top of page
COVER.png

Internship @ Meta | Jun - Aug 2022 | 8 Weeks

Ads Manager: Info Label Adoption Improvements

Designing for increased adoption of info labels by advertisers on Meta’s Ads Manager platform.

Ads Manager is the central platform where advertisers can create, manage, and track their advertisements. As such, the platform accounts for 90% of Meta’s annual revenue.
During my internship, I worked on Meta’s Creative and Guidance Team, the team that designs and maintains the L1 template where advertisers actually create ads. Within Ads Manager, the L1 template accounts for 50% of all revenue from Ads Manager.

Intro - AM.png

Ads manager, L1 template

The page focused on ad creative experiences

Within info labels, I designed for 3 projects

Sticky Inputs for Manual Info Labels (Main)
Offers and Discounts Info Labels (Stretch)
CFP Adaptation of Info Labels (Stretch)

My Role.

Product designer

Design Lead

Responsibilities.

Product Strategy
Visual Design
Interaction Design
Proactivity and Drive
Intentionality

Team.

Product designer

Content designer

Project manager

Engineer

Data scientist

The Initial Experience.

Info labels allow advertisers to display business and account information on their ads.

L1 Template: Entrypoint into Info Label Modal

Today's Exp.png

Info Label Modal: Manual Info Labels

Manual Info Labels (AKA Custom) display information manually inputted by the advertiser

Currently, there are 3 types of manual info labels:
Shipping Policy, Payment Type, Return Policy

Info Label Modal: Auto Info Labels

Auto Info Labels (AKA From your profile) display information automatically pulled from their account or page.

This is currently defaulted-on in the new ad creation flow, resulting in high adoption.

The Problem Statement.

How can we increase adoption of manual info labels by advertisers from the L1 template?

Initially, I was given a simple, broad goal, with little direction and specificity:

To increase adoption of info labels.

My team and I subsequently discovered the opportunity space around manual info labels in particular by analyzing and understanding data and insights from info label and L1 data tracking, identifying possible areas of improvement for both advertisers and also business.

Discovery.

From my initial goal of increasing info label adoption, I worked with data science XFN partners to get more insight and knowledge in the problem space:

AR breakdown.png
Manual v. Auto Ads Revenue disparity

Manual info labels experience 5x less adoption than auto info labels, shown by the difference in ads revenue.

datavis breakdown.png
Little variation within manual info labels

The majority of manual info labels only use two variations, and over half have just one variation: advertisers don't typically change manual info labels between ads.

Low discove.png
Low discoverability of info labels

The key driver in this discrepancy is mainly due low awareness from a lack of default-on for manual info labels. The vast majority of Auto Info Label adoptions come from advertisers never even clicking into the Info Labels modal.

Definition.

With more understanding of the problem space from data science insights, I was able to narrow our problem statement to focus specifically on manual info labels. We can see that a lack of awareness of manual info labels due to no default-on contributes to low adoption.

I then led a brainstorming session with my XFN team to narrow down our product strategy and design direction.

We analyzed the tradeoffs and constraints involved with each idea to determine our design solution.

Brainstorming.png

Our solution

Sticky Values for Manual Info Labels

Exploration & Iteration.

To guide my design explorations, I broke the solution into 3 areas of consideration.

Activation.png

Sticky input activation.

How can users activate sticky inputs, saving their unsaved changes for future info labels?

Messaging.png

Messaging for sticky activation & unsaved changes.

How can we inform users of auto-populated sticky inputs or encourage them to save their unsaved changes?

Eye.png

Confirming use of auto-populated info labels.

How can we increase visibility into the info labels selected to minimize risk of mis-selection or applying the wrong info labels?

01

Sticky input activation

Initial explorations.

For the visual design of sticky input activation, I chose to use the checkbox component after looking into different activation-related components from the Geodesic design system.

Activation - Initial.png
initial exp activation.png
Ads Message Chip Popover.png

The main question I was exploring was “how can I encourage advertisers to opt-in to save their unsaved changes?”, which prompted me to do explorations about confirmation modals and disruptive flows.

UX design crit and XFN meeting feedback.

I took my initial explorations into meetings for feedback: at a Creative and Guidance DES team crit, and an Info Labels X-team meeting, recieving the following feedback points:

The confirmation modal isn’t suitable for info labels.
  • A modal as disruptive as a confirmation modal can lead to drop-off rather than adoption.

  • Confirmation modals should be reserved for larger and more important features; L1 Ads manager has many different features, and info labels are one of the smaller ones.

Can we make the experience as simple and easy as possible?

Confirmation modals and needing to click into sticky inputs requires both a lot of messaging / guidance and clicks for advertisers. How can we make the creative experience as seamless and low friction as possible?

Post-feedback design proposal.

Custom Info Labels - Sticky Opt-in 2.png
Key changes:
  1. Defaulting-on the Sticky Inputs checkbox and auto-saving changes.

  2. Moving the checkbox down to reduce likelihood of misconception + unchecking

02

Messaging sticky activation & unsaved changes

Initial explorations.

For my initial explorations, I focused on what in-product messaging can we use to highlight the feature for users.


I came up with two scenarios that could use messaging:

  1. Changes have not been saved, opt-in to save

  2. Values have been auto-populated (previously opted-in and saved changes)

ipd msg.png
Intial Exp - meg.png

UX design crit and XFN meeting feedback.

Too much messaging in the modal...

If we default-on sticky inputs, how necessary is in-product messaging?

Post-feedback design proposal.

Key changes:

With Default-on, we are choosing to remove in-product messaging altogether, as the auto-population is relatively low risk and intuitive. (The only exception is the Blue NUX).

Ads manager also faces an issue of too much guidance and in-product messaging.

Custom Info Labels - Sticky Opt-in 2.png

03

Confirming use of auto-populated info labels

Previous entry point

L1 - Prev.png

Design proposal

Using the L1 level to show info labels selected instead of a confirmation modal.

Our intent with this design proposal is to:

  1. Allow users to view the info labels they have applied to minimize legal risk

  2. Encourage entry into the info label modal through more visibility into options available

Start_new_1 1.png

The Solution.

Sticky Values for Manual Info Labels.

Start_new_1 1.png

Increased info label visibility from the L1-template entry point

From the L1 level, users can now see both the info labels they have selected and those still available without needing to enter the info label modal.

This decision not only reduces risk of mis-selection of info labels, but also increases adoption of manual info labels due to increased awareness.

Rules for Content.png
Custom Info Labels - Sticky Opt-in.png

Auto-populated values on entry to the info label modal

Custom info label values are automatically filled in upon entry (from previous ads) if the user has opted in to sticky values.

Upon making a change to the info label settings...

The sticky values checkbox will render and be activated by default.
We will also apply CTA changes for subtle in-product messaging.

CTA Rules.png
Custom Info Labels - Sticky Opt-in 2.png

The Impact.

From working with data science to predict the impact of our project, we estimate that sticky values for manual info labels will result in

a 5% increase in adoption of manual info labels.

This would give Meta a 0.0005% iRev lift.
With around $110B of annual revenue, this would be around half a million in additional revenue each year.

Key Takeaways.

How to communicate, collaborate and win buy-in for designs by involving others in my process.

Proactivity, communicating well, and making sure everyone on the team is aligned and has visibility into the process.

Working with XFNs, especially in Product Strategy and Roadmapping.

Collaborating with data analytics to understand the product strategy, and understanding how my project can help my team reach our business and impact goal.

Using Meta’s many design resources, and being receptive to feedback.

Taking advantage of the many design critiques and team meetings, design system support groups, and peers around me to make sure my work is as aligned with the Geodesic design system as possible (to pass design review :D)

Reflections.

I'm still astonished by how much I was able to experience and learn over the course of one summer at Meta. While I came in expecting to hone my skills in visual and interaction design (and that I did), I was surprised by the amount of product strategy, business analytics and team roadmapping I was able to practice. Being given such a broad goal for my main project allowed me to explore and practice my product strategy skills, something I'm extremely grateful to my team for.

The other biggest impact from this summer on my growth as a designer is learning how to communicate and collaborate with both my project and my design teams. In involving those around me in my process, I gain more buy-in and support from my teammates, helping me to pass design reviews and create greater impact. I'm very appreciative to my XFN partners for treating me as they would a full time designer and giving me responsibility over my projects.

Huge thank you's to Meredith Xie, Trevor Robertson, Crystal Ye, Shivani Singla, Dan Li, Hannah Berry, Shenxun Wang, Wilson Zhang, James Tilinghast and everyone else who helped me over the course of my internship!

Thank you for reading!

bottom of page