How Miro Acquires 30% More Paying Customers with No Extra Budget
By the numbers
Overview
Company
Miro is a digital whiteboard platform used by over 80 million people to brainstorm, plan, and collaborate in real time
Industry
SaaS / Subscription
Headquarters
San Francisco, CA
Campaign Type
Google Search
.png)
Snapshot
Not only have we brought more paying accounts with the same media spend, we brought accounts with higher value. We brought more annual plans (compared to monthly plans). We also acquired teams with a higher number of members for each account.
Miro offers digital collaboration tools with upgrade options for business or corporate plans. Sign-up starts free and conversion to paid subscription typically spans days, if not weeks.
Campaigns before Voyantis relied on sign-up rates as a proxy conversion event for Google campaign optimization. Voyantis reshaped the goal for Google campaigns, optimizing for probability of converting free trial users into paying corporate clients.
Armed with the new predictions, Miro conducted an A/B test with value-based bidding and a tROAS goal. Miro achieved a 30% uptick in paying customers at the same spend, alongside a 33% boost in pROAS.
Impact at a glance
Increased paying corporate customers by 30%
at the same media spend
Boosted pROAS by 33%
through value-based bidding with predictive signals
Grew annual plan share by 35%
over monthly subscriptions, reducing churn and improving revenue predictability
Lifted average seat count per customer by 14%
at renewal, acquiring larger teams that compound over time
The challenge
Miro's funnel begins with a free trial. Most conversions to paid users take place after the first seven days, if not longer. Given their lengthy conversion funnel, Miro used a proxy KPI - corporate sign-up - to optimize its Google campaigns.
Optimizing for sign-ups wasn't optimal because it didn't directly reflect the likelihood of converting to paid users. Miro needed to optimize for a signal further down the funnel, and ideally move from tCPA to value-based bidding.
The solve
Predict conversion likelihood with a custom model
Voyantis developed bespoke prediction models using only anonymized data, generating conversion forecasts based on Miro's anonymized engagement, transaction, and onboarding data. The platform analyzes Miro's customer behaviors and attributes, identifying signals and patterns that predict the transition from trial users to paying customers. The model achieved 90% accuracy.
Deliver predictions to Google in real time
Voyantis generated user-level predictions to feed them to Google using their Server API, informing value-based bidding algorithms.
Validate through a controlled test
Miro conducted a test with five non-branded search campaigns targeting a US audience, equally dividing the budget between a control group and a Voyantis experiment group.
The control group focused on optimizing for the old goals - tCPA and corporate sign-ups. The Voyantis group optimized for the likelihood of converting to a paying user, based on Voyantis's proprietary prediction models.
Maintain accuracy over time
The model is continuously tested and verified by the Voyantis platform to generate the highest accuracy of future predictions and to aintain effectiveness as Google's algorithms change.
Voyantis’ impact
33%
increase in ROAS
30%
increase in new paying customers
35%
increase in annual plans
More paying customers at the same spend
Miro experienced a 30% surge in paying customers and a 33% increase in predicted ROAS, with the same campaign budget. Beyond immediate ROAS gains, long-term revenue impact was achieved, with renewals boasting an increase in average seat count per customer and annual plan subscriptions taking 35% more share over monthly ones.
Voyantis's impact remained robust as Miro increased budgets and scale, proving widespread scalability across Miro's operations.
We were able to bid more effectively while bringing in more users, it is important because it sets the stage for future scaling. It was important to verify that the new bid strategy with Voyantis is scalable.
