Growth case study: Self-serve growth
A detailed look at the journey to double Pinterest's self-serve advertisers
Introduction
In the summer of 2018 Pinterest embarked on a mission to jumpstart self-serve advertiser growth. This began with a small tiger team to prove out the concept and led to multiple teams, disciplines and executives coming together to drive a 2.5x increase in active advertisers. This post breaks down each stage of the initiative into the funding rounds of a startup and summarizes the learnings and outcomes from each.
0-3 months - Seed capital
A tiger team was initially created with the mission to prove out the self-serve advertising opportunity on Pinterest. This period most closely resembles a startup that has just received seed capital. The time frame given to show traction was 3 months.
Team
1-2 Engineers
1 Designer
0-1 Product Manager
0-1 Engineering Manager
0-1 Product analyst
Strategy
The initial hypothesis asserted that the self-serve product generally worked well enough to begin growing it. If incorrect, there was a possibility of churning out advertisers before getting traction. As with any new idea, risk will be prevalent in the early stages and high stakes decisions are often unavoidable.
We believed we had obtained product market fit for a subset of our advertisers. So we simply needed to get businesses to the ad product and they would become advertisers (easy right?). We wanted to do this by introducing new entry points within the highest traffic areas of the business app to bring awareness.
Tactics
The team focused on conversion optimization tactics to accomplish this strategy. These tactics generally include creating different types of upsells and entry points to bring awareness to all users. During this period the team ran 35 experiments of which 12 ultimately shipped to production.
A banner across different surfaces on Pinterest letting businesses know they can promote content
An entrypoint to allow businesses to directly “Promote” when viewing their own Pin
An upsell on the homepage that highlights a specific pin to promote
Outcome
Through a series of wins including the highlights above, we saw a reacceleration of the topline advertiser growth rate by about 15%. Although we performed proper experiments we didn’t understand whether or not our changes were causal. The data we did collect, however, was compelling enough to convince leadership to officially fund the team.
We also saw signs of entrypoint saturation (e.g. too many red buttons on a surface) due to our initial conversion tactics. The team even began to receive feedback that businesses were seeing conversion upsells that were out of context. With this in mind we knew we needed to change our strategy if we wanted to deliver consistent results.
3-9 months - Series A funding
After the initial tiger team’s success, we were officially funded with a full team. At this point the team could be compared to a Series A startup that has shown some success but needs to show consistency for further investment.
Team
3-4 Engineers
1 Designer
0-1 Design Lead
1 Product Manager
1 Engineering Manager
1 Product analyst
Strategy
Product
In light of the entrypoint saturation risks we discovered with the initial conversion strategy, new areas of development were essential. With a fully staffed team and more experience within the space it became easier to make strategic investments. Specifically we wanted to invest in onboarding. The team observed that even if there was a focus only on new businesses, an outsized impact on advertiser growth was viable. This resulted in a deep dive into the new business funnel which gave the team focus and constraints to generate creative ideas.
Sample advertising funnel for the first week a business signs up (Not real data)
Data
As a tiger team most data projects were dedicated to finding additional insights to create new experiments. With full product analyst support there was more capacity to invest in data infrastructure. We specifically focused on visibility of mission critical metrics, availability of fast tables to perform ad hoc analysis and automated measurement capabilities to understand experiment impact.
Tactics
The largest drop-off in the new business funnel was when an ad campaign was being created. The initial experiments were around education and guidance focused on this specific flow. During this period the team ran 56 experiments of which 24 ultimately shipped to production.
For new users with content to promote we offered an advertising credits to see if we could lower the bar for trying the advertising product
When a user creates their first ad, give them some education on how the process works and what to do
Outcome
With the new dashboards built the team could clearly see impact on the topline. At this point the number of advertisers grew by about 50%. One of the major contributors to this growth was the new technology to measure per experiment impact. Specifically an easy to run report was created which outputted the total number of new spenders generated per experiment. An internal tool may not seem as impactful as delivering a feature but in this case it allowed the team to make much better decisions. An example of this was internationalization feature work that had been originally written off. Due to the small population size we guessed it wouldn’t be impactful, but we ran an experiment anyways. The results showed it was a huge success according to the report. At the end of the 3-9 month period internationalization contributed about 25% of the impact delivered by the team.
With clear goals and metrics other teams began to pick up the flag and worked towards the same advertiser growth goals. This led to several foundational surfaces and platforms being built out for businesses. These core features didn’t necessarily contribute a step change in growth on their own but paved the way for future growth for Pinterest’s platform.
9-20 months - Series B funding
The team began to deliver more predictable results and could clearly articulate how more funding would drive growth. This is comparable to a startup that just received it’s series B and is injecting more talent and capital to scale the business.
Team
4-7 Engineers
1-2 Designers
0-1 Design Lead
1 Product Manager
1 Engineering Manager
1 Product analyst
Strategy
Product
The team was running out of runway in the current surfaces and channels it had explored. It felt like fundamental pieces of the advertiser experience were missing but what was needed wasn’t necessarily new growth features. Fortunately new platforms and surfaces, built by our partner teams, were coming online to bridge the gap. This gave the growth team more room to take bets on these brand new surfaces to pave the way for the next step change in growth.
Data
At this point the team had mostly focused on getting new advertisers to begin spending. Because this direction represented acquisition but didn’t consider activation or retention there was a concern that this strategy may lead to churn. A lack of visibility on a per experiment basis of activation impact compounded this concern. As a result, we shifted our data strategy to begin focusing on whether advertisers were returning to our platform and spending again.
Tactics
Given that the number of engineers on the team doubled, we could employ multiple tactics to drive growth at once. One of the new surfaces brought online during this period was advertising through the mobile app which brought a host of new opportunities to drive new advertisers. Another new opportunity presented itself from the previous conversion tactics we had implemented in the last 9 months. This resulted in 2.3x the traffic on the campaign creation page making it a great target for improvements in advertiser comprehension and ease.
An additional benefit of more resourcing was the room to reinvest in bets that were already paying off but needed more technical effort to scale. During this period the team ran 97 experiments of which 50 ultimately shipped to production.
Enabling users on mobile devices to advertise
A number of experiments to optimize the quick promote experience including budget changes, audience information and more
Redesigned the homepage for businesses to make it more creation and advertising focused
Outcome
Over this period we saw a step change in advertiser growth. The growth engine was firing on all cylinders, leading to the number of advertisers on our platform doubling over 12 months. The tactics highlighted above accounted for the lionshare of impact during this time period.
These results underscore the importance of building a foundation in order to truly have breakout growth. Visibility, measurement and core product improvements don’t necessarily drive explosive impact directly but without them a growth team will quickly run out of gas. With this said cross functional alignment proved pivotal to this success. The Data, Ads and Growth teams all worked together to deliver the final result.
Conclusion
The highlights mentioned at each stage cumulatively led to a step change in growth. The decisions and ideas detailed aren't particularly impactful alone, but combined created a new business opportunity. This case study is meant to explain some of the processes for building a successful growth team from scratch with the below being the primary takeaways.
0-3 months - Seed capital:
Take bold risks and focus on simple high impact experiments
3-9 months - Series A:
Build out measurement tools and let them guide the team to the most impactful work
Invest in visibility so peers can begin to rallying around the initiative
Ensure there is a plan to build a strong foundation so breakout growth is possible
9-20 months - Series B:
Take bets on new platforms and surfaces to attempt for a step change in growth
Begin harvesting the low hanging fruit in the rest of the growth funnel (e.g. activation, retention)
Reinvest and revisit tactics that are already working