How did Notion improve its ROAS by 38% without increasing its budget?

Case Study

Notion

Industry

PLG, SaaS, productivity and note-taking web application

Highlights

Customer Journey: Notion, is a productivity and note-taking service. Notion's 'try before you buy' freemium model extends the path to becoming a paying user, often stretching over several weeks.

Proxy KPI Limitation: Initially, Notion relied on sign-ups as a proxy KPI for Google campaigns, which was suboptimal due to its weak correlation with the ultimate goal: conversion to a paying user.

Conversion Prediction: Notion implemented Voyantis' value prediction, integrating it with Google's value-based bidding to predict the likelihood of free users converting to paid subscribers.

Impressive Outcomes: Notion experienced a 38% boost in its predicted 6-month ROAS and a 21% uptick in customer activation.

“Conversion to paid users was very hard to measure. We knew we drove users but the causal relationship to paid users wasn’t clear. It was hard to feed
google back with relevant inputs”

Fabien David

Performance Marketing Manager

The Challenge

Campaign funnel with/ without Voyantis | Visual

Campaign funnel without Voyantis | Visual

Notion's freemium model leads to a prolonged conversion timeline, with transitions to paying customers taking from a few weeks up to several months.

Notion focused its campaign strategies on early-funnel events like sign-ups due to their immediate trackability.

Relying on sign-up metrics for optimization proved suboptimal as it didn't accurately indicate the user's likelihood to convert into a paying customer.

Notion aimed to transition to value-based bidding, utilizing predictions to guide Google algorithms towards targeting higher-value customers.

The Pilot

Notion set up an experiment on Google with two non-search campaigns, each having the same budget cap.
‍
The control group focused on cost, using tCPA optimization for sign-ups, while the Voyantis experiment group focused on optimizing for the likelihood of conversion to a paying user, based on predictions from Voyantis' AI models.

“It wasn’t just easy to work with Voyantis’ signal optimization solution, it generated great results too. We saw a 21% increase in team activation rate and a 38% predicted 6-month increase in ROAS”.

Fabien David

Performance Marketing Manager

How does it work

Data
processing

Voyantis created custom prediction models to generate conversion prediction based on Notion’s anonymized data.

Prediction model generation

Voyantis' AI engine analyzes Notion's unique customer behaviors and attributes, correlating them with conversion trends to construct a specialized prediction model designed exclusively for Notion.

Value-based bidding

Predictions at the user level, are fed to Google using Google’s server API enabling Google to optimize campaigns for tROAS using value based bidding.

Accuracy verification

The model is constantly tested and verified by the Voyantis AI engine to generate the highest accuracy of future predictions and adapt for future upgrades of the Google algorithms.

The Impact

The trial clearly demonstrated how customer value predictions can improve campaign performance, suggesting a shift away from focusing solely on CPA

pROAS increase

ROAS Increase: Notion saw 21% increase in team activations, and 38% increase in predicted 6-months ROAS, all while keeping the same overall campaign budget.

Longer term impact

Cost efficiency: Notion significantly lowered the cost of acquiring paying customers, maximizing earnings without expanding the budget.

Scalable effect

Attracting larget teams: Beyond the 21% increase in team activations, Notion attracted higher-value customers, evident by a 35% rise in accounts with more than three users.

+21%

New team activation

-32%

Cost per a paying customer

+35%

Customers with +3 users

“Here at Notion we’re excited to continue our partnership with Voyantis as we
expand our growth strategy”

Fabien David

Performance Marketing Manager