Dynamic Product Ads

The Problem

Large retail/ecommerce advertisers hire Twitter to acquire new customers and maximize the value of existing customers. Our target customer wants to drive efficiency in Cost Per Purchase and Return on Ad Spend (ROAS), leverage their diverse customer base and product catalogs which range from thousands to millions of products, to create a scalable and sustainable way to grow their business. Performance marketers are also busy and want to spend the minimum time necessary to achieve results.

Lead the design and strategy of how to design and introduce Dynamic Product Ads onto Twitter, with a primary focus on our customer-facing experience. This involved developing knowledge on how advertisers organize their product catalogs, how our back-end systems functioned, and how our users on Twitter desire to interact with shopping ads.

The Process

Dynamic Product Ads are the perfect example of a product that seems simple to the consumer, but is very complicated to build and design for any and all use cases for it. As a means to gain alignment across design, engineering, product, and data, I helped lead a workshop that aimed to unify our team so our approach could be quick, effective, and founded on team buy-in.

Workshop Results

Out of this workshop, we started to align on a few different elements that we wanted to prioritize pursuing. This included concepts such as price, ratings, inventory updates, social proof, CTA buttons, sale ending, etc. We generated a lot of great ideas, but I noted that if we pursued them all at once, we wouldn't be able to properly understand what was and wasn't working. We would simply be putting far too many variables in at once to get a meaningful result.

Coming out of the workshop, I was able to iterate upon each individual idea, as well as develop a plan for how they might all dynamically work together (e.g. Price + CTA, Sales Ending + Rating, etc.) Working alongside our Product Managers, I proposed a phased experimentation plan so we could get cleaner data, while simultaneously having a plan to continue to iterate on the best performing designs.

Beta Results

We partnered with a handful of companies to participate in our beta test, such as New Balance, Cole Haan, Home Depot, Office Depot, Zenni Optical and more. The results of our beta experiment were extremely positive, with the intention to launch to GA in Q4 '22. With this launch will come further iteration upon the format to experiment with added features such as inventory updates.

For 4/4 SVO (small value order) experiments, DPA Ads outperformed non-DPA ads across the board with 30-88% lower Cost per Click Through Purchase, and 6-50% lower Cost per Site Visit.

For 4/4 WCO (web click optimization) experiments, DPA Ads outperformed non-DPA ads across the board with 24-74% lower Cost per Click Through Purchase.

Zenni Optical

Cole Haan