Main content

Ex-Twilio team launches a platform to help SaaS companies build analytics products

Barb Mosher Zinck Profile picture for user barb.mosher April 8, 2024
Building on experience, Propel's Data Cloud platform addresses a common problem.

Data scientist and specialist extract knowledge and insights from data. Data science analytics, machine learning control, big data analytics concept © Visual Generation - Shutterstock
(© Visual Generation - Shutterstock) (© Visual Generation - Shutterstock)

Open up most SaaS applications and look at the reporting they offer. It's pretty simple, right? A few dashboards with charts and the ability to apply filters like date range. The experience typically isn't great. Sometimes, you need to see your data differently, and it's just not possible. Creating a reporting and analytics tool inside a SaaS application isn't easy. It's a different set of resources and takes time and money. Or at least it did. 

Three ex-Twilio employees - Nico Acosta, Tyler Wells, and Mark Roberts - have launched Propel, a platform that lets SaaS companies implement analytics capabilities inside their applications faster and easier, tapping into the need for better analytics solutions. The idea is that Propel is helping growth-stage SaaS companies ship analytics products without the need for vast engineering resources and money. 

Identifying the need for better analytics products

Acosta is the CEO of Propel. He's a serial entrepreneur and was employee number 65 at Twilio, joining the company ten years ago. Acosta worked in multiple areas, first on the developer experience, then voice, and then started the AI/ML business unit. 

During his time at Twilio, he saw the need to surface analytics to end customers who were generating vast amounts of data. At the time, most of that data was used internally and accessed via a BI product. It was slow, and the data wasn't real-time, but none of that mattered to the internal team. 

However, the requirements were very different when they wanted to surface that data to end customers. Customers wanted their data to load quickly and in real-time. SaaS companies had customers who not only wanted to consume data via dashboards and consoles, but also wanted to integrate the raw data with their own product or access it via an API to build it into their internal tools. 

At Twilio, these requirements were met by building a custom analytics app in each area. It struck Acosta and his co-founders that there must be a platform (middleware) that isn’t that different from SaaS product to SaaS product or company to company:

Essentially you ingest data, you put it into a real-time analytics DB that can serve data really fast, that can transform it how you need it. And then you serve it via API's that product teams can integrate into their products in the form of dashboards, in the form of customer-facing API's. And that's how Propel was born. From our experience building customer facing analytics at Twilio and kind of our desire to build them faster and cheaper.

That’s where the idea for Propel started.

An industry-wide problem looking for an answer

There are companies out there building their own analytics products, including Twilio. Acosta says the founding team at Propel talked with many companies, and found those doing this well were investing considerable time, engineering resource, and costs. All the other SaaS companies were struggling, creating sub-optimal dashboards that were slow, did not use real-time data, and had poor user experiences. 

Propel was launched with a narrow feature set, with data initially exposed from Snowflake through a set of APIs. Webhooks and then Kafka were later added. Today, Propel supports a wide range of data sources and has a breadth of APIs.

There are several parts to the platform:

  • Connect and ingest data, unifying disparate sources.
  • Transform, enrich, and get the data ready in real time.
  • Expose the data via an API.

It also provides a library of React components pre-integrated into the API to help developers visualize the data in their applications.

All of the above would take a company three to six months to complete on its own. With Propel, the first three things can happen in minutes, said Acosta. A developer can use their Graph QL APIs and build a data experience in hours.

A couple of examples help here. One of Propel's initial customers was Tackle, a cloud GTM that helps SaaS companies sell in cloud marketplaces like AWS and GCP (Google Cloud Platform). Tackle pulls data from all these marketplaces, such as business installations, subscriptions, revenue, and even high-level metrics like views. It uses Propel to surface that data back into its application so that its customers can track how their applications are used.

Another example from Acosta: 

A different company, for example, Lumeo, more in the IoT space, they do computer vision analytics. So they tap into enterprise CCTV streams, and they run a series of computer vision models to do a variety of things, for example, like counting people, or detecting if people coming in or out of an area or counting cars or and they pipe all those events in real-time to propel and then they serve as those real-time dashboards to their customers that are looking at the Loomio product seeing kind of like the flow of people by location by time.

Security, AI, and productizing analytics

The Propel team has extensive experience building operating cloud systems. Co-founder Tyler Wells was Director of Engineering at Twilio and worked for Skype, so they understand what’s required to build an analytics platform that can operate and scale efficiently. The platform is also SOC2 Type 2, GDPR, and CCPA compliant.

One key feature Acosta mentioned was support for multi-tenancy. SaaS applications are multi-tenant, and these companies can have 10, 20, or 100,000 customers. So, access control is essential. Acosta said Propel provides multi-tenant tokens that when used, the API can smartly filter the data for that customer (tenant) and handle all the data access controls.

Then, there is the question of how AI fits in. Acosta says AI is very important for what they do and that they are working on new capabilities that will affect how end users interact with their data, citing two approaches: 

  1. Custom reporting: Customers have different requirements and different units of analysis. AI can help them get their custom reports by asking for things according to the dimensions they care about.
  2. Ad hoc questions: Sometimes, a person needs an answer to a one-off question. They don't need a dashboard or chart; they just need the answer. 

Propel also has an entire semantic layer built on top of the data, making it easier for AI to work. 

Code generation is another important aspect of AI capabilities. Many embedded dashboards aren't great. Companies are embedding Looker and Tableau, but there's no control over the look and feel. The option to do custom builds is out of most companies' reach. But AI will bring the cost of custom builds down. Acosta envisions building a copilot and an API:

You can build with those conversational data experiences, but you as a developer can say, hey, give me give me a React widget with a time series chart broken down by this and that. So here it is, here's the query. Our models now understand your data, they understand the API is to understand everything that you need there."

Propel’s AI capabilities are currently being tested in the lab. A full version will take time, but they intend to get developers' hands on what they are building as soon as possible.

Productizing analytics

Acosta is often asked about how a company should prioritize its roadmap. He believes that the best indicator of customer value is what customers are willing to pay for. 

Twilio always productizes the analytics product it builds. Acosta recalls that these were some of the fastest revenue-growing products in the company’s history. What if you added your analytics product as a new SKU? Have a free tier, he suggests, but think about what high-value customers need, what long tail customers need, and prioritize features:

Then the analytics also have kind of that secondary effect that drives retention. A lot of businesses, like when you make a payment, the value that Stripe gives you goes to zero,  you’re already charged, right? Or you send the message, make a call you, right, like the value goes to zero. Except if you have the collective intelligence of all your activity over a period of time, then you think twice if somebody's offering you a slightly cheaper option, because you're you're losing all that data that you have and all that intelligence. Right. So, like that drives a lot of stickiness with with SaaS products.

My take

Analytics in SaaS applications tend to be an afterthought. Get the main functionality done and then integrate some tools to provide basic reporting and dashboards. It’s a common theme and when you listen to Acosta’s experiences at Twilio, you realize that it doesn’t make sense, especially today.

Analytics are critical to understanding how your product is being used. It’s critical for customers to understand how their business is doing. If ever there was a time to start giving it more respect, it’s now. Propel’s vision and approach can help many companies create robust analytics products alongside their applications. The Propel Data Cloud is now generally available.

A grey colored placeholder image