Heap announces new generative AI CoPilot

Heap, the leader in product analytics, unveils AI CoPilot's open beta today. This generative AI tool empowers product managers, designers and marketers, with a chat interface, AI-generated chart summaries, and insightful follow-up questions, making advanced analytics accessible to all.

Heap pioneered auto-capture in product analytics, empowering digital teams to ask new questions on historical analytics events and get instant answers. Later, Heap introduced a proprietary data science layer to the auto-captured data, automatically surfacing hidden friction points and unhappy paths that were never thought about in tracking plans but were impacting conversion, user engagement, and retention. 

As part of Contentsquare, Heap is combining its best-in-class quantitative analytics tools like cross-platform journeys, retention, and lifetime value analysis alongside Contentsquare’s industry-leading experience analytics capabilities such as session replay, zoning, and more.

These represent a powerful set of tools, but all powerful tools require some expertise to fully leverage their advanced capabilities. There remains a challenge – how do you drive widespread, self-serve adoption of these advanced analytics tools in a team that is less experienced with analytics?

What’s wrong with analytics today

As we speak with digital teams at both our current and prospective customers, we see a common trend – every team has an “analytics enabler”. 

This person is usually a blessing to us. They are the advocate for investing in analytics, the educator who uplifts their teammates, and the creator of data-driven culture in their organization, but there just aren’t enough of them.

Ask yourself this: how big is your product management or digital marketing team?  How many members of these teams are able to, independently dive into your analytics data, ask the right questions, and interpret the resulting data to get actionable insights without any specialized training workshops or talking to an analyst?

In most organizations, we’ve found that this number is a small fraction of the overall team – the majority, on the other hand, have limited experience with any kind of digital analytics tools. 

What if we could do better? And I mean a lot better.

The opportunity with Generative AI

You can’t go online or attend any conference today without spending the day reading or talking about AI. But is it hype or reality? And how should you best leverage it?

We believe that what makes generative AI uniquely practical and valuable to the world of digital analytics is its ability to understand users’ natural language and explain complex concepts back to a user in natural language. 

Let’s look at what those “analytics enablers” are doing, but with a lens of what’s now possible with generative AI:

• They take requests in natural language, usually in the form of a quick ad-hoc question about a feature, a marketing channel, etc.

• They know where to look – what chart should you create? Which events are the right ones to use? Should you be applying a group-by or a filter for this question?

• They iterate – you can always ask follow up questions like “how does that vary by country? On web vs mobile?”

• They explain their reasoning – an important part of being an enabler is teaching, and great enablers do this in context as they answer ad-hoc questions.

What if we could take these capabilities and bake them right into our product? We could help significantly reduce the workload of these “analytics enablers” and in doing so, empower a whole new population of novice users to get value out of digital analytics.

Introducing AI CoPilot in Heap

AI CoPilot is a set of generative AI powered capabilities that are designed to empower, educate, and enable users new to the world of analytics to quickly get started and get to valuable, actionable insights in Heap. These capabilities include:

Chat:

Ask questions in plain English and Heap's AI CoPilot will select the right chart, events, segments, properties, filters and groupings for you automatically. AI CoPilot will also build the right charts while considering past conversations to answer your follow-up questions. Plus, our unique chat combines help articles with the analytics engine, guiding you step-by-step. Just ask "What's next?" or "How do I analyze a feature launch?" and AI CoPilot will create a personalized plan for you leveraging Heap's help center documentation, best practice guides and online training materials.

Summaries:

When creating a new chart, AI CoPilot will automatically create a title for the chart and describe it in the form of a plain english question describing what the chart is trying to show.  CoPilot will also highlight key choices like events, groupings, or filters that impact the analysis, without the complexity of showing the same query builder interface used to create the chart. This makes for much more approachable, readable charts when sharing insights with your peers.

Follow-ups:

Using the auto-generated description on charts, AI CoPilot will identify similar questions asked by others at your organization and highlight a set of up to 3 follow-up questions to show alongside each chart. These questions are aimed at encouraging iterative analysis (for example: breaking down a time trend in signups by country or user segment to learn where new users are coming from). With Follow-ups, you don’t need to have a question in mind – a teammate can share any chart with you, and in one click you can be chatting with AI CoPilot to dive deeper into the data. 

Transparency and Privacy

We believe that AI shouldn’t be a black box. At any point when using Chat, you can click the "understand my approach" button and the AI CoPilot will explain why it chose the type of chart and the specific events or properties used in the analysis. This is a great way for new users to get familiar with their account’s dataset and Heap’s data model in context.

Heap also maintains a privacy-centric approach to applying generative AI. While AI CoPilot is powered by the OpenAI API Platform, no customer data is used to train third party AI models. In fact, unlike other solutions on the market, customer query results are never sent to a third party AI service at all. Heap sends a proprietary set of metadata including help center content, names of relevant events, properties, etc. from the customer’s Heap account, and the user’s question to OpenAI, which returns instructions on what analysis to perform. That analysis is performed securely in Heap’s interface, just as it would be if the user completed it manually themselves.

How to get started

As of today, AI CoPilot is now in a public Beta for all Heap customers in our Growth, Pro, and Premier plans, available at no additional cost. 

Over the next few weeks, we will be rolling out access to AI CoPilot for eligible customers. After a qualified Heap administrator opts into Heap’s AI beta agreement, users in the account will be granted access to use AI CoPilot. We have also provided a variety of controls for administrators who would like to manage their own rollout of Heap’s AI CoPilot within their organization.

Using Heap to power your AI

While today is an exciting day for generative AI capabilities in Heap, we already have many customers who have been actively using Heap data to power AI applications in their own businesses (and you can too!).

Heap Connect is Heap’s industry-leading managed ETL solution that makes it easy to sync user behavioral data between Heap and your data warehouse. This is particularly helpful when building AI models for:

• Cross-sell: Use data on which product pages, collections, and variations a user viewed and purchased to drive cross-sell and increase order values.

• Upsell: Identify which users or accounts are good candidates for upsell to add-on SKUs based on aggregate user behavior; which features are users using? How often? What is the tipping point for purchase of an add-on with a trial?

• Fraud detection: Use behavioral signals alongside user demographics to identify fraudulent behavior, and leverage Heap’s auto-captured dataset to quickly iterate on features in fraud detection models.

Data teams have loved Heap Connect for years because of its ability to automate the tedious data engineering work of creating and managing data backfill and migration pipelines. Our managed ETL saves them countless hours, and our auto-captured dataset enables them to start training models months sooner than they would otherwise, yielding a faster time to market and more impact from their AI capabilities. 

What’s next for AI at Contentsquare?

We are so excited about what generative AI can do for Heap and for analytics as an industry, and we hope to welcome many of you to use AI CoPilot in Heap in the coming weeks as a first step.

Chat, Summaries, and Follow-Ups are just the beginning – with our north star focused on bringing the “analytics enabler” directly into our product experience, we expect to see generative AI appear in many more places across Heap and Contentsquare in the coming months, with a special focus on understanding, summarizing, and educating in context. 

When we look back at this first step years from now, we aspire to live in a world where entire digital teams, regardless of their experience level, are able to easily dive in, get the personalized in-context help they need, and get valuable, data-driven insights. And hopefully learn something along the way.