Overview

What makes an ad

Everyone knows what an ad is, but most people probably don’t understand the complex hierarchy and settings that go into creating an ad. Because advertising is a complex world with lots of unique terminology and structures, that can even differ platform to platform, I wanted to provide some background information that would make this more understandable.

Components

Campaign: Campaigns are the highest component in the advertising hierarchy- they include groups of line items.  They are normally used to set goals for advertising campaign.  Some common types of campaigns are based around events (super bowl, back to school), holidays/seasons (Halloween, Spring), new products, and sales.  To promote relate products

Line Item: Line items are where the user will set the duration, targeting options, and budget of the campaign.  Normally within a campaign the user will create different line items and to try different types of “attacks” and based on what performs best will add or detract budget from different line items.  

Creatives: A creative is the set of visual ads the user sees.  The individual ads will differ by size and where they appear on the site, but within a creative the ads will have a consistent theme. Ads on walmart.com consist of text, images, and CTA’s.  These are associated at the line item level, but can be recycled and used on different line items within or even outside the same campaign.  

Users

The users of DSS would be any company that sells products through walmart.com. Internally we refer to these companies that create ads as advertisers. Within an advertiser there will be different types of users depending on the division of labor within the company. Depending on the company there can be separate users for creating campaigns/line items versus creatives. Additionally, many large corporations will contract out their marketing work to agencies, sometimes even having a different agency for each part.

Platform

The Display Self Serve (DSS) platform consisted of a dashboard and three main user flows.  The dashboard shows the hierarchy of campaigns, line items within campaigns, and the creatives associated with each line item.  It also shows some of the most important statistics around campaigns and line items: dates, budget/impressions (advertisement views) delivered, and pacing (is the correct number on impressions being delivered each day). The three flows are the campaign creation flow, the line item creation flow, and the creative creation flow otherwise known as the Creative Builder.  

Design Process

When I joined Walmart Connect they were preparing for the launch of DSS. The plan was to have two internal tests, followed by an external demo, an alpha release, and a beta release, before finally releasing to general audiences. For each of these phases my process was as follows:

1. QA Testing: Before any user testing, myself and the product team would test the product to find most bugs and test common and edge use cases. I would additionally, prior to the general QA testing, perform a VQA to make sure the site matched the approved mocks.

2. User testing: It was very important to me that we not only collect feedback but conducted user studies. I would conduct Think Aloud Protocol (TAP) with select users to watch them interact with the platform to identify any pain points. Additionally I would work with PM’s to lead focus groups to collect feedback at the different stages of the campaign creation.

3. Synthesizing of Feedback: I would look for patterns from the pain points noted in the TAP sessions to identify UX issues. Additionally I would collect all the feedback (across all sessions) into an excel document organized by flow/page and issue type (visual, UX, copy, feature request, bug). I would also track which users would request which changes, which was especially important when advocating for a new feature or a UX redesign.

4. Discussion of Next Steps: At this time the product team, engineering, and myself would discuss what initiatives would be put forward for the next release. This was heavily based on the feedback log I created.

5. Design and Iteration: After going over feedback and deciding what the next design changes and features would be, I would create the new designs first sketching them out and then building them out using our style guide and component library. I would iterate on these with the feedback from PM’s, engineering, and designers on other teams.

New Features

Throughout my time working on DSS, I contributed heavily to the product by designing new features as needed. The need and prioritization of these features was backed up by the research we had done with users. For each feature, I worked extensively with engineering and product to make sure that the specs and workflow were clear and that the feature was ultimately worked as it was supposed to.

Keyword Targeting

Problem: One of the most common types of targeting that our users expected to be able to utilize on DSS was keyword targeting. This was type of targeting allowed users to target ads by search keywords. However, for our first iterations had only the most simple targeting options. Some issues to consider are that users can input hundred of keywords, and assign a match type to each keyword (broad or exact). Also, there are some keywords advertisers might not be able to target.

Solution: I designed a keyword targeting tool that allowed users to add target keywords and negative keywords. The tool allows users to put in 100-200 keywords at a time, and do bulk actions so save users time. I also designed an internal tool that allows our operations team to reject or approve these keywords, and well as add their own. Some important features include collapsible sections for groups of keywords, to make finding keywords easy, keywords search, and error messages for different keyword submittal issues.

Macro Contextual Targeting

Problem: With the pre-existing contextual targeting feature, advertisers were having issues meeting their impressions quota. They also wanted recommendations for categories to target. On the Walmart side, we wanted to incorporate a tool that our internal technicians used, which used ML to expand to related categories. However, the tool was very complex and took specialized training to use and we wanted to translate this into something intuitive and non-expert friendly.

Solution: I looked into our existing internal tool to see how it works and developed two different prototypes with differing levels of user control and complexity. We then A/B tested the two version for user comprehension and preference. We found that all advertisers we tested with not only understood the functionality of the more advanced version but also preferred the level of control. The final version allowed the user to turn on the related categories and select the reach they wanted.

Audience Targeting

Problem: Another type of targeting, that we originally did not offer, but that was also important to our advertisers was audience targeting. Audience targeting allows our users to target groups of customers based on certain characteristics. Because of the engineering heavy work this type of targeting would require we needed to start with an MVP version before working our way up to the full scope of audience targeting.

Solution: Based on advertiser feedback we picked the initial options for audience targeting which were a primary selection if they wanted to target customers by their brands or by category, and a secondary selection between likelihood of purchasing versus purchase history. We also added in custom audiences which advertisers could request very specific parameters and have that specific audience made for them to address the advertiser need for more complex options.

Tags/CTA’s/3P Impressions

Problem: Advertisers to use third-party companies to get metrics about the performance of their ads. Walmart does have their own reporting center, but some companies prefer to view these metrics from a neutral third party. One issues I had to consider is that advertisers could potentially use different url's for each ad within a creative, or use the same url's throughout the creative. Also, all url's had to be checked by different API's to make suer they were correct/not malicious. Not only could these API's take up to 15 minutes to return a result but custom warnings had to be given to the users based on which API call the url failed. Additionally, this feature was originally slated to be a part of the creative builder, didn't make sense from a UX perspective as it was most likely different users would be doing these different tasks, and that these would be tasks done at different times of the advertisement setup.

Solution: I used advertiser an internal feedback to advocate for a move of the feature from the creative builder to our line item setup flow. I designed the page for third-party urls to include warnings that would let the user know if there were any issues.

Notification Center

Problem: There are several issues that can delay or cause issues when serving ads. Right now there is no in app way to communicate these to our advertisers, and we rely solely on email. However, these emails can get lost in a sea of other emails users receive, and it is important to communicate these issues through the platform itself.

Solution: I designed a notification center that allows the user to view all the notifications they receive. Notifications that will negatively affect the flighting of a campaign are labeled with a warning symbol, so the user can quickly pick out the most urgent notifications. Additionally new alerts are labeled with "NEW" and bolded. Users can chose to delete notifications and turn on/off email notifications.