A step-by-step guide to building and delivering embedded analytics

Sharon Zhang, Senior Consultant, Looker Professional Services, Google Cloud

Sep 10, 2020

There’s no question that analytics can be beneficial to your organization’s growth and success. Embedding analytics into your product can create new revenue streams. You can use data or analytics to create opportunities for upsell. You can even offer it as an entire new feature set or product line.

But even with the benefits of rolling out a productized analytics offering being as immense as they are, the task of actually building one can be daunting.

After helping deploy more than fifty embedded analytics projects, I’ve seen all the common concerns to this challenge. Most of my customers worry about the people and financial resources required to build out the offering, the time it will take to implement, and how quickly user needs change. One of the questions I hear most often is ‘Can we adapt quickly enough as customer needs evolve?’

The answer is yes.

Over the past few years, I’ve developed a success framework that has helped many companies de-risk the embedded analytics journey. In this blog, I’m going to share that framework with you and walk you through a few key things you can do to build a strong foundation before you get started on your embedded analytics journey. From there, I’ll share how to structure your deployment in phases with some simple building blocks to maximize your success.

Build a strong foundation: four key ingredients

Before you build your analytics offering, I encourage you to consider four things that will become the foundation of your deployment:

  1. The database you choose
  2. The analytics platform you choose
  3. Your software development resources
  4. Your data product owner

Let’s dive in!

1) The database

The database that will ultimately support your analytics product will need to be scalable enough to support the volume of data and type of analysis you’ll be offering. I recommend selecting a database that supports high concurrency — meaning it can handle a large number of users accessing and interacting with dashboards and running queries simultaneously. If you already have an internal use case that requires a similar database, you’re even closer to being able to deliver embedded analytics than you realize.

You can learn more about the database ecosystem here.

2) The analytics platform

Whichever analytics platform you choose, you’ll want to make sure it provides agility and flexibility because your data needs will evolve. Powered by Looker (PBL) supports any external analytics use case, enabling you to deliver the full functionality of Looker as an external service.

With our OEM offering, you can expose your Looker instance directly to your customers or partners with no software development, which results in quick time to value. You can expose elements of the Looker UI, like dashboards, in an existing application through embedding for a more integrated end user experience. You can also create a fully customized application using Looker’s application programming interface (API) as a backend. This takes longer to develop, but provides your customers with a highly customized experience.

You’ll never be locked into a single PBL solution. You can easily move between PBL offerings and mix in other options, depending on your analytics use cases and what you’re productizing for your customers.

3) Software development resources

Before you launch your offering, determine the technical resources you’ll need to model your data and to build your embedded analytics application. If you have limited internal resources, don’t worry! Looker’s Professional Services team and Partner Network are here to supplement your in-house resources. We’ve done this many, many times for organizations of all sizes and across various industries.

4) Data product owner

This is one of the most critical components to have in place before you start building your productized analytics offering. When you select a product manager, ensure that they are aligned around and understand the vision for the analytics product offering. This person will become a product and customer evangelist, driving communication about your product, helping determine the features that should be included, and controlling the timeline for launch, so it is highly important that they are equipped with the knowledge and authority necessary to drive the project forward.

The three building blocks for a successful launch

At this point, you’ve put in the time and thought through the foundational elements that will underpin your analytics offering. As you prepare to launch your product, I have three recommendations to help you ensure that it is a successful launch. By approaching this framework as a phased process instead of a massive, monolithic goal, you’ll enjoy a smoother path as you work towards achieving the ultimate goal you’re reaching for.

Block 1: Whet your customer’s appetite for data

Getting your customers excited is the first critical step in this phase. I like to think of this as building an embedded analytics “starter pack.” Build three to five high-value dashboards that are technically straightforward to implement. With these built out, you can choose to embed these as part of your customer portal through SSO embedding, or can use them to show meaningful data — in view-only mode — to a handful of customers. The goal here is to get customer feedback so you know where you should invest time and resources next. The initial investment should be in engaging with your customers after you release these dashboards. In my work on the PS team, we’ve seen customers successfully beta launch this type of offering in as little as three to six weeks.

Block 2: Add new functionality and increase user base

Once you’ve solicited customer feedback from your first few dashboards, continue to build on that starter experience. For example, you can use the information from the beta launch to create additional features or deploy a tiered analytics offering with certain features available for a premium upsell. You can also start to create different levels of user access by adjusting Looker embed user permissions. During this phase, many Looker customers also start to build additional screens to expose user generated content, offering a more seamless content navigation experience. These features are straightforward to add, as the foundation comes out of the box from Looker.

This is also the building block where you can explore how to monetize your embedded analytics offering. For example, some customers offer a tiered plan that looks something like this:

  • Basic Tier
    • View preset reports within a limited dataset date range (example: last six months)
  • Advanced Tier
    • Basic + the ability to drill down, download, and schedule reports and see 12 months of data
  • Premium Tier
    • Advanced + Ability to explore, save, and share content with other team members and see data for all time ranges
    • Access to support team, custom report building

Check out how Looker customer SupporTrends built an embedded analytics solution in only 6 months and took just 2 weeks to implement Looker visualizations.

Block 3: Customize the application experience

By the time you reach this building block, you should have a really good understanding of what your customers want. You’re already delivering value to your customers, so you’re ready to begin customizing your offering. If you tried to deliver a customized experience right out of the gate, this would be daunting and a bigger lift. But you’ve already set up the infrastructure — users and APIs, for example — so your developers can now focus on building new features. We recommend you start by adding in commonly requested, high-value custom user interface (UI) features, like more advanced analysis, a custom cohort builder, or a report builder, which is essentially a lighter version of the Looker explore experience.

You can use the Looker API as a “backend engine” to power this more customized experience. This enables you to fully integrate your data into your application UI so it looks and feels like your application, rather than the Looker UI embedded in your application. You can also use the extension framework to jumpstart this type of custom development. This framework significantly reduces the effort and complexity of building custom data applications and tools by handling functions like authentication, access control and permission management, and API access.

Did you know? If you’re already a Looker customer, you’re probably already 70% of the way towards creating an embedded analytics solution! You’ve already cleared some of the toughest hurdles we see in PBL implementations. Your data is ready. Your developers know how to use Looker and LookML. And, you have your data model, logic, and dashboards already built. All you have to do now is expose what you’ve already built to your customers. Down the road, you can iterate and customize as you see fit, and you don’t have to start from scratch.

You’re on your way towards delivering an embedded analytics solution

I hope this blog has given you a better understanding of the embedding analytics process and shown that taking the first steps on your journey towards offering an embedded analytics solution isn’t as scary as it might seem. The benefits are clear, and if you think about a few key things before you get started, you will have a strong foundation to build on.

And don’t forget, our Looker teams are here for you throughout your journey. Below are some great resources that you can learn from as you approach embedding analytics at your organization, and you can always reach out to our team to learn if PBL is the right approach for you.

Additional embedded analytics resources

Next Previous

Subscribe for the latest posts