Embedded for the win.

Kris Curtis
CMD'ing Data
Published in
4 min readSep 10, 2020

What is embedded analytics?

It’s a term that has been around for a few years, but fairly recently emerged as an “added feature” in some of the software vendors around. You could argue that players like Microstrategy have been always had this as part of their offering.

I’ve worked on an embedded analytics project before, taking the form of charts or visuals and embedding them in another application, normally into a custom website or similar interface.

“Taking the form of charts or visuals and embedding them in another application, normally into a custom website or similar interface”.

Hosted website with proposed layout of embedded charts in iframe.

Why do this? Surely the native gui are perfectly set up for users already? Software companies invest millions of R&D budgets into creating these user friendly interfaces.

In that first experience of working on this type of project I didn’t get it.

I really struggled to see the need or the value in doing this. Other than the stakeholders wanting it, there was no real other reason and eventually the project slowly died.

Fast forward to this project.

A new client joined and there was an agreement to provide a reporting portal in which these customers could view their performance.

We were already providing this service to other customers using ThoughtSpot as the analytics platform in which to deliver data to our external customers. Why did we need to create a different solution for these customers?

So fairly reasonable question right?

The answers justified this.

  1. The customer was of high strategic importance. We had to go the extra mile to show the client that we valued them as a customer and needed the “wow” factor
  2. Our competitors were also providing a custom reporting portal for the client, meaning there was already an expectation that we would also fulfil this requirement
  3. Our standard performance reporting contained lots of metrics and kpi’s, some of which were not relevant to this client. We wanted to avoid confusion from these and focus purely on the metrics that mattered to them

The brief

So we have a high profile client that we want to impress. Who already has exposure to a similar type of insight portal from a competitor. We need to surface the right metrics and insight to them to help them understand their data and grow their business. We had to provide insight to two user types, different from each other, but in order to be efficient, deliver this solution in the same environment.

When you read that, its summerises for me the exact reasons why you would want to explore an embedded solution for analytics.

The benefits

In working on these type of projects there are always benefits. I’ll just make a note of some of these right here.

  • Fully customisable user experience — incorporating small but important details which really show the end users you care about them. Branding and the company logo help create this trust and bespoke feel
  • Additional features—links to help pages or emails and feedback links incorporated on the page which users can easily use. Feedback call to action buttons and other bespoke features.
  • Removing the clutter and distractions — depending on your company policy and governance users may find themselves lost in content overload. Outdated reports, multiple metrics with different definitions. An embedded solution can ensure users only see what you want them to see.

Learnings & Advice

We were able to design and build our solution in around 6 weeks with a very small team. The benefit that comes with data visualization software being used as the basis for an embedded solution is how fast you can deploy.

However- we hit a few issues. Here is my advice on anyone looking to start a similar project.

  1. Get the team on the same page — we had plenty of people who had not worked together before. Introductions and working styles were overlooked. This meant there was a different level of buy in at day 1.
  2. Identify a leader — we didn’t have one clear person who was leading the project. In our case it was a commercial proposition, being developed by data, wrapped up in a tech product.
  3. Stop the creep — Working for an external client, there were lots of demo’s and meetings. Great excitement from the client, but this also got the multiple parties more than willing to help out.

Summary

My view point on embedded analytics has changed after this project. I’ve seen the light and am a convert.

I wouldn’t rush into a project like this, start with the basics but keep embedded up your sleeve like a wildcard.

The effort is worth it and you can really blow some minds in how easy it is to incorporate data and insight seamlessly into a custom website.

Good luck on any projects in this area.

Supporting documentation

Embedding in ThoughtSpot

Runtime filters

--

--

Kris Curtis
CMD'ing Data

A data professional for 17 years, focusing on educating and creating possibilities for business users to embrace the use of data.