound. Keen API for September delivery: $6.28 per arbitrary unit.
- Not necessarily predictable. Have you ever gotten an electric bill that was inconsistent with your expectations? Yeah, me too. Running a grow house is totes expensive.
Not value as in low-cost, but value as in “how much value do you get out of using Keen?”. Good analytics have value. Duh. So, why not set a price that we believe reflects the added value our customers will get from using our service. We could then price per user, per application, etc..
- Makes us about demonstrating value. If we say that the cost of using Keen is $x/user, then we’d better damn well produce at least $x in added-value.
- Very straightforward. Price = unit * $x.
- We might inadvertently limit our customer potential to only those with high-ish per app or per user revenue.
- Potential unpredictable price for you. We don’t want you to have to think about restricting new users because you don’t want to spend too much on analytics. “Sorry, Timmy. The CFO says we can’t afford to give you have access to the Lassie Lifeline app. Better steer clear of the well.”
- Could be problematic for us if we encounter a lot of resource-intensive (i.e. storage, bandwidth, processing) customers. Readers who have been to a Las Vegas buffet know what I’m talking about.
Tiered Flat Fee
We could offer a few pricing levels with incremental features. Pretty common for all of the BuzzWord as a Service companies out there.
- Extremely predictable. You tell us you want to pay $x per month and then you pay $x per month until you tell us that you don’t want to pay $x per month anymore.
- Extremely familiar to the customers we have and expect to have.
- As our customers are going to use our service in different ways, we think it’ll be difficult to set-up the tiers such that we account for the variety of uses.
- Creates a slight separation between use/value and cost. There’s something about you signing-up for a tier that you never fully use that is inconsistent with the company we want to be. We don’t want to be that juicer you bought because it’ll pay for itself after only 3,728 gallons of carrot juice made at home.
Customers can buy bundles of credits and each of our features has a per credit cost. Microsoft, for example, does this with Xbox.
- Offers tremendous flexibility. You effectively pay for only the individual features and services that are most relevant to their business.
- Gives Keen excellent insight into which features are actually generating value for our customers. If it doesn’t help, they won’t pay for it.
- This is the most complex way of pricing. And, at the very least, it lacks some harmony with our transparency values.
- Higher management costs on the Keen side. We’ll have to meter lots of different activities with a high degree of accuracy. I’m not worried about our ability to do it; it just has all the thrill of doing your taxes.
- I felt the need to reference Microsoft in explaining the pricing model.
A La Carte Add-Ons
Regardless of the route (or combination of routes) we choose from above, we could also include some added-value services that are priced separately. A few possible examples:
- Phone Support vs. Email Support vs. Self-Service
- Data Modeling Support
- Data Retention Beyond Published Timeline
- Miscellaneous Money Laundering
I’m willing to wager a significant sum of money that the right answer is some combination of many of the above ideas. We’ve been thinking about the problem a lot, and we’re extremely committed to getting this (as close to) right (as possible). Now, we want your feedback. Send us an email or, better, start a conversation in the comments. What makes the most sense to you and your work? Anything you hate?
Alright, that’s all I’ve got. Again, we really want to hear from you about this. Not in the habit of commenting on blog posts? Well, I’m not in the habit of publicly detailing my thoughts on how to build a revenue stream. It’s a new world.