Sarah Haas

Meta in-app browser Autofill

Save Payment consent experiment, with the framework extended to Use Payment

Facebook and Instagram, iOS and Android

Lead content designer and content strategist

Optimizing Autofill content for trust, clarity, and scale

Meta's in-app browser Autofill helps people complete checkout on third-party sites by saving contact and payment information and offering it again during later browsing and shopping sessions. People who had not opted in could see a Save Payment prompt after checkout; people with saved information could receive Use Payment prompts while completing form fields.

Research identified trust as a barrier to adoption, while an earlier UXR study found no meaningful difference among subtle messaging variants. I proposed a larger change and led content strategy for an experiment comparing a more visual, lower-text Save experience with a more explanatory treatment built around stronger trust signals. Because there was no single source of truth for the product history, Legal and Privacy guidance, content decisions, definitions, requirements, decision thresholds, or state-specific context behind the experience, I assembled the existing knowledge, resolved gaps with partners, and documented a shared baseline for future Save and Use Payment work.

Role, scope and partners

Lead content designer and content strategist. I owned content strategy for the Save Payment consent experiment across Facebook and Instagram on iOS and Android, with related research and the resulting framework extended to Use Payment, in partnership with Product, Product Design, UXR, Data Science, Engineering, Legal, and Privacy. I proposed the content variants for the initial study and partnered with UXR to integrate them into the research, then authored the Save Payment treatments, mapped how requirements changed across 16 Save variants, and extended the resulting baseline and review process across 11 Use Payment variants.

Goal

The product goal was to increase Save Payment acceptance and ultimately increase successful payment credential saves by comparing two approaches: more explanation with stronger trust signals, and less text with a more visual experience. My content goals were to determine how far the disclosure could be simplified while preserving required coverage, establish a clearer value proposition grounded in user needs, and create a reusable baseline for future Save and Use Payment changes.

Outcome

The work established an approved content baseline and Legal and Privacy review process that supported subsequent Autofill experiments. Individual treatments also produced positive secondary signals. The broader experiment did not produce statistically significant results against its primary goals, so the UX changes were not launched. The team planned to continue investing in content optimization as a UX quality priority, independent of immediate metric lift.

Results

Content outcome

I reduced content across the complete Save disclosure experience by 46%, from approximately 140 to 75 words, and shortened the primary disclosure by 58%, from 63 to 26 words. This reduced how much people had to process, brought the value and trust information forward, and made the decision easier to scan while preserving required Legal and Privacy coverage.

Experiment signals

The combined content and design experiment was neutral on its primary metrics. Individual treatments produced positive secondary signals, including a 2.2% lift in Save Accepts on Instagram iOS and a 1.8% lift in form completions on Facebook iOS. Because the content and visual changes were tested together, the results could not isolate the effect of either one. Neither showed evidence of hurting the primary metrics, but neither demonstrated an independent lift. A content-specific experiment was planned for later that year to evaluate optimized language on its own.

System outcome

The approved content baseline and review process gave Content, Product, Design, Legal, and Privacy a shared reference for subsequent Autofill experiments.

The challenge

Saving payment information is a high-trust decision distributed across a dense content system.

Saving payment information is a high-trust decision. People needed to understand what would be saved, where it would be stored, how it would be used, why saving it was valuable, and how they could manage or decline it. The required information was distributed across the main prompt, expandable details, and footer, making the overall decision harder to understand quickly.

The disclosure was not one static experience. The content and required coverage changed based on region, platform, security model, consent history, and user state, producing 16 Save variants and 11 Use variants. A clearer experience therefore had to work as a coordinated content system, not as a rewrite of one prompt.

Earlier research showed no meaningful difference among subtle messaging variants. Testing a more substantial direction required close partnership with UXR, Product Design, and Product Management so the content, visual hierarchy, and product behavior worked together while still supporting the product goal.

The Save experience had to answer five connected questions clearly enough for someone to make an informed decision.

Design thinking

Designing a clear decision and reusable content guidance.

I proactively pitched and scoped a UXR study to understand whether different content emphasis changed how people understood and responded to Save Payment. I proposed the research experiences and content variants, then partnered with UXR to integrate them into the study. The results showed little meaningful difference among the subtler approaches, supporting a larger content and design exploration.

I also structured the guidance so Content Design, Legal, Privacy, and Product could see which language was required, where it applied, and how much flexibility each element allowed. Teams could combine state-specific requirements with updated core content, making future changes easier to develop, review, and scale.

I combined the research findings with the broader Autofill audit, source documentation, Data Science and Product input, Legal and Privacy requirements, and my content analysis to establish five principles:

Lead with the value

Make the benefit of saving information for future checkout immediately clear.

Make trust visible

State how information is secured, stored, and managed in clear, prominent language.

Make the language actionable

Use headings, labels, and calls to action that clearly describe what will happen.

Strengthen the decision language

Help people understand what they are agreeing to and what control they retain.

Structure disclosure by priority

Keep the immediate decision concise while making required supporting details easy to access.

Using these principles, I defined the content strategy and authored content for two treatment directions with Product Design and Product Management. Product Design developed the final visual treatment for each direction alongside the content. One paired concise content with a more visual interface; the other used more explanation and stronger trust signals. Product Management connected the work to acceptance and credential-save goals, and UXR helped ensure the treatments answered a meaningful research question.

Together, the research, principles, and content guidance created a coordinated content strategy that could be tested and reused across the Autofill system.

The five content design principles used to restructure the screen around what people needed to understand before deciding.

What I did

Building a reusable Autofill content system.

  1. 01

    Reconstructed the baseline

    I gathered product documentation, source language, prior decisions, and Legal and Privacy guidance, then worked with partners to clarify why each content element existed, which requirements applied, and where the experience had room to change.

  2. 02

    Mapped the product states

    I mapped the logic behind 16 Save Payment variants, including regional, security, consent, and user-state differences. I later applied the same model across 11 Use Payment variants, identifying the shared content foundation and the requirements triggered by each state.

  3. 03

    Authored the experiment treatments

    I deconstructed the main prompt, expandable details, calls to action, and footer, then authored the content for the more visual and more explanatory directions. I adapted both directions across the required Save states while preserving a coherent value proposition, trust model, and interaction language.

  4. 04

    Created a repeatable review process

    I built a cross-variant review matrix that showed the control and proposed treatments across applicable product states. This gave Content Design, Product, Design, Legal, and Privacy one place to review changes, document decisions, and reuse approved language in subsequent Autofill experiments.

The cross-variant review matrix carried the content strategy into a testable comparison across the Control, Variant A, and Variant B treatments.

The solution

A clearer consent experience and a stronger foundation for future work.

  1. 01

    Optimize the content for a clearer decision

    I rewrote the headlines, supporting copy, calls to action, and disclosures in direct, active language. The content led with the benefit, made security and data use more explicit, clarified the controls people retained, and described each product state more precisely. I also removed repetition and replaced broad language with more specific guidance where requirements allowed.

  2. 02

    Reduce cognitive load through hierarchy and structure

    I shortened the primary and complete disclosure experiences and prioritized what people needed to understand at the consent moment. Supporting requirements were organized within the existing disclosure layers, preserving the information people needed to assess value, security, control, and data use.

  3. 03

    Document a new decision and approval baseline

    I documented the approved language, requirements, rationale, and conditions that applied across Save Payment states. Legal and Privacy could reference previous decisions and focus future reviews on what was new or changing. When appropriate, teams could inform them of an update using the approved baseline rather than restarting the full approval process.

Together, the optimized language, clearer hierarchy, and documented baseline made the consent decision easier to understand and created a more efficient foundation for future Autofill work.

The word-count impact across the disclosure, expanded details, and footer.

Reflection

Document the full decision

A durable content record includes the decisions, requirements, approvals, rationale, and final language. Memory fades, product ownership changes, and future partners should have the context ready when they need it. That record also makes Legal and Privacy approval easier by showing what has already been decided, what requirements apply, and exactly what is changing.

Make it easy to say yes

Content quality depends on advocacy. I brought Legal and Privacy into the work early to understand their existing guidance, identify what context they needed, and establish a shared starting point. I then prepared complete review packages that made the requirements, rationale, options, and proposed changes easy to evaluate. The process gave each partner a voice in the decisions while keeping the work organized and moving toward a shared, defensible outcome. For a global, revenue-driving product with significant compliance requirements, that made it easier for everyone to understand the changes, weigh the tradeoffs, and say yes.

Advocate for quality beyond the metrics

Content optimization will not always produce enough metric movement to unlock further investment. Clear, reusable consent content aligned with Product, Legal, and Privacy guidance is still good UX. Content Design connects user needs, product goals, system behavior, and regulatory requirements. Part of the role is protecting and advocating for the user experience when immediate results do not tell the whole story.