As the leading grocery delivery service in North America, Instacart connects customers with personal shoppers who handpick items from local stores. Offering items from over 1400 retail partners across the United States and Canada, customers can enjoy same-day delivery or pickup, all conveniently through their mobile devices or computers.
INSTACART | 2023-2024
Creative Team
Philip Lowe, Creative Director
Tiffany Terra, Art Director/Lead Designer
Maxwell Pendergrast, Lead Motion Designer
Sofia Perez, Copy Lead
Embracing Dynamic Creative Optimization (DCO)
Dynamic Creative Optimization (DCO) revolutionizes digital advertising by using data and automation to tailor ad content in real-time for each viewer. By leveraging various data points—demographics, browsing behavior, location, time of day, and device type—DCO dynamically assembles and delivers personalized ad creatives that deeply resonate with individual users.
In the fall of 2023, the Instacart marketing org formed a dedicated strike team to explore and test the potential of DCO. Despite working with limited resources, we creatively repurposed assets from previous video and photo shoots to develop a flexible versioning strategy. This approach allowed us to scale personalization across multiple markets, demonstrating the immense value and impact of DCO in enhancing audience engagement and driving brand growth.
Phase I: Dynamic Canvas Testing
Our first step into DCO was creating a series of dynamic canvases to test. A dynamic canvas is variable frame that allows you to pull in assets such as copy, image or video from multiple sources.
As we were starting from scratch, the team went through several rounds of ideation for the first canvas test. The initial request from leadership was vague (as ad personalization tooling was new to the whole team) and there was no creative testing strategy in place.
Refining the process, we developed rules around fixed and flexible testing elements, narrowing down the key visual elements which needed to have a constant presence on-screen as well as the variables that we would test against each other. This led to a more standardized look and feel for the first iteration, but left room to experiment as we scaled our testing.
Fixed Elements:
Instacart logo should be present throughout the duration of the video and fixed at the top of any messaging hierarchy.
There must be a CTA, whether in a headline, button or badge (badges historically boost performance across our paid marketing channels).
Solid Kale background to support brand without overshadowing or adding extra noise to video content.
Flexible Elements:
3FD/Three free deliveries messaging
Regional retailer logos
City/Geo name
QR codes
Geo-specific promo codes/offers
Performance-Based Evolution
Several rounds of iterative testing produced our strongest performing template to-date, with an average CPIA (Cost Per Incremental Action) 20% lower than channel benchmarks. Audiences overwhelming responded to the creative format of placing a phone with messaging next to a branded video. This gave persistent and explicit visual cues: the service is an app, what retailers are included along with additional promotional messaging.
Phase II: Dynamic Videos + Retailer Localization
After getting the canvas testing up and running, we started Phase II of our personalization testing. We wanted to create modular videos where we could insert contextually relevant info for the audience, bringing personalization into their viewing experience.
Planning modules before moving to storyboards
Final Back to School storyboards/video flow
Expanding Back to School Messaging
This testing timeline aligned with the weeks after our larger Back to School 2023 campaign, continuing the narrative and driving audiences of parents to reuse the Instacart app throughout the school year.
Personalized Seasonal Testing
Leaning into seasonality, we also developed a fall-themed video format to test against our more evergreen control creative, which was still running in canvas format. We were hoping to get insights on whether or not these specific moments were worth developing creative for in the future, or if more evergreen videos performed better.
Planning modules before moving to storyboards
Final Apple Season storyboards/video flow
Seasonality + Retailers: A Winning Combination
After running these videos, we saw a large spike in conversions when showing local retailers vs the control/national retailers. Apple Season videos drove a 15% better CPIA (Cost Per Incremental Action) than channel benchmarks and outperformed our control, giving the team a good indication that switching up creative for holidays and seasonal moments was the right way to go.
Phase III: Building a Scalable Product
Following the success of our Back to School and Apple Season personalization tests, the team sought to create a dynamic video template with more testing variables baked in. Collaborating with the copy and motion design leads, I pitched an ambitious project aimed at targeting audiences during strategic times—just before meals and during craving windows.
We hypothesized that incorporating dynamic elements based on the day of the week and time of day, focusing on meal and snack windows, would enhance viewer relevance and boost brand consideration. Additionally, showcasing a preview of the app would demystify the user experience. This strategy not only offered a scalable opportunity for paid placements but also allowed us to pitch to CPG brands. By highlighting their products, we could not only drive sales, but also offset our advertising costs, and strengthen ad performance through strategic brand partnerships.
Planning modules before moving to storyboards
More complexity is added to the storyboarding process
Launching a Scalable Product
Not only did we engineer our most expansive DCO (Dynamic Creative Optimization) video to date, producing 50-60 unique versions with limitless potential for expansion through CPG partnerships and flexible image refreshes and prompts, we created a solid template for future tests to iterate off of.
Our streamlined approach, from storyboard to versioning, enabled us to complete the entire project in just two weeks—a remarkable achievement compared to the typical three-month timeline for similar initiatives. This efficiency not only showcases our ability to innovate quickly but also underscores our commitment to delivering high-impact solutions that drive growth and engagement.