Lean Analytics: Use Data to Build a Better Startup Faster by Alistair Croll and Benjamin Yoskovitz
Book about how to use data to make decisions in startups. Every metric you collect should be useful information in making a decision, so avoid collecting vanity metrics like the number of page views or sign-ups. It’s easy to lose information by simply averaging, so look for segments of users or behave as a certain way, or cohorts of users who signed up at a certain date and track them across time. A good business direction should have three key properties: it must be profitable, you should like doing it, and you should have some kind of advantage over others.
Too many metrics is distracting, narrow them down to a few that you care about; different metrics are good for tracking different dimensions of your business, such as how sticky are your users, how much money they spend on your product, how likely they are to pay, etc. At any given stage in your company, you should define one single metric to focus on, this will give you focus on what matters. Which metric to choose depends a lot on your business model and which stage your company is in:
- E-commerce businesses sell a product to the customer, they can be divided into two groups: those that are focused on one-time buyers, and those based on loyalty, this mostly depends on which category of products you sell. If customers are one-time buyers, you should focus on acquisition, so consider referral programs and creating awareness. If your business is loyalty based, then the focus is on making customers buy more frequently. Most people will arrive through search, so keyword optimization is important. Track the abandonment rate at each step of the checkout process; also operational metrics like shipping rate and stock.
- SaaS (Software as a Service) usually operates as a subscription model (although exceptions can be made), your main goal is to build something sufficiently “sticky” so that LTV can recover the marketing cost, so churn is a crucial metric. Freemium is a useful sales tactic to get people to eventually pay.
- Mobile apps: most of the traffic will come from the app store instead of traditional forms of marketing, so app store dynamics is highly important. Otherwise, the churn metrics to track are similar to SaaS businesses.
- Media sites (like news and blogs): make money by creating content that people want to read, while monetizing them with display ads and affiliate links. The challenge is maintaining a balance of content to ad ratio to keep users, while maximizing inventory (display ads available to sell).
- User generated content: platforms like Reddit or Facebook, where users generate content that you monetize. Define a hierarchy of engagement levels for users from casual, logged in, voting, commenting, posting, and finally moderators – look at how the users go between these categories (engagement funnel). Expect to spend a significant amount of time preventing vote manipulation and spam.
- Two-sided marketplace: kind of a combination of e-commerce and user-generated content business models. When starting a two-sided marketplace, focus on the demand first, ie, the side spending the money. You can artificially acquire supply at first; after it is passed the initial stage, look at the health of the market: percentage of repeat buyers, percentage of searches with results, fraud and quality metrics, etc.
Your startup should have different priorities depending on which stage you’re at. First, make sure the product is useful enough before worrying about revenue and growth. In the first stage, validate that problem really exists and is painful enough to be worth solving. Talk to customers and rate their responses on a quantitative scale, you can do interviews or surveys, but with surveys be careful that users understand them properly. Validate the riskiest assumptions of the MVP first before building it, and have analytics installed to measure the engagement for every step of the MVP.
The second step is stickiness, here you focus on retention and engagement. Analyze what steps makes users drop off during sign up, or when they are confused by a feature which means that the marketing is misaligned with the feature. The goal at each iteration is to learn: come up with top questions and design experiments to find answers to them. The next step is viral growth, this is useful because it is cheaper than paid marketing. The ideal case is if virality is inherent part of the product, such as a travel app for sharing plans with friends; otherwise, artificial virality is deliberately adding incentives for referring friends to the product.
In the revenue stage, you want to build a machine that takes in pennies and produces nickels, ie, produces more money than it takes in. Oftentimes you can have an impressive number of users and customers, but are not making money from them: in this case, the best strategy is often to pivot to a different market for your existing product who are willing to pay, such as selling to publishers instead of readers. In this scaling stage, you have saturated one market and are trying to expand to more markets and channels. This will often require making different versions of your product suitable for the new markets.
The next section of the book presents typical rates to target for different types of business models, such as churn rate, growth acquisition cost, page load time, etc. For new entrepreneurs, it is often unclear which metrics you should be improving, but with these rules of thumb, there are certain thresholds that, once passed, are probably good enough and your time is better spent working on something else instead (eg: cancellation rate per month in the best-case scenario will never be less than 2%). I won’t list all the threshold numbers (there are a lot of them) but they seem approximately correct and serves as a useful reference. When you have no idea what threshold to target, try improving it for a while until the improvement plateaus, then switch to something else.
The final two chapters discuss enterprise products and ‘intrapreneurs’ within larger companies. The difference with enterprise sales is that there are fewer customers, so you can afford to talk to all of them, but you only have one chance to make a first impression. Be careful not to provide services without building up a product, as many startups that follow this route basically become consulting companies. Be willing to fire customers that take up too much time, or charge them enough to be worth the effort. In big organizations, entrepreneurs can exist within an existing organization, but they need to get executive buy-in for their proposals and then iterate. Similar to a startup, you have the advantage of the resources of a larger organization, but there are also more constraints, such as not alienating existing users.