User-focused architecture

The problem

The creation of an ads campaign consists of a number of interconnected steps. The sequential progression from step #1 to step #2 and so on feels very intuitive, as it allows user to focus on one delineated task at a time before moving on to the next. This core principle has been a cornerstone for creation flows on ads platforms such as Google, FB and Amazon for many years, where an experienced multi-platform advertiser has had long to get acquainted and accustomed with this model. This was strategically smart for these platforms.

However, eBay is neither of those platforms and we needed to think of a model best befitting the majority of our specific user segments, a substantial portion of which were large C2C and SMB advertisers (>40% of revenue). With that, I needed to design a more user-friendly model that would not over-index on a specific ads acumen, but rather on efficiency across all seller groups on eBay.

The erratic decision-making chain

Initial steps of this flow require nothing but a working assumption. Advertiser makes strategic decisions (scheduling and budgeting) based on a hypothesis. It is only in the subsequent steps important bits of factual data begin to surface. These commonly include ad group guidance, keyword data and match type insights. This guidance, being placed towards the end of the flow frequently challenges the assumptions that had already been made earlier in the flow. To factor them in and to course-correct, a seller is forced to retrace their steps through multiple pagination steps they had already completed.

As a result, the crisp linearity of creation a campaign is only effective in one direction. In reality, such experience is full of multiple u-turns and cross-corrections. For example, on step #4 a seller might realize that based on the keyword cost they’re seeing, their budget (on step #2) needs to be adjusted accordingly. In turn, these changes may force them to also reconsider their Ad Group strategy (step #3). The otherwise linear process quickly becomes erratic and laborious. It also breaks a number of UX laws, specifically the Miller Law that acknowledges the limitations of working memory in users.

The solution for eBay (launched in 2019)

I have designed a more efficient model for complex, multi-step campaign flows. It takes forward the crisp delineation of tasks that traditional models share, but it also locks the completed steps to the single surface.

Key advantages

• Linearity of overall flow keeps seller focused on a single sub-task at a time.
• Progressive disclosure of upcoming steps upon completion (and backend validation) of completed steps. Real-time validation also minimizes the number of compiled errors.
• Fast and easy access to earlier completed steps without losing mental track of collective campaign parameters. Seller scrolls up the page, rather than paginates backwards.