Hey friends,
From 2017 until around the start of Covid, I made my living helping early-stage SaaS companies set up their product analytics and reporting.
I’ve helped many startups “turn on the lights” to make better business decisions and in this post I’m going to help you do the same.
In this extensive guide I’m going to cover the following topics:
The 5 steps of the product analytics implementation plan.
The frameworks you need to understand to shape your product analytics strategy.
Best practices for building and maintaining a tracking plan.
How to scale your product analytics events, and tie it to your product development process.
Introduce you to PostHog, the king of product analytics.
How to setup and use PostHog.
The difference between client-side and backend events.
A list of the first events you need to start tracking.
A list of important product questions, and how to answer them.
By the time you’ve finished reading this guide you’ll know exactly how to gain visibility into your own SaaS product, the important questions you should be asking, and how to answer them.
You’ll know if you’re moving towards product-market fit, where users are getting “stuck”, and how many are experiencing your “wow” moment.
We have a lot to cover so let’s get started.
The 5 steps of the product analytics implementation plan
Before jump into the “meat and potatoes” of this post, I think it’s a good idea to have a high level understanding of what we want to achieve, and the plan to get us there.
Our primary goal is to set up the necessary tooling, infrastructure, frameworks and processes to allow the key decision makers to answer business critical product questions.
To do that we will follow the product analytics implementation plan shown in the image below.
The plan has 5 main steps which will be covered in detail in this post.
Before we move onto step #1, building a tracking plan, I first want to cover 2 important frameworks.
2 important frameworks to help shape your product analytics strategy
While working as an analytics consultant, I developed two useful frameworks that are important to know before discussing the tracking plan.
Framework #1: The core funnel
If you’re a long-time reader of this substack then you should be familiar with the core funnel framework.
The 12 minute video below covers the framework in detail.
The main idea behind the core funnel is to map out the major steps in your users journey. Most SaaS typically have 3 - 6 major steps which make up the core funnel.
The reason it’s important to understand the core funnel is because it gives us a starting point for our event tracking plan, and forces us to stay focused on what matters. We first want to make sure we can track the core funnel before we focus on other areas.
A SaaS that doesn’t have visibility into how users progress through the core funnel are “flying blind”.
Framework #2: Include product analytics in your product development process
The second framework is to integrate product analytics in your product development process.
Product managers need to be trained on product analytics basics and include a section in their product requirement documents (PRDs) that cover product analytics. Some businesses hire product analysts who are responsible for covering this area instead of the product managers.
The kind of data I like to see included by product managers are:
A list of questions which the product analytics need to answer. For example, “how often is this feature used to users?”, and “can we track when this feature is enabled or disabled?”.
The actual events themselves so that the developer who builds the feature can make sure they are covered before the feature goes live in production.
In a few minutes of reading, you’ll have the knowledge you need to implement framework #2 in your startup. The main takeaway is that once you’ve decided to start adopting product analytics, it needs to become a part of your product development process.
We’re now ready to move onto the first step of the product analytics implementation plan, building a tracking plan.
Best practices for building and maintaining a tracking plan
A tracking plan is a company-wide resource to help everyone stay aligned on the events that R&D have, or will implement. It’s a “data dictionary” of sorts that help keep things organized.
A good tracking plan has an owner, usually the head of product, and is consistently maintained. A tracking plan has its own set of rules which should be written down in the internal knowledge base, and shared among the product and R&D team.
I recommend building your tracking plan as a table, either in Google Sheets, or an equivalent solution.
I created an example tracking plan which you can view here. Feel free to steal it: https://docs.google.com/spreadsheets/d/1JQ3WlxyCwYFeHi2TjS_aTizm_vSHkaQS8lnI2leM18A/edit?usp=sharing
I only filled out all the details of the first 4 events, but you have enough to get a good idea of what a tracking plan should look like.
Let’s go over the columns I’ve included in my version.
# - The “id” of the event in the document. It also acts as a counter so you can see how many events are in the document. Having a number for each event is also helpful for referencing events in the document (“check out event #4 in the tracking plan”).
Event name - Standard. Each event needs a name. This name is what will appear in PostHog.
Date Added - Just good to have. Helps see how the event list grows over time.
Added By - The individual who added the event to the plan. Helps involve the product managers, analysts, etc. Helps everyone understand ownership.
Status - You should have different statuses for each step in the funnel of taking an event from concept to live in production. This way it’s easy to see where the event is in the process. Use statuses which make sense to you based on your own process. In my example I have “Work In Progress”, “Ready for Dev”, “Ready for QA”, “Live” and “Canceled”.
Default properties - This column is optional but it does help having it to remove confusion for QA. PostHog has certain default properties which are out-of-the-box and can’t be manipulated. You can list those in the tracking plan so QA knows to ignore them.
Custom properties - This is by far the most complex field in the tracking plan. In this column you need to list all the properties you want to include in the event. Here are a few helpful guidelines to know for assigning properties:
Include ids - This helps analysts (and also agentic agents) run joins and connect events to each other, or other entities in your database.
It’s fine to include ids and their values as separate properties - Notice I included both “source_id” and “source” as properties for the Request Created event. Having both allows for quick understanding in the PostHog interface, as well as allowing for a join.
Reuse properties to keep the distinct list as small as possible - For example, don’t have two properties, “button_color” and “color”, just use “color”. Use common sense, and try and keep the names of properties “high level” so they cover a wide range of use cases. Use simple, descriptive words to avoid confusion.
When you need more than one property to distinguish between possible actions, use multiple properties, don’t create a unique property - Lets say you have multiple red buttons on a page and want to track which button exactly is being clicked. How would you do it? In such a situation I would have an event called “Element Clicked”, and 3 properties, type = Button, color = Red, and location = top right corner. Notice I have 2 descriptive properties to help me distinguish between the actions. I would map each button’s location clearly in the PRD when these events are created and create a button click event for each button.
Be consistent - This includes maintaining the naming convention rules, using the same relevant values across multiple events, etc.
Trigger - The trigger column is used to explain clearly the trigger logic for each event. Make sure it’s clear when the event should be triggered and cover nuances and rare situations so the developer creating the event can cover all scenarios.
Task URL / Story URL - Depending which product development tool you use (usually Linear or Jira, but sometimes Notion, Trello, or some other solution), you would have a url to a document, table entry or “story” related to the event. Whoever is working on the product update that is related to the event should include the url to the document in the tracking plan. This allows others to click through and see the full context behind the event. We only need to include the link to the document or story for the first time the event is created.
Notes - A general field for notes by the person who added it, or the head of product. It could be used for starting a comment thread where a discussion on the event might be needed. This field is very much optional.
Now that you’ve got the outline of your tracking plan ready, which events should you add first?
A list of the first events you need to start tracking
Each SaaS will have it’s own initial set of events so I can’t give you an exact list of events to add to your plan. What I can do is give you some guidelines to help you come up with your initial list.
Limit your initial list to no more than 10 events, and focus on the core funnel
As the headline above says, start with no more than 10 events.
That usually involves 3 - 6 core funnel related events, and a few others related to primary usage / value exchange.
For my SaaS, Project Echo, a user starts his journey by signing up, going through a short onboarding process and then landing on the request page.
During the onboarding the user picks a few requests to help seed his workspace but the real “wow” moment is when one of his users submits a request.
To help map out this process using events, I need the following events:
User Created - The first step. It’s triggered when a user either signs up using email, or their Google account. Since one user can have multiple workspaces, I need an event that specifically indicates a new user signed up.
Onboarding Completed - Triggered once a user has completed the onboarding process. At this stage the user is ready to see value from the SaaS.
Member Created - A member is an end-user of a Project Echo user. A member is either created manually, or automatically if the user integrates their SaaS with Project Echo. A member needs to be created before that individual can submit a request (the “wow” moment), so we need to track this event.
Request Created - The first real value exchange is when a user creates a request in their workspace. This can be done manually by the user, via an agent they have connected to their workspace, or via a script leveraging the API. Either way, when a user creates a request its a good sign and we want to track it. This event is also triggered when a member creates a request (the “wow” moment). We would use specific properties such as created_by_id, to distinguish if this request was created by a user, or a member.
So in total I just need 4 events to map out the most important part of Project Echo’s core funnel. I could add a few more such as “Subscription Started”, or “Invitation Sent” (for when a user invites their team to their workspace), but at the end of the day I need to first nail the value exchange part of the funnel before I focus on anything else.
Include events which help you understand usage patterns
I recommend including events which tie to direct, primary usage of your SaaS. In my case it’s Request Created but in your case you might have multiple features you expect users to use repeatedly.
The reason it’s good to have events for each feature use is because it can help you identify usage patterns that could change your product roadmap.
You might expect users to do A but you see in the data they do A once and then come every day to do B. Now you can focus on trying to understand what is special about B.
Usage patterns are really useful to know early because it can help you pivot in the right direction, rather than guessing.
Well done, you’ve built a tracking plan, decided on your initial list of events, and you’re ready to implement them, but which tool do we use to collect and house these events?
The king of product analytics: Introducing PostHog
PostHog is an all-in-one product analytics platform that was founded in 2020. Over the last few years the company has grown into a “unicorn”, and become extremely popular among early-stage SaaS startups.
The two main reasons PostHog has become so popular is:
It’s multi-tool approach.
Its extremely generous free tier.
A single PostHog account gives you access to event tracking infrastructure, session recordings, surveys, “feature flags”, a data warehouse, data pipelines and more.
I could write an entire guide on using PostHog but since our focus is on event tracking, I’m going to focus primarily on this part of the PostHog platform.
PostHog’s free tier includes just one project (one project per SaaS), but you can move to a paid tier, enter your credit card, and increase this limit to 6.
You still get the free events and other credits so even though you’re no longer on a free tier, you won’t pay anything unless you exhaust the free credits.
How to set up PostHog in your SaaS app?
The initial setup of PostHog takes just a few minutes. You signup using an email address, Google, Github or Gitlab account, go through a short onboarding flow, and then you’ll get access to your project details.
The next step is adding PostHog to your app. This can be done with a prompt to your coding agent since PostHog’s SDKs and documents are public. What you will need is your project’s token, region and project id.
Your coding agent will instruct you to add these details as environment variables (this is best practice), and then it will go ahead and add PostHog’s relevant SDK to your app.
It’s important to understand that PostHog, just like Segment, Rudderstack, and other event tracking tools, fundamentally covers two types of events.
The first type is client-side, or “front-end” events. These are events which are generated when users visit pages within your app, or engage with elements on the page. Clicking a button, filling out a form, etc would be considered client-side events.
The second type are events triggered in the “backend” of your app. An example might be “Subscription Canceled” which may be triggered without the user doing anything, it just happens in the background.
PostHog is used to collect both of these event types but the initial setup focuses on the first type.
PostHog provides a lot of client-side events out of the box, including “pageview” which is very important.
If you leave these default events on (PostHog asks you during the onboarding), you’ll end up capturing a ton of useful events without needing to code them individually. I recommend keeping these events on and to make sure PostHog’s SDK is loading on every page within your app.
You can use PostHog to track activity on your public website but just keep in mind that if you have a lot of traffic you are more likely to hit your monthly limit. If you’re just getting started, go ahead and add PostHog to all your pages.
Once PostHog has been added, make sure to test it by logging into your app, or visit your website, and confirm that PostHog is receiving events. This can be done by visiting the Activity page in PostHog.
Congratulations, at this point you’re collecting useful client-side events, including pageviews. You can use this information to see how users are navigating within your app, which pages they visit the most, and which pages get ignored.
The next step in the process is to start implementing the initial set of events from your tracking plan.
If you’re using Cursor, or Claude, write up a detailed prompt sharing the details of each event so the agent knows what to do. Make sure to include the trigger, properties, the relevant source of each value for each property, and the naming convention rules.
Make sure to test each event once the coding work is done and make sure the correct values are being passed to PostHog.
The sooner these events are being tracked, the sooner you can start gathering data. The more data you have, the more statistically significant your analyses will be so don’t wait. Prioritize the implementation of your initial events and push the changes to production ASAP.
Once the events are live, you may need to wait a few weeks to collect enough data to move onto the next step in the process, answer the product questions.
If you’re fortunate enough to have enough new signups and regular usage to collect a lot of data, then you can move onto the next step after a few days, or a week or two. You may still be forced to wait a while if users typically take a long time to go through the entire user journey.
The most important product questions for early-stage SaaS, and how to answer them
Everything is in place for you to start answering your most important product questions but you aren’t sure where to start.
Below is a list of questions you should focus on answering first. These questions will help determine if the product is doing what its supposed to do, if you are moving towards product-market fit, and most importantly, where are the main issues that need to be addressed.
I want you to notice that the questions are sequential, following the shape of the core funnel.
How many users signed up in the last 7 days / 28 days / 90 days?
The first question deals with the top of the funnel. We need to know how many users are signing up so we can determine if we have enough “throughput” to make decisions. We need data to make data-driven decisions and if we only get a handful of users signing up, we don’t have enough to make statistically relevant decisions.
This question can be answered by either counting the “signup” event and grouping by day, or by running a query on the database (counting the distinct user_id and grouping by date).
If you have very little signups at this stage, then forget about the rest of these questions and focus on growing the top of the funnel. You shouldn’t bother moving beyond this initial step if you don’t have the data you need to answer follow on questions.Out of the users that sign up, what percentage complete the onboarding?
The first step in the funnel after signup is completing the onboarding. We expect the vast majority of our users to complete the onboarding since those that don’t will never convert to a paid subscriber.
Since we are tracking our core funnel, we need to have a way to determine if a user completes the onboarding.
This can be done either via a simple pageview event (looking for pageviews of the page users reach after the onboarding is completed), or by having a dedicated “onboarding completed” event.Out of the users that completed the onboarding, what percentage of them experience the “wow” moment?
After onboarding we expect our users to quickly get value from our software. This is typically referred to as the “wow” moment. We should have a dedicated event for this moment, either in the form of a usage event (“the first time the user uses the feature they experience the wow moment”), or a serious of events (“uses the feature 3 times and then adds more credits”).
To answer this question we first need to have a clearly defined “wow” moment and make sure we are tracking the necessary events. Once the logic is determined and the events are in place, we can run our analysis.
We also want to start using cohort-based tables and reporting to answer these questions. Since we expect a drop off of users during the onboarding, we only want to include those that completed the onboarding in our analysis.Out of the users that completed the onboarding, and experienced the wow moment, what percentage of them had repeated usage?
As you have undoubtedly noticed by now, the questions I recommend you focus on should be cohort-based, and funnel orientated.
You ultimately want to know on average how 100 users go through their journey and what percentage get through to each step in that journey.
The typical paying users will have completed onboarding, experienced the “wow” moment, had repeated usage, and ultimately wanted more value which required moving from the free plan to a paid plan.
The video below shows a core funnel dashboard which shows how different cohorts perform throughout the core funnel. The table in the video doesn’t have “repeated usage” but that is a column you could add to your own report.The bottom line is that each core funnel event we added to the tracking plan needs to map to a step in the journey thus allowing us to see which users reached (or didn’t reach that step).
Once all the necessary events are in place, we have what we need to build a core funnel report like the one shown in the video above.The core funnel report shown above is your end goal. This single report is all you need to understand where the problems exist in your SaaS business.
Notice how the report shows both absolute and percentage drop off. We need the absolute numbers to see if we have enough data to take the percentage drop off seriously. If we only had 10 users in a cohort, we couldn’t determine the true conversion between the steps.
The way to analyze this report is to focus in on the big drop offs. For example, in the table above over 90% complete the onboarding but only around 60% of those users experience the “wow” moment. That big drop off is then followed by another, repeat use, and we then end up with a conversion to paid of about 10 - 13%.
We would focus in on the three steps in the middle of the funnel (Wow moment —> repeat use —> conversion to paid) since this is where the users are dropping off.Out of the users that completed the onboarding, experienced the wow moment, and had repeat usage, what percentage of them subscribed to a paid plan?
The final step in the core funnel that we care about at this point is the percentage of users that convert to paid. Now some SaaS don’t have a freemium plan so their core funnel will be different and they may want to look at the percentage of users that start a trial, enter their credit card, etc.
In my example we do have a freemium plan and therefore need to first track adoption, usage and ultimately the percentage that convert.
The approach to answer this question is the same as the previous, we need a dedicated event tied to this step in the journey (would usually be “Subscription Created”, or some variant), which we can tie to our users and add to a cohort table.
Since most SaaS use Stripe, it is possible to get events directly from Stripe instead of recreating them in PostHog. There are pros and cons for each approach but I prefer to have all my events in one system so I’d recommend creating them in PostHog.Out of the users that started a subscription, what percentage of them are still subscribed 1 month, 2 months, 6 months, 9 months and 12 months later?
The final question in my list has to do with retention, specifically subscriber retention.
This is the final step in the core funnel. We want to know what percentage of those that start paying us are still paying us as time goes by. I like to focus on monthly subscribers and their retention over the first 12 months.
What I have noticed is that if monthly subscribers generally stick around for more than 6 months, the SaaS is in a healthy position. The reason I focus on monthly subscribers is because you need years of data to determine true retention of annual subscribers. I’ve found that if monthly retention is high, annual retention is usually also high.
There are a few ways to answer this question. The most accurate requires getting your hands on the charges data from your payment processor so you can see monthly payments from your subscribers and build a useful dataset for analysis.
This question wouldn’t typically be answered using PostHog but more likely directly from the subscription management solution (Stripe in most cases). Tools like Stripe even provide out-of-the-box reports to help answer questions like the one above.
In Summary
This guide has walked you through the exact process I’ve used to help many SaaS startups master their product analytics. We covered the entire product analytics implementation plan.
The process started with understanding important frameworks, building a tracking plan and signing up to PostHog.
You then implemented that tracking plan and started collecting data.
Once enough data was collected, you moved onto step #4, answering select product questions.
I gave you 6 questions which helped you build a core funnel cohort report which paints a clear picture on how users progress on their journey from signup to paid subscriber and beyond.
This report is the most powerful report you can build as an early-stage, pre-product market fit SaaS. It not only clearly paints a high-level overview of your core funnel, but points you directly to where you and your team need to focus.
If you have any questions on this guide that you’d like answered, please don’t hesitate to ask in a comment below.
That’s it for this week, thanks for reading.
Justin
PS: If you need further help setting up PostHog, building a tracking plan, or answer key product questions, please reach out, I’d love to chat and see how I can help.










