4 Key Metrics for Early Stage SaaS Startups
1. Introduction:
Start-ups are an exciting world with all kinds of insane and unworldly successes and failures made possible with the rise of computer technology and accompanying technologies like the internet, smart phones, etc.
One startup story that stands out to me is how Instagram launched on October 2010, and was acquired by Meta (Facebook) less than two years later in April 2012 for $1 billion dollars having 27 million users. What’s even more impressive is that they only had 13 full-time employees in April 2012!
So what exactly is a startup and what role does analytics play in them? Silicon Valley legend and founder of Y Combinator Paul Graham (PG) believes that the defining characteristic of a startup company is fast growth. As PG states, startups are differentiated from other businesses by two key characteristics:
(a) They are designed to make something that lots of people want
(b) They are designed to reach and serve all those people.
Software is a perfect way to accomplish both goals, because unlike a typical business like a barber shop or restaurant which have limited capacity and can only scale to a particular point, software companies like Instagram can reliably serve 27 million users using only 13 employees and a lot of well written code.
But building something a lot of people want is much easier said than done. There have been catastrophic failures where companies have invested millions only to create a product that nobody really wants. Take Quibi for example, which raised $1.75 billion from investors in April 2020, only to shut down 7 months later in December 2020 because ultimately they built a product people did not really want.
My goal in this article is to show how analytics can help increase your changes of start-up success and avoid building a product that is destined for failure. Since the most important metrics for start-up success will change depending on the business model and business maturity, I will start with a SaaS based start-up in its early stage. In later articles, I will delve into other business models and business maturity stages.
The 3 Engine Growth Framework:
Before we delve into analytics, I want to introduce a start-up growth framework to provide some background. Eric Ries, author of The Lean Startup, conceptualized the idea of the three “engines” that drive startup growth and associated KPIs (key performance indicators).
1. Sticky Engine — focuses on getting users to return and keep using your product.
2. Virality Engine — focuses on getting existing users to attract new users to your product.
3. Paid Engine — focuses on creating a sustainable income stream from users to allow new investment into growth.
Since there are limited resources, startups will need to focus on one of these engines depending on the stage in which a company is operating. Early on, it is recommended to focus on the Sticky and Virality Engines for growth and NOT on the Paid Engine because having a winning product will make getting paying customer a lot easier.
By focusing on each engine at a particular point in time, one can generate sustainable growth and increase the chances of creating a successful startup business.
3. The 4 Key Metrics:
Data Analytics is the study of using computational analysis to objectively discover and interpret meaningful patterns in data.
In the context of this article, the goal is to use computational analysis to drive growth for SaaS based startups. As already mentioned, as a company matures the most relevant metrics will change accordingly as more resources are available and different “engines” are used to drive growth. In the beginning, the goal is to find product-market-fit by testing a hypothesis and pivoting as needed.
I will introduce the 4 key metrics I recommend your start-up begins to measure as soon as possible, and then go into a more detailed discussion for each metric.
- Usage — How often and for how long do users engage with your product in a given time period?
- Churn — What percentage of users stop using your services in a given time period?
- Virality — How likely customers are to invite others and spread the word, and how long it takes them to do so?
- Lifetime value — How much your users are worth from first engagement to them not using your services?
1. Usage:
Being able to track how long, how often, and which features within your application users are interacting with is one of the most important metrics showing overall health of your product. Ideally you would want to track daily usage over time but if your product is not a daily use app, you should establish a minimum baseline engagement for usage. For example, a dentist reservation tool is not expected to be used daily but should be used at least a few times per year.
- Daily, weekly, and/or monthly active users?
- How long it takes someone to become inactive?
- How many inactive users can be reactivated when sent an email?
- Which features did users spend time with most, and which did they ignore?
The top priority is to build a core set of features that gets used regularly and successfully, even by a small group of initial users. If you cannot get 100 users to have a positive experience using your application or accomplish their task, you are definitely not going to be able to do so for a million users. On the other hand, you want to make sure your target market is sufficiently large enough to support growth.
User Segmentation can be very useful in analyzing usage data because it can help lead to a better understanding of users. For example, is there a certain age group or demographic that is using your application more than others? A word of caution, be sure to normalize the data to account for inherent differences in population size. If there is a subsection of users who seem to use your product much more than other users, one needs to figure out what is common to them, refocus on the product requirements, and you can grow your user base from there.
You should also consider the technology adoption lifecycle. New products initially appeal only to early adopters comfortable with change, or to that segment of the market so desperate for your solution that it’s willing to tolerate something that’s still rough around the edges. One must also be cautious that early adopters may be vocal, but their needs might not reflect those of the bigger, more lucrative mainstream market.
When measuring engagement, do not just look at a coarse metric like visit frequency. Look for usage patterns throughout your application. For example, it’s interesting to know that people log in three times per week, but what are they actually doing inside your application? What if they’re only spending a few minutes each time? Is that good or bad? Are there specific features they’re using versus others? Is there one feature that they always use, and are there others they never touch? Did they return of their own accord, or in response to an email?
2. Churn:
Churn is a popular metric that measures the percentage of users who abandon your service over time. The time period will vary depending on the application but you want to measure on a weekly, monthly, quarterly, and yearly basis.
If you have users who pay and not pay, be sure to track churn for both groups separately. Unpaid users churn by cancelling their accounts or not coming back (e.g. a user has not logged back in after account creation), while paid users churn can be measured by cancellations. By capturing the data for customers that have churned, you can invite them back to your product if you have a significant feature update or by offering them a discount code.
It is also recommended that you differentiate between active and non-active users. The definition is specific to your product and industry, but a common definition for non-active users is failure to use your product in the last 90 days.
The calculation for churn rate is as follows:
The reason we want to take the average of the number of users at the beginning of the period and end of period is to account for how churn calculation is a moment-in-time snapshot and therefore can led to misleading results where growth is variable. Below is a sample churn rate calculation for a mock SaaS application:
An example churn calculation for Paying Users in the month of January is as follows:
80 / ((1000 + 970) / 2) = 8.12%
Some alternatives to the above churn rate calculation is to use a daily churn rate, as well 7-day average churn rate to be able to response quicker to user behavioural changes. An alternative approach to churn rates is to measure churn by cohort of users by comparing churn user rates based on when they started using your services. This can be especially helpful when trying to determine the effects on churn by the addition of a new functionality or promotion.
Finding churn rates for your users is just the first step in creating a successful product. The most important work is actually finding out WHY users have churned and how to decrease this rate as much as possible. To help with this, you should be collecting as much data from users once they have initiated the deactivation process. Churned users can provide a wealth of knowledge for what users actually want from a product. For example, if a majority of churned users are complaining that the cost is too high, maybe you should consider decreasing the cost or try other methods of payment like ads.
The best SaaS sites or applications usually have churn ranging from 1.5% to 3% a month. For other sites, it’ll vary depending on how you define “in-active.” Mark MacLeod, Partner at Real Ventures, says that you need to get below a 5% monthly churn rate before you know you’ve got a business that’s ready to scale.
3. Virality:
Virality is a very important method for growth and can be defined as how many additional users your current users bring to your product.
Virality can be sub-divided into three types:
- Inherent virality — built into your product and happens through general use of your product. Example, you send a colleague a link to sign a document using DocuSign, which requires them to make a guest account or download the app. If your product requires your users to connect with other users this can be a great method for growth.
- Artificial virality — built artificially by through some type of reward system that incentives a user to get others to use your product. This can come off as awkward and not the ideal method to achieving high virality.
- Word-of-mouth virality— built by genuine word-of-mouth due to user satisfaction independent of your service. This is extremely effective but very hard to track.
Virality can be measured as a “viral coefficient,” which is the number of new customers that each existing customer is able to successfully convert and can be measured in the below steps:
- Get the total number of users using your product.
- Get the number of invites sent by your users (inherent or artificial).
- Calculate the invitation rate — number of invites sent divided by total number of users.
- Calculate the acceptance rate — number of signups divided by number of invites.
- Multiply the invitation rate by the acceptance rate.
The viral coefficent is a useful metric because in theory 26% of users will in turn invite another 26% of users and so on. In practice, it is unlikely that users will continue inviting friends as time goes on, and instead in the beginning they will invite who they think is relevant and then stop inviting users as time goes on. As such, the number of users gained from virality will get saturated eventually but is still a powerful concept nonetheless.
Ideally, we want to aim for a viral coefficient above 1, which means the product is self-sustaining. It means that every user is inviting at least another user, and that user is inviting another user, and so on.
Another factor that is very important is known as the cycle time. This is the defined as the amount of time before someone invites other users to the product. If for app1 the average time it takes for a user to invite others is 5 days versus the average time for app2 is 1 day, then it means that after 10 days, app2 will have 256 times the number of users compared to app1:
There’s no “typical” virality for start-ups. A sustained viral coefficient of greater than 1 is an extremely strong indicator of growth, and suggests that you should be focusing on stickiness so you can retain those new users as you add them. If you are over 0.75, things are pretty good. Try to build inherent virality into the product, and track it against your business model. But even a lower viral coefficient is useful, because it effectively reduces your customer acquisition cost.
We can improve the virality coefficient metric by a number of ways:
- Increase the acceptance rate
- Extend the amount of time a user is using your product in the hopes of increasing the odds of them inviting other users.
- Shorten the cycle time
- Work on create more artificial virality
You should aware of the risk that you build virality and word of mouth at the expense of engagement. For example, perhaps you are bringing in new users who are different from your earlier adopters, and as a result they don’t engage with the product. Or maybe your unique value proposition is getting lost in your marketing efforts, and your new users have different expectations from earlier ones. If you’re investing in adding users, but your churn is high, you may not be getting a good enough return on investment.
4. Customer Lifetime Value (CLV):
Once you have built momentum with the sticky and virality engines, it is time to consider the paid engine to build a sustainable business. Famously, twitter did not have paid advertising until April 2010 after amassing 40 million users and 65 million “tweets” were posted each day.
Being able to generate significant revenue is a way to really prove that your idea is right and that your start-up can make money in a scalable, consistent, and self-sustaining way. Rather than measuring raw revenue, might be going “up and to the right,” revenue per customer is a better indicator of actual business health. For example, if revenue is going up but revenue per customer is going down, it tells you that you are going to need a lot more customers to continue growing at the same pace. The ratio helps you focus on making real decisions for your start-up.
This is when you should start to consider Customer Lifetime Value (CLV), which measures revenue per customer, which ties in churn rate and revenue.
To find the CLV, we need to first estimate for how long you can expect an average user to use your product. We can the multiply this value with the estimated revenue generated per customer for this time period.
One way to achieve this is to start with the churn rate. For example, the example churn rate for Paying Users in January was:
80 / ((1000 + 970) / 2) = 8.12%
Another way to think about this value is that if we start with 100 users, then each month you can expect around 8 customer to stop being paying customers, so we are left with 92 paying customers. In theory, next month another 8 users will churn, leaving 84 paying customers. If we keep going with this pattern, there will be 0 users left by 12 months.
As such, for the purposes of CLV we can assume that on-average a user will be be a paying customer for:
(100/8.12) = 12.31 months
If the subscription rate for your SaaS application is $9.99 / month, then then CLV is $122.98.
Having a healthy CLV is a great way to demonstrate the health of your SaaS business that combines a number of fundamental metrics into a single metric. One may also consider tweaking the business model or bundle services to improve the CLV. Perhaps you want to consider offering a transactional pricing model instead of subscription-based pricing. You need to test different price points qualitatively (by getting feedback from customers) and quantitatively. While a subscription model lends itself to more predictive financial planning and less volatile revenue numbers, it doesn’t always fit the value proposition, or how customers expect to pay.
A related metrics to CLV is Customer Acquisition Cost (CAC), which is the opposite of CLV in that you want to look at the user enrollment rate (instead of user churn rate) and marketing costs related to user acquisition. By looking at the ratio of CLV and CAC, you can get an even more complete analysis of business health. A CLV/CAC of 5–6 means that for every dollar that a company invests in finding a customer, it makes back $5 to $6 dollars per customer.
4. On the Contrary:
Start-ups are a very complicated business with many things to consider other than data alone. Quantitative data is great for testing hypotheses, but it’s lousy for generating new ones unless combined with human introspection. Rather than be a slave to the data, we should use it as a tool to optimize the business while being fully aware of the bigger picture.
Founders that blindly optimize based on customer data, regardless of its relationship to sales, may have unintended consequences — like bad PR. Data-driven machine optimization, when not moderated by human judgment, can cause problems and can be dangerous — even fatal for a start-up. A machine can find the optimal settings for something, but only within the constraints and problem space of which it’s aware. Data-driven optimization can perform this kind of iterative improvement but cannot inform you how new technological breakthroughs can disrupt how you do business.
In Conclusion:
Running a start-up successfully is one of the hardest things one can do in the 21st century with so many pitfalls and dangers lurking around the corner. Although analytics is not a “magic pill” that will solve all your problems, it can certainly help in providing clarity in many aspects of a company’s operations. It is not easy to architect a well designed analytics system so I would advise hiring a data engineering generalist that has experience with startup metrics or is willing to learn.
Moreover, it’s important to understand that there is a natural progression of metrics that will change over time as the business evolves. The metrics start by tracking questions like “Does anyone care about this at all?” and then get more sophisticated, asking questions like “Can this business actually scale?” As you start to look at more sophisticated metrics, you may realize your business model is fundamentally flawed and unsustainable. This is quite normal in the start-up world and I would advise against starting from scratch. Instead, what you may need is a new market, not a new product, and analytics may help in identifying that market.
Hope you enjoyed reading my article and feel free to reach out to me on LinkedIn to connect and learn more about everything data & technology. Please follow me if you are interested in learning more about the modern data stack and associated tools.
In writing this article I have not gained financially or otherwise. My goal is to spread awareness in the ways analytics can help in the startup world.
