Lessons learned running a SaaS product for cryptocurrency enthusiasts
This article offers a wrap-up of the key learnings from running my latest side-project: a machine-learning-powered cryptocurrency price bot. CryptoProphet is a price tracking bot that uses machine learning to forecast prices and alerts subscribers of unusual market movement. Customers get alerts only when something unusual happens and don’t have to watch volatile prices.
As of today, 1,300 people registered, the bot sent 75,000 individual alerts and the forecasts got about 50,000 page views. You can read my previous blog post to learn more about the inception of the project.
In November 2017, I released a new version of the tool and started experimenting with acquisition channels, features, and revenue. In his talk Startup Metrics for Pirates, Dave McClure outlined a framework to get a high-level view of important metrics for a startup: Acquisition, Activation, Retention, Referral and Revenue. I will use these criteria to analyse my side-project.
Acquisition
If it is complex to build a technology product, it is even harder to get people to know about it and actually use it. “If you build it, they will come” simply does not work. The first step of the analysis is to get an idea of how prospective customers become aware of your value proposition. At CryptoProphet, there were two distinct periods of customer acquisition strategies.
During the first couple months, most of the traffic came from me sharing the tool in online communities such as cryptocurrency focused subreddits or Telegram groups. This hands-on user acquisition strategy was useful in the early days of the project. It helped me gather feedback from the community and drove the first iterations of the tool.
Unfortunately, it was not a scalable way to acquire users. First, it required constant work to identify new groups to promote the project to. Second, the traffic acquired through this channels drained out a couple of days after the promotion.
Later on, I realised that I was getting an increasing amount of traffic from search engines (organic traffic) and decided to invest time in search engine optimization (SEO).
Broadly speaking, for a given search query such as ‘crypto alert tool’, Google will rank your website on the result page based on its content and its authority. A first SEO tactic is to ensure that your content matches the type of queries target customers might type in Google.
So I added descriptions of cryptocurrencies to the relevant forecast pages and started to post updates about the best performing cryptocurrencies. The results were significant.
From one week to the other, the amount of organic traffic reach new levels. Google is now the source of approximately 70% of CryptoProphet traffic.
While the spikes in organic traffic can be attributed to my SEO work, I suspect the increase of interest in cryptocurrencies in the past couple of months drove most of the upward trend.
Activation
It is not enough to get traffic on your landing page. Visitors need to sign up and become active users. Each product has its own set of criteria to define what an active user is. For example, Facebook considers a user to be active if he or she has logged in and as had an interaction with the website (sharing, liking, etc.).
On CryptoProphet, users receive messages on Telegram whenever there is unusual market movement. A user is considered active if they have signed up for the tool and connected their Telegram account to the bot.
In the Activation step, the objective is to optimize the conversion of visitors into active users. Most of the work I have done there was to improve the website’s copywriting to make the value proposition clearer to visitors and to optimize the onboarding process.
I also added Drift messaging to enable users to talk to me from their dashboard if they need help, for example in connecting their account to Telegram.
It is worth mentioning that the value proposition of my tool is to keep user away from always checking cryptocurrency prices. The product is by definition not engaging: it is a one-way push communication to the user.
In today’s technology products, it is a norm to increase engagement using psychological tweaks (see books like Hooked). This rationale might makes sense if your revenue model is tied to time spent on your product (e.g. advertising) or if your defensibility comes from its stickiness (e.g. a productivity SaaS). However, a movement is now emerging to promote calm technologies. That is, those that require the smallest possible amount of attention. And it is worth having a thought about.
Retention
Customer retention is a measure of the number of users that stick with your product overtime. A first good proxy for retention is to track the number of users that unsubscribe overtime, known as churn. It is useful to find areas of improvement through surveys with churned users and to make sure that new cohorts of users are more engaged. In my case, as users get more out of the product, their likelihood to bring in revenue increases.
It is fair to say that I have not spent enough time improving retention numbers at CryptoProphet. I only recently started tracking churn by tracking if a user had blocked the Telegram bot.
For CryptoProphet, it appears that the rate at which users unsubscribe remains fairly stable.
I believe that adding new cryptocurrencies tracked through the bot helps to retain users as they get more value out of the tool. But there is not always a straightforward link between adding features and improving user engagement. Many product suffers from feature creep: they get over-complicated to satisfy every users needs which often leads to negative results.
A good first step to improve user retention is to set up automated email campaigns based on the usage of your product. At CryptoProphet, users that signed up but never connected their Telegram account to the bot receive a reminder email. So far, this experiment brought back 5 users out of 511 (a ~1% conversion ratio).
Referral
Once users have been active for some time, they might recommend your product to their friends, colleagues, or business partners. The best referrals come organically from users who recommend your product simply because they like it. It is also possible to incentivise this behaviour. Doing so can turn referrals into a powerful engine of growth.
Dropbox had their early growth driven by a unique referral program. Users and their friends gained extra free storage space for each referred user. With this tactic, Dropbox grew from 100,000 to 4,000,000 users in about a year.
As with every new feature, there is a cost to adding a referral program to a product. Once it’s there you’ll have to reply to new type of customer requests and maintain active the incentive program.
Thus, CryptoProphet does not offer a referral program. Users do recommend the tool to their friends, though, as you can see below. Always motivating to read.
Revenue
The last step of the framework raises the most important question: how much money are you making. Of all the steps of the framework, this might be the one that is the most specific to the business you have. No two business will share the same revenue profile.
CryptoProphet is a freemium service. Free users can only subscribe to one cryptocurrency price alert, premium users to an unlimited number of alerts. The subscription price started as $12 per year but I kept increasing it as I was adding new cryptocurrencies to track. Today, a premium subscription costs $30.
The price increase did not negatively affect the number of new subscriptions per week. It did lead to an increase in revenue. Indeed, there is a lot of literature about the correlation between a product price and its perceived quality by the customers. A price of $30 sends a strong signal of the value a premium subscription will deliver.
I chose to charge a yearly subscription rather than a monthly subscription and generate as much revenue as possible upfront. I based my decision on two assumptions:
- My tool is dedicated to the highly volatile cryptocurrency market, the longevity of which is questionable.
- My value proposition targets users that are not much involved with the market: they want to stay in touch with the prices, but their interest is likely to drop off sooner, rather than later.
From the graph above, you can see that each period of high volatility of Bitcoin’s price is followed by a increase in web traffic on CryptoProphet. On top of that, the number of daily visits is growing with the price of Bitcoin and its increasing media coverage. I suspect that the same pattern applies for most cryptocurrency related websites.
I am certain that my service (and most other cryptocurrency tools) will die or at least see a massive decrease of users following any major crash. If this happen in the next six months, a customer registering today for $2.5/month and disappearing with the crash will bring at most $15 in total. By charging upfront a yearly subscription, I am getting $30 today.
Conclusion
Developing CryptoProphet has been a great way for me to learn about product development. I did not achieve sufficient traction to make a sustainable business. As long as I keep a calm-tech value proposition the level of user engagement will stay low. Further, the service is evolving in a risky and volatile industry.
I hope that this case study demonstrated a structured way to look at your business metrics. Which metrics you need to focus on depend on the type of business you run, and this is not an exhaustive list but rather a first high-level framework. To gauge your business’s health in depth you will need to consider measures such as Customer Acquisition Cost, Customer Lifetime Value, and many others. For further reading on the subject, I suggest a16z’s Startup Metrics series.
