So you’ve set up Segment (maybe on WordPress using our how-to guide?). Data is flowing from at least one source such as your website or app. It works. You see the data flowing. That’s cool!
But what now? You’ve got a large stream of data coming in, and it needs attention. How do you even use Segment effectively and go beyond just installing Segment?
Where Does Segment Fit In Your Product Analytics Infrastructure and Strategy
We should first start by understanding what’s the purpose of Segment. Where does it fit in your architecture but also in your overall analytical strategy?
Segment In Your Data Architecture
In regards to where Segment fit architecturally, here’s how we view the ideal data infrastructure at Lantrns Analytics (which we go into greater depth in our Ultimate Guide to Product Analytics).
Segment is the broker of your product’s data. It mines data from your touchpoint’s sources (website, app, CRM, helpdesks, surveys, payments, etc.), transforms your data in a unified format, associates it to a uniquely identified user and sends that data to an extensive list of destinations.
Segment In Your Analytics Strategy
But it’s not enough to understand how Segment fits in your data architecture, you want to leverage Segment for your analytical strategy. So let’s use an example.
Say your product is a 2-sided marketplace that connects influencers and brands in order for the latter to market their product through social media posts from the former. This sort of ecosystem is rich in dynamics and there are quite a lot of outcomes that could/should be measured.
But only a few metrics are really key to measure the health of a 2-sided marketplace. It’s mostly all about the number and quality of members in each party (brand and influencers), as well as the activity surrounding the deals you’re brokering between those 2.
Surrounding all that are a lot of user behaviours that will provide indicators as to what those numbers will look like in the end. Think of registrations, frequency of logins, deals posted, applications to deals, etc, etc.
Those are the events you’re tracking through Segment that map out the user journeys and give context to your product’s KPIs.
Consolidate Your Investment
I’ve seen Segment customers invest a substantial amount of money per month to track user behaviours through that service without any real plan as to what their objectives are and how to make that investment worthwhile.
Use Segment with A Purpose
As Spiderman is well aware: “With great power comes great responsibility.” Segment is indeed quite powerful and easy to use. You could easily overindulge and end up with a clutter of meaningless data. This is where I’ve seen good intentions, excitement and ambition crash against a wall of useless data.
You need a strategy to link data that is to be tracked to your product’s KPIs and user’s journey key touchpoints. As we’ve hopefully overemphasized in our guide, not all dynamics are created equal, meaning not all outcomes are worth tracking.
I won’t go over what’s already been described in-depth in our Ultimate Guide to Product Analytics, but there are choices to be made based on the growth stage of your product, the business model, users’ struggle(s) you’re aiming to solve, current challenges you’re facing, strategic initiatives you’re currently pursuing, etc. There are many questions that should guide the development of your analytical strategy and infrastructure.
That whole strategy will lead to a tracking plan that links your strategy and implementation. Segment has a good article on tracking plans that you should definitely read before even thinking of tracking a single event with their library. I’ll also come back to tracking plan a bit further in this post to talk about enforcing its implementation and making sure that it’s always relevant and accurate to link your analytical strategy with your data infrastructure.
Invest In Your Data Warehouse
Using Segment is the first step in owning your data and that should be a central concern of yours. Of course, you could just use Segment to pipe your data to Google Analytics, but what’s the purpose of that? You are paying for a service that generates granular and high-quality data that you can own. Take advantage of that.
As the first destination of your data, you should absolutely set up a data warehouse on any of the platforms you’re comfortable with. I highly recommend Snowflake (for its separation of storage and compute, for its ease of use and for the richness of its SQL language and tools), but you could definitely use other analytical databases such as Redshift, Google BigQuery, or even transactional databases such as Postgresql.
Whatever you choose, just set up that data warehouse now and you’ll thank me after.
That should be the first step, really. You don’t even need to do anything with that data yet, but just have it there and rejoice with the fact that you’re now owning your product’s usage data.
But what can you do with that data?
This is to be the source of all your data exploration. Seeing a weird value in your KPIs? Explore the underlying data in your data warehouse! Wondering what all influencers from Sweden have been doing on your platform this last Sunday? Build custom KPI reports with Tableau / PowerBI / Looker / etc. to create an ad-hoc analysis. You can even eventually train your predictive models. The possibilities are endless once you own that data.
Setting up a working data warehouse requires raw data first and you’ve done that by setting up your DW as the first destination in Segment. Now you need an ELT process to clean up / transform / enhance that data to make it easily usable by yourself and other analysts.
Take advantage of SaaS service’s free packages
Now that you’ve done the heavy lifting by defining your strategy, writing and maintaining a tracking plan, and setting up your data warehouse, it’s time to explore the endless possibilities that Segment offers. Just have a look at Segment’s catalog of services.
Besides services for Analytics, you can connect your Segment sources to destination categories such as A/B Testing, Advertising, CRM, Email Marketing, Heatmaps, Live chats, Push notifications, etc. Many of these options have free plans to evaluate.
I would recommend not going trigger happy and firing off multiple services right from the start, but think about what you’d like to better understand now and find the service that could help you explore that question. Chances are there’s a free plan out there that can be set up in minutes that will help you answer those questions.
Be A Power User
Now that you’ve consolidated your investment in Segment, time to push the envelope and become a power user. Segment is a facilitator in your product analytics’ journey and now that you have a solid data infrastructure in place, there is no reason to not reach for one of those few holy grails.
Constantly accelerate your development cycles
Product management is a speed game and if you’re into the Lean philosophy, you know that accelerating your build-measure-learn cycle is core to reaching product-market fit fast (or pivoting faster).
Segment is here to help with this as well. Protocols is an add-on to their core offering.
Protocols’ mission is to:
- Align everyone involved with your analytics on what’s to be tracked and how;
- Validate that what is being tracked has no quality issues;
- Enforce that destinations are only using the data they need and in the format required.
Even if you have a solid Segment implementation, it’s easy to lack the flexibility to adapt your analytics infrastructure to changing needs and for errors to creep in.
For example, let’s say our 2-sided marketplace enters a new phase of its growth and we are now more interested in community building aspects of the marketplace, then we will focus on different user events and KPIs. But how to easily make those changes throughout the infrastructure? It would be easy to just add an event without going through the implementation strategy and tracking plan. If we do this, we might not enforce our naming convention for events (for example Object-Action, such as User Login). All kinds of problems can creep in with time.
Protocols are an integrated line of defence. No events can be implemented without being added to its tracking plan. That takes care of enforcing naming convention, making sure we define properties to it and assessing that its implementation follows all our requirements.
High-quality data is key to accelerating your development cycle. You can’t measure and learn with a dysfunctioning data infrastructure, which is hard to change and for which you cannot trust the data.
Deploy improvements through experiments and measure impact
That brings us to a second holy grail: deploy improvements through experiments.
With your whole product instrumentation in place, you have the infrastructure to deploy improvements that are anchored on a hypothesis of how that should impact your user’s behaviours and product’s outcomes. Having baselines of users’ metrics such as posts read in a session, all product improvements should be an educated guess as to what can be done to influence behaviours that have a positive impact on their experience and your outcomes.
Experimentation isn’t as complicated as we tend to think it is. There are statistics involved, but it’s not that complex and shouldn’t prevent product owners from plunging into experimentation. And even if your methodology ain’t perfect, it will at least be more robust than the Hail Mary approach.
That said, Google also has a pretty solid A/B testing platform available called Google Optimize. You can easily manipulate your website’s appearance and content to evaluate how users react to those changes.
Or how about just doing it yourself without the help of any tools? I personally use Asana to track experiments and document their results. Nothing fancy about it, but it works and as I said, it’s still better than the Hail Mary approach.
Whatever your method, experimentation is a holy grail you shouldn’t look past because you might be scared off by a few statistics. The advantages far surpass the little amount of learning you’ll need to do.
Personalize product to your users
One holy grail that is way easier to get access to now is personalization. Not only is there a solid catalog of Personalization services offered by Segment, but they themselves are offering another add-on which is called Personas.
Let’s start from an example. A visitor comes to your site, signs up for a trial, eventually subscribes, uses the app further and contacts your support service. Those are just some of the actions done by one visitor and for which data is generated and captured by different systems (Website, Payment system, CRM, etc.).
So let’s say our user signed up for a free trial, contacted your support service, didn’t subscribe and became inactive for the last 30 days. Personas can help you target those users and launch campaigns to reactivate them.
Personas aggregates data from all your sources under a unified identity, lets you build profiles and synchronizes that information to your marketing channels.
As behaviours are coming in real-time to Segment, new traits are automatically computed and attached to your users. This means you can define audiences right from the source (e.g. trial users who contacted support and haven’t subscribed) and sync those to your marketing channels. No need to configure that audience on all channels.
Now that you’ve defined an audience, you can automatically sync it with marketing channels such as Facebook Custom Audiences, Salesforce Marketing Cloud, Twitter Ads, etc. That means you do not need to define that audience on all of your marketing channels – Segment Personas takes care of all that for you.
You’ve just taken the first step towards unleashing a flurry of insights that will enlighten the path towards product-market fit. Our hope is that this article will inspire you to take you towards product analytic’s holy grails.
Lantrns Analytics is dedicated to accompanying you on that product analytics journey, may that be by sharing knowledge and educating you on how to do it on your own, or by giving you access to your own team of dedicated analytics experts.
Want to pursue that journey with us?
- Subscribe to our weekly newsletter. We share data analytics’ best practices and new ideas with product owners, so they can incrementally and intelligently improve their product.
- Read our Ultimate Guide to Product Analytics. This is where we share our up-to-date ideas on how to do product analytics right.
- Have a look at our packages. This is our entry-level service package where we get your analytical infrastructure in place + all that’s needed to start reporting your product’s KPIs effectively.
- Our just contact us directly. We’d be happy to chat.