When Voiceflow’s head of Growth Daniel D’Souza looks at his self-serve funnel today, he sees that since going live with Toplyne, Voiceflow’s activation rates have increased 5x, and their 30-day free-to-paid conversion rates a whopping 3x.

Daniel uses Toplyne to identify high-potential users early in their lifecycle to engage through community invites, workshops, and meetings with their BDRs.

Meet Voiceflow

Co-founder and CEO Braden Ream describes Voiceflow as a “Figma-quality” collaborative platform for building conversational AI agents.

Voiceflow started around 5 years ago in the conversation design space, initially working with companies like JPMorgan, Home Depot, BMW, and other Fortune 100 companies on building and launching their AI agents. 

‍

These AI assistants connect with your back end, grab information from your database, combine that with information from GPT or other large language models, and then deploy that to a web chat, mobile app, or website embed via their dialogue API.

“Voiceflow allows for integration with any existing natural language understanding platform or technology stack, allowing teams to supercharge their design and collaboration capabilities without costly vendor replacements or technology changes.” - Braden Ream

At Voiceflow, Toplyne is used by the two people who lead their monetization engine - Sam Burns, head of Marketing and Daniel D’Souza, head of Growth.

Go-to-market Challenges

Riding the publicity wave of the ChatGPT announcement in late 2023, the market for tools helping users build conversational AI experiences exploded.

“Voiceflow’s signups and usage have roughly doubled since the launch of ChatGPT in November, as the interest in conversational AI grows,” said Braden Ream talking to TechCrunch.

Voiceflow is free to get started, with three paid plans on offer - a starter plan for individuals, a teams plan for businesses, and an enterprise plan with the full suite of features.

As Voiceflow grew, their GTM team faced three primary challenges:

1. Declining activation & conversion rates - Daniel needed a way to use product usage signals to nudge new sign-ups to activate and then convert to paid plans

2. BDRs had a needle in a haystack problem - They needed a way to identify enterprise Product Qualified Leads (PQLs) from self-serve users

3. Data literacy - Sales & CS teams at Voiceflow had no access to product usage data of users.

‍

Considerations

As Daniel and team looked for ways to address these GTM challenges, they had a few considerations. Daniel wanted any solution to be able to:

1. Connect and stitch first-party data from different sources - Voiceflow uses Segment, Mixpanel, BigQuery to track user data. He wanted something that wasn’t another layer, but instead helped them connect different sources and put it in the right place.

2. Build Custom AI models - Daniel needed a way to use plug-and-play AI models on top of Voiceflow’s first-party customer data, with the flexibility of layering on their own understanding of an Ideal Customer Persona.

3. Work with messy data - Voiceflow’s warehouse data was messy, as is the case with every other company at their stage of growth. They needed any solution to be adaptive to this messy data

‍

‍“The Toplyne team has really helped us actually better leverage the data, which was a huge blocker before and why we couldn't actually connect it into a lot of these places ourselves.” - Daniel

Voiceflow's Toplyne AI Agents

Each Toplyne agent is trained on Voiceflow’s first-party customer data. Daniel and team configures the metric they want to move and trigger the Toplyne agent to then complete particular GTM tasks on a recurring basis. Voiceflow currently uses two Toplyne agents and is experimenting with two more.

‍

Agent 1: Activation rate boost

‍A Toplyne AI agent trained on Voiceflow’s product usage data in Segment identifies users early in the lifecycle who are primed for activations and targets them with calls, invites, and workshops.

The AI agent identifies the audience of users primed for activation through a combination of two factors:

1. ICP Definition - Detects accounts that satisfy Voiceflow’s definition of an Ideal Customer Persona - accounts where at least one conversation designer and a developer who have signed up.

2. Intent - Accounts that have the highest scores based on product usage in Voiceflow.

Impact: Toplyne’s activation agent increased Voiceflow’s activation rate by 5x leading to a 3x increase in downstream 30 day self-serve conversions.

“We started having more enterprise customers focused on production who were coming in because they were getting activated easier. And so that's really where Toplyne  had the biggest impact for us.” - Daniel

Agent 2: Product signals in your CRM 

Toplyne syncs product usage signals in the CRM for Customer Success teams to access through the Toplyne extension on Salesforce.

The extension distilled the most useful information from product usage data for sales and CS:

• For AEs to look through companies going through an active trial and see the actions they’ve taken so far + assess the next best action they should be taking

• For CS to look at onboarded accounts and assess account health and whether they’re progressing towards value

Impact: Data literacy improving efficiency of Sales & CS teams.

Uncovering more use-cases with Toplyne

Daniel’s team is keen to uncover more use cases with Toplyne. The Toplyne workflows that have worked at the top of the funnel are now being deployed further down the funnel:

Agent 3: Enterprise pipeline for BDRs

‍Toplyne’s AI finds product qualified leads primed for enterprise users from free & teams plans through both usage signals and Voiceflow’s ICP definition.

Agent 4: Next best action

‍Toplyne predicts what the optimal “next best action” to take on a customer is, in order to drive a custom goal like conversion, activation, or expansion - be it an in-app nudge, an email, a human intervention, etc.

With his first-party data sources all consolidated by Toplyne, and AI models running on top, Daniel and Sam are excited to uncover more use cases with Toplyne.

But more on that later.