Embedded analytics in B2B SaaS: A comparison

With the inception of many new Software-as-a-Service (SaaS) companies, the focus is shifting from just building a product that works to a product that is fueled by data & AI. Especially for Business-to-Business (B2B), it is increasingly important to have insightful decision-making fueled by real-time data.

This is where embedded analytics emerges as a game-changer. Beyond traditional analytics tools, embedded analytics seamlessly integrates data analysis and visualisation directly into the core applications, empowering SaaS B2B companies with the ability to offer their users a uniquely customised experience. In this blog, I’ll share my experience from the angle of working in a B2B SaaS company (Solvimon) that went through the journey of embedding analytics and explored the different options on the market.

Solvimon’s profile

To give perspective on Solvimon’s profile here’s a few characteristics:

1. Industry: B2B SaaS Fintech
2. Stage: Seed; We just raised our seed round of EUR 9M led by Northzone.
3. Stack: Postgres, BigQuery, Airflow, Spark, Kafka
4. Sources: Invoice, Usage, Customer data sources

A new field in the making

In this blog we’re primarily looking at embedded analytics from the angle of embedding visualisations and dashboards into an already existing application. This also means that the tools that are compared here mostly came from the angle of business intelligence and then built the functionality to facilitate embedded analytics. This is a field that is still evolving quite a bit and there’s new players and propositions coming into the picture every month. The main advantage of opting for this approach to embed analytics is that it doesn’t require a dedicated full stack team to set-up and manage analytics in your application. It furthermore ensures data consistency as you can have one source of truth for analytics (external and internal).

Note that comprehensive comparisons of BI tools already exist, see for example this evaluation matrix from Datateer & Chartio Community.

Criteria

In our case the most relevant technical criteria to focus on are:

1. Allows to embed single visuals and dashboards into an application with multi-tenancy support.
2. Visualisations and dashboards can natively deal with timezone and currency conversions.
3. Visualisations can be customised to fit the application.
4. The embedding is performant in terms of loading and query time.

From a business perspective the main angle that was looked at is pricing.

The compared options

A few options were disregarded from the start due to a hefty price tag, these were Looker, Tableau, Power BI, GoodData. A few options like Trevor.io, Preset, Observable were disregarded as they did not seem to fit our criteria (based on the evaluation matrix).

In the end we chose to compare the following options:

1. Holistics
2. Metabase
3. Cumul.io (now known as Luzmo.io)

Holistics

Holistics is at a core a BI tool that looks and feels similar to Looker. It has quite some functionality and is still delivering quite some new features (such as the dashboards as code). It offers the opportunity to embed visuals and dashboards into applications through an iFrame implementation.

Metabase

Similar to Holistics also Metabase is a BI tool at its core. It however feels nothing like Looker and has a unique feel to it. It felt quite intuitive how its set-up but there’s still quite a steep learning curve, even for a well-seasoned data professional. Also Metabase offers an iFrame implementation for embedding. An added advantage of Metabase is that they are open-source but do offer a cloud offering and paid options.

Cumul.io

This is a different piece of cake. This is one of the few tools that is natively built to support embedded analytics. It offers a no-code builder and also offers SDKs to embed visuals and dashboards into applications.

Evaluation

In short I’ll walk you through the observations I had.

While Holistics seemed like a good option from the start, it already felt a bit “clunky” when interacting with the UI. For example, clicking a button might not do anything (a re-click was necessary) or it did but was not evident as there is no animation. Moreover, the query performance wasn’t great. This made it such that dashboards would that a few seconds to load and that updating filters again took a few seconds.

Metabase on the other hand felt smooth and the set-up was quite straightforward. There was a loading time for the iFrame element to fully show but other than that both performance and customisation were acceptable.

Cumul.io pretty much was in another league here, as it supports better embedding options through pre-build components. In the end, however, we did a further deep-dive in pricing and noticed that Cumul.io limits the amount of viewers. As a B2B SaaS company this isn’t preferred as we might have an arbitary amount of viewer in our application.

All-in-all Metabase was picked as the best option overall. Its performance is acceptable, customisation options sufficient and pricing did not limit the amount of viewers.

Moving away from Metabase

Now the anticlimax, we are moving away from Metabase. While Metabase gave us the head-start we needed to embed analytics into our application and the customisation and performance is acceptable, we noticed that a few things gave us headaches:

• While there are workarounds to support multiple timezones and currencies, this creates a lot of data duplication and does not work seamless.
• The loading time to load a Metabase iFrame in our application does not make it a seamless experience. It takes around 1–3 seconds for the element to fully load after which the query executes. You can imagine, this isn’t a nice experience. Also it never really felt like part of our application, eventhough it had the same coloring and fonts.
• We’d like to offer our embedded analytics directly to customers for them to embed. The iFrame only option and performance issues do not make this an option.

To this end we’ve discussed to throw the above overboard and look for a new option (if existing).

New beginning

To keep it simple we looked for a tool that supports our criteria, does not limit the amount of viewers and can facilitate our embedded analytics use cases.

Two other options came up:

• Embeddable: A spin-off from Trevor.io, is finishing their lighthouse program and is now offering a beta program. It’s build for embedded analytics and promises a truely unique experience.
• Vizzly: A relatively new company also built around embedded analytics. Built from the ground-up to support embedded visuals and dashboards.

Vizzly looked like a good option at first but at the time (it does now in private beta though!) it did not offer a comprehensive cloud solution which meant we’d have a fully manage it ourselves. Embeddable, being a spin-off from Trevor.io, gave us the impression of a more polished product. Also the embeddables would be Javascript native (yay no iframes!) and it supports your own visualisations (to craft those truely bespoke experiences).

I’m happy to say that we’ve enrolled in the beta program of Embeddable. After learning all the above it seems like this is the option we’d want to invest in. We’ll keep you posted on how this pans out, but we’re excited about what Embeddable is building and is going to offer.

EDIT

Together with Embeddable I’ve written a customer story, read it here: Solvimon Customer Story | Embeddable. In short, we are extremely happy with Embeddable and it should be the go-to choice for anyone wanting a fast and fully customisable embedded analytics experience.