How WeTransfer used a Growth Model to predict marketing ROI and identify biggest revenue opportunities
How WeTransfer used a Growth Model to predict marketing ROI and identify biggest revenue opportunities
I helped WeTransfer build their first Growth Model. I cover why and how we built this model; as well as how it was leveraged by product and marketing teams.
In this article, I’m trying something new by sharing details of a project I’ve done for WeTransfer. The goal is to make the use case and benefits of building a Growth Model as tangible as possible. I hope you enjoy the read!
Used by over 80 million people to send files every month
Launched in 2009 in the Netherlands, WeTransfer is eponymous with file transferring within the creative industry. It’s a well-loved brand (B Corp™ and Climate Neutral certified) that has had organic hypergrowth over the last decade. The company has two monetisation models:
Subscriptions: Users can use the product for free, or upgrade to one of their paid plans for increased transfer size, storage and collaboration tools.
Advertising: Free users get served premium advertising on the background of their transfer pages.
The problem: the intersections between product + marketing and subscription + advertising muddle the picture
WeTransfer already had marketing measurement best practices in place: server-side tracking, triangulated attribution, incrementality tests and quarterly MMM analysis. They used that to understand marketing performance of their campaigns, which were used to drive new subscriptions.
But it still didn’t provide the full picture. The indirect impact of marketing on product virality and on advertising revenue wasn’t included. There was a set of questions the company couldn’t answer:
What’s the impact that product metrics (e.g. percentage of users that transfer a file) have on a campaign?
How should the company navigate the balance between driving traffic for advertising; versus driving Pro and Premium conversions?
How do free users acquired with paid media impact product virality and future organic subscriptions?
What’s the optimal budget to spend on awareness and conversion campaigns?
Best practices measurement wasn’t enough for a complete picture. The indirect impact of marketing on product virality and on advertising revenue wasn’t yet quantifiable.
Without understanding how product and marketing affect each other, the company couldn’t identify the most impactful initiatives. And, without the most impactful initiatives, there was a lack of clarity on where to best allocate resources.
To summarise, WeTransfer needed to uncover:
The overall ROAS of campaigns, that’d take into consideration: advertising revenue, subscription revenue and virality (referred users)
Biggest opportunities to increase ROAS, that’d take into consideration non-marketing initiatives that’d increase conversions, activation, upselling and “shareability”
The goal: a holistic product and marketing roadmap including both advertising and subscription revenue
WeTransfer identified Growth Model as a potential solution to their resource-allocation question. The model was to serve two primary functions:
Pinpointing high-impact regions, or “levers”. Identify the biggest areas of revenue opportunities for the business across the whole funnel.
Ensure alignment of objectives. Align Marketing and Product teams to focus together on enhancing the key levers.
The model had to go beyond attributable subscription, and provide a holistic view of all revenue streams, including advertising and free users.
It also needed the ability to calculate short term and long term impact, in order to measure the organic viral aspect of the product in its integrity. The first use case we chose for the model was to identify improvements for paid media.
Muhammaud Daud, my manager on the project and WeTransfer’s Growth and Marketing Analytics Director, shared this with me:
When it comes to conversions, there's a ceiling to the improvements we can achieve on ad platforms. We needed to understand the network effect and pinpoint the inefficiencies across the entire customer journey.
The execution: building the model to measure WeTransfer’s growth engine
The process I followed for building WeTransfer’s Growth Model was similar to what I’ve done for other PLG brands (like Homerun’s), albeit a bit more complicated due to the secondary revenue source (advertising). If you’re looking to read more on Growth Models, Matt Lerner has a lot of resources on the topic.
Part 1: Mapping the funnel and identifying the levers
A qualitative analysis was the first step taken to build the model. We mapped user journey, identifying the different touch points that lead to revenue being generated, either via the product or via advertising.
After the funnel mapping, metrics were defined for each step in the user journey, like “Average Downloaders per Transfer”, “Average Conversion Rate from Free to Pro” and “Average Advertising RPM”.
Part 2: Sourcing the metrics
We sourced the benchmarks for our identified levers / metrics. To get a full picture, we used an array of data sources, such as Google Analytics, Snowplow, advertising sales assets and marketing platforms.
Part 3: Constructing the Model’s Logic
This phase of model construction focused on the logic underpinning WeTransfer's growth model. This involved:
Identification of Levers: Levers are metrics that can be modified by the business and can influence revenue. For instance, “Average Downloaders per Transfer” is a lever. WeTransfer can encourage users to share their files to many people, which can move the metric, the product’s virality impact, and later on the revenue.
Connecting the Levers: Using Google Sheet, we connected the different levers to calculate the final predicted revenue. This enabled us to measure the financial impact of modifying each lever by X%.
For the WeTransfer Growth Model, we included 15 individual levers. They covered product usage (e.g. “Monthly Retention”), onboarding (e.g. “Conversion to Free Account”), advertising revenue (e.g. “RPM”), among others.
The result: interpreting insights and recommending marketing initiatives
Visualising most impactful and most movable levers
With the complete Growth Model, we could now move towards the analysis and recommendations stage. Our first step was to score each of the 15 levers as “High”, “Medium” and “Low” in two different criteria:
Impact: The impact that moving this metric has on the revenue. For example, “Conversion Free to Pro” is one of the highest impactful levers for most PLG companies.
Confidence: A guesstimation on how likely it is that we can move this lever. This was based on industry benchmarks and the company history (how many activities they’ve already launched to move this lever).
We analysed all levers to identify which ones we were most confident we could impact. The name of the levers is blacked out.
Based on these two scoring systems, we created a visual “opportunity map” to be shared with the business:
Recommending and prioritising initiatives
With a clear overview of the highest impactful + highest confident levers, the business now had a data-driven recommendation on what metrics to focus on. As part of this project, we also delivered an overview of different marketing and product initiatives that could move each of our priority levers.
The recommended initiatives covered both product and marketing. It included lifecycle activities, product growth optimisations, landing page suggestions, among others.
Getting Product and Marketing on the same page
The Growth Model proved that Product and Marketing can’t work in silos. They’re both in constant friction on how users discover and use the product. Growing a business through marketing alone isn’t as efficient as making it a combined effort.
It also brought clarity to the business on what were the opportunities (levers) to drive revenue, and what are some of the ways (initiatives recommendation) of getting there. Armed with this data, Product and Marketing could work together on shaping a roadmap that drives the biggest impact to the business.
This is what Muhammad had to share on the importance of getting Product and Marketing on the same page:
To make an impact, we have to think of marketing and product as two levers in our growth engine. The hardest part is to quantify this symbiotic relationship and this model definitely helps achieve that.
Moving forward: Automating the model for full-funnel analysis
The first iteration of WeTransfer’s Growth Model was finalised in 2022. By Q4 2023, most recommended initiatives of the first version had already been implemented or are currently in the roadmap.
The Business Performance team is taking the Growth Model further, with a more mature version. The new iteration will be automated and based on their first party data (Snowplow) with the goal of providing a full-funnel analysis of their hard-to-attribute top of funnel initiatives (like YouTube campaigns).
Some additional thoughts to close this out
If you’re interested in building something similar for your brand, then do reach out (personal website here). And, if you enjoy reading articles on the intersection of marketing and data, don’t forget to enter your email and subscribe: