UX | Data Privacy | Case Study
Everything You’ve Ever Agreed To (and Never Read) — A Case Study
How a team of University of Washington graduate students reimagined the way we read the Terms & Conditions.
People care about their data and digital privacy. Yet, almost no one reads the Terms and Conditions that dictate how their data is used. These documents are long, dense, and designed for lawyers, not the people agreeing to them.
For our HCDE 518: User-Centered Design (Fall 2025) course at the University of Washington, we set out to reimagine that moment of consent by designing a browser extension that makes the fine print simple, visible, and easier to act on.
The Team of Secret Agents 🕵️
Meet Termsly: Your Privacy Companion
Termsly is the outcome of a 10-week case study culminating in a proposed browser extension. This extension summarizes and grades a website’s Terms, giving users a contextual understanding of clauses at the level of detail the user chooses.
The Privacy Paradox and Our Research
While users claim to be very concerned about their privacy, they nevertheless undertake very little to protect their personal data.¹
While users are often skeptical about how their data is handled, most still accept the terms without reading them and continue using the products anyway.
Users accept the inherent tradeoff between sharing personal data to access services. The breakdown occurs when users feel that these services are overreaching or that providers intentionally obscure the extent of their rights (e.g. AI model training, targeted ads, etc.). Yet, this kind of overreach is often disclosed — at least technically — in the Terms of Service (ToS), Terms and Conditions (T&C), or privacy policy. For simplicity, we will refer to these collectively as Terms in this project.
The research makes one thing clear: Terms are still written in ways that make them difficult to understand or easy to ignore.
Survey Says…
From our survey with 41 respondents, most said they value transparency but find Terms long, confusing, and discouraging to read. One participant summed it up simply:
“They’re designed in a way to make you not want to read them.”
There was also a shared sense of frustration around fairness and accountability. Several respondents felt companies benefit disproportionately from user data. One wrote:
“If my data is being collected and used for the profit of corporations, it would be fair for the corporations to have to pay for the use of my data.”
Experts revealed the challenges inherent in data privacy disclosure
Elements of consent: freely given, specific, informed, unambiguous, and revocable. — Daniella A. Kim
We interviewed experts across academia, nonprofits, and the tech industry, who helped us see how messy “consent” becomes once business, UX, and technology collide. A few themes stood out:
- Business vs. user benefit. Regulatory compliance often drives how companies design their privacy flows, which can conflict with what would actually help users understand and control their data.
- Tradeoffs and “spiky” interests. People tend to accept data collection as a tradeoff for useful services, but have one or two non-negotiables (like location tracking) that can become instant deal breakers.
- Design as a lever for trust. Visual layout and interaction details, such as clear structure or simple animations confirming a data deletion request, can significantly improve comprehension and users’ confidence that their choices are honored.
Learning from other solutions: Competitive Analysis
To understand what it actually feels like to read Terms — and what existing tools do (and don’t) fix — we combined a “Try It Yourself” reading exercise with a competitive analysis of Terms of Service; Didn’t Read (ToS;DR). We sat down with the Terms for Meta, Spotify, and Otter.ai and tried to read them as other users might.
Spotify alone had nearly 15,000 words just in its Terms and Privacy Policy, took about an hour to read, and hid important details inside dense legal text. It made the cost of “just reading the Terms” very tangible.
Existing solutions like ToS;DR and the Usable Privacy Project are effective in explaining the Terms more simply. However, these tools require navigating to a different site and rely on manually updated (and therefore out-of-date) policies. This highlighted the gap in the market for an in-situ, glanceable Terms assessment that stays current and always traces summaries back to the source.
Research Triangulation… or Quadrangulation?
We pulled together key findings from our research and distilled them into five themes that described concrete paths for improving the Terms experience:
- Educate and Inform. Help users understand what they’re agreeing to in plain language.
- Power to the User. Give people real choices and control over how their data is used.
- Less Frustrating, More Comfortable. Reduce friction and emotional fatigue when encountering Terms.
- Less Boring, More Engaging. Make the experience feel worth paying attention to, not something to rush past.
- Pressure on Providers. Encourage government and companies to write better Terms and be more accountable.
These themes became the bridge from research to design. They led us to our guiding design question:
How might we help end users comprehend a product’s Terms more clearly, understand what happens with their personal data, and empower them to make the best choice?
Our Personas and Design Goals Emerged
Regina represents users who understand online risks but have grown numb to them, experiencing consent fatigue and rushing through Terms to avoid friction.
Oliver represents older users who don’t fully understand the consequences of their online actions, and accept Terms because they “have to.”
Get Termsly Secret Agents’s stories in your inbox
Join Medium for free to get updates from this writer.
Along with our personas, we set 5 design goals:
- The design should be concise and use simple language.
- The design should provide comprehension of the impact of the Terms.
- The design should instill confidence through transparency: Users should find our solution trustworthy, accurate, and complete.
- The design should be contextual.
- The design should be personalized and personable.
To the Drawing Board ✍️
We ideated and sketched over 40 ideas, then affinity mapped commonalities in our approaches. We organized these sketches by the user journey touchpoints, pulling out common traits we gravitated towards.
The Moment of Decision
Using our sketches, we identified our key touchpoints:
- Before agreeing to the Terms (i.e. signing up for a service).
- How are users affected by the Terms (e.g. a victim of a data leak).
We refined into these three ideas accordingly:
1. Summarization Overlay Plugin
A browser extension that provides AI-generated summaries of key Terms points like data sharing and tracking. This reduces reading effort and learns a user’s priorities over time to surface the most relevant highlights automatically.
2. External Device Summarization
An AR or phone-based tool that overlays key clauses on scanned screens or paper agreements. It can translate legal or foreign language into plain text, making agreements more accessible across mediums, languages, and literacy levels.
3. “Terms Wrapped” Visualization
Inspired by Spotify Wrapped, a story-style recap that shows a user’s consent activity over the past year: accepted Terms, data-sharing graphs, and tracking across companies. It reframes an invisible, legalistic process into a personal narrative that informs and empowers.
Our Refined Direction
We determined from our research that users preferred a browser plugin over an external device; relying on a device causes too much friction for consistent use. Additionally, Terms are often a click away for sign-up flows, making the plugin more accessible.
With our touchpoints in mind, we chose the Summarization Overlay Plugin and the Terms Wrapped to build upon.
The plugin would help users at the moment before they agree to the Terms, while the Wrapped provides reflection on their data privacy standing after already having signed up for services. Our solution would educate and empower users in multiple contexts, demystifying the Terms.
Prototyping and Usability Testing
Using RITE (Rapid Iterative Testing and Evaluation), we advanced low-fidelity wireframes to a final mockup. As we observed users and received feedback, we were able to correct confusing interactions, fine-tune our wording, and create a product that felt intuitive and reassuring.
Some key insights from our testing:
- 💡 Users wanted the signal displayed directly on the Terms page ⇒ ✅ Embed ratings on the webpage and extension button.
- ❌ Letter grades were hard to understand ⇒ ✅ Use smileys for more intuitive ratings.
- ❌ Users couldn’t see ratings before opening ⇒ ✅ Display rating on the extension button.
- 💡 Users skimmed bold headers and missed details ⇒ ✅ Simplify UI for “at a glance” comprehension.
- ❌ Users needed clearer links between summaries and full Terms ⇒ ✅ Show full Terms with contextual highlights and explanations.
- 💡 Testers focused on a few key “dealbreaker” issues ⇒ ✅ Add customizable settings to prioritize what matters most.
- ❌ Users preferred a pop-up over the sidebar ⇒ ✅ Use a pop-up design that feels integrated.
- ❌ Users were confused where the website ended and the extension began ⇒ ✅ Improve branding to clearly show the popup belongs to the extension.
Final Product: Termsly
Summarization Plugin
In our final prototype, Termsly appears as a pop-up over the site you’re visiting. At the top, a single mood-based face and label (“Bad”) give a quick read on the overall Terms for an at-a-glance assessment.
Below that, plain-language summary items (e.g. “Minors are still tracked.”) are grouped into sections, and each has its own face rating for fast scanning. To drill down, click an item to view the Terms on the right and highlight the exact clause, showing a small explanation popup and making it easy to move from a high-level summary into the specific legal text when needed.
We created a simple Help view to explain the rating system. The Settings view lets users personalize Termsly by adjusting feature toggles and ranking priority categories — such as “Data They Collect” and “Purchases & Billing” — so summaries focus on what matters most to them.
Terms Wrapped
Terms Wrapped turns a year of inattentive “agrees” into a colorful recap. It shows how many Terms you accepted, where your data travels, what companies have learned about you, and how long reading everything would have taken. It also points out opportunities to revoke cookies or data access, helping you understand your footprint while giving you ways to take back control.
It’s Showtime 😎
On December 1st, 2025, Termsly debuted at the HCDE 518 Showcase. One of our favorite moments was a visitor who exclaimed “Wow! I would love to use this!”
Other feedback we received throughout the process:
I love this pop up. I want this. Make this please.
The “TOC wrapped” is super clever and fun
Such a good idea, simple and well executed.
i really like the coloring and short description of the specific details! as someone who doesn’t really ever read the terms, it’s nice to see it broken into a much more digestible form compared to all the jargon its normally filled with
omg the wrapped is so cute and a nice way to engage the users more/create more motivation while encouraging them to use your tool
Takeaways
Lessons Learned
Advantage of fidelity in prototyping and sketching. Although we learned about tuning the fidelity of your artifacts for the kind of feedback you’re seeking, we didn’t appreciate it until receiving really amazing (and product-shaping!) feedback from our peers as we rapidly iterated with lo-fi prototypes.
The limitations of a third party to solve this issue. Our research continually reinforced that governments and companies should — but fail to — play a role in protecting users’ privacy. Legislation has made progress in recent years, but there’s still more to do. We hope Termsly encourages these entities to construct better terms for users.
Future Considerations
Designing Termsly surfaced a lot of ideas we didn’t have time to fully build but feel like natural next steps for the product. In the future, we imagine:
- Bringing Termsly to mobile. Many “agree” moments happen on phones, not laptops. A mobile version (or OS-level integration) would guide users while they tap through the prompts.
- Stronger tools to revoke access. Users wanted to move beyond awareness and actually cut specific ties, such as third-party data sharing, from a single place. Termsly could offer clearer paths to revoke or limit that access.
- More human oversight on AI analysis. As the system scales, we see value in layering expert review, clearer confidence indicators, and explicit flags for ambiguous or hard-to-classify clauses. Users should understand both what the system knows and where it’s less certain.
We’d also like to spend more time defining the visual language and brand for Termsly. Most of our effort went into research, flows, and interaction patterns. A future iteration would refine the look and feel to better match the tone we’re aiming for: calm, approachable, and trustworthy.
Try it Yourself: Our Prototype
If you have any questions, feel free to contact any of the team members: Anna Czoski, J.P. Nguyen, Katarina Shewchuk, and Tyler McBride. Thank you for reading our case study!
Acknowledgements
We would like to thank our 41 survey participants, our 6 usability testers, Tracy Kosa, Tiffany Chen, Kasia Chmielinski, Stephen Hamilton, the entire class of HCDE 518 Section C, and — finally — our instructors, Daniella Kim and Lubna Razaq, for all the support throughout this project!
References
¹Barth, Susanne, and Menno D. T. de Jong. “The Privacy Paradox — Investigating Discrepancies between Expressed Privacy Concerns and Actual Online Behavior: A Systematic Literature Review.” Telematics and Informatics, vol. 34, no. 7, 2017. ScienceDirect, https://doi.org/10.1016/j.tele.2017.04.013.
Mousavi Nejad, N., Jabat, P., Nedelchev, R., Scerri, S., Graux, D. (2020). Establishing a Strong Baseline for Privacy Policy Classification. In: Hölbl, M., Rannenberg, K., Welzer, T. (eds) ICT Systems Security and Privacy Protection. SEC 2020. IFIP Advances in Information and Communication Technology, vol 580. Springer, Cham. https://doi.org/10.1007/978-3-030-58201-2_25.
Ibdah, D., Lachtar, N., Raparthi, S.M., & Bacha, A. (2021). “Why Should I Read the Privacy Policy, I Just Need the Service”: A Study on Attitudes and Perceptions Towards Privacy Policies. IEEE Access, PP, 1–1. https://ieeexplore.ieee.org/document/9624976/.
⁴Aly, H., Liu, Y., Anaraky, R.G., Khan, S., Namara, M., Byrne, K.A., & Knijnenburg, B.P. (2024). Tailoring Digital Privacy Education Interventions for Older Adults: A Comparative Study on Modality Preferences and Effectiveness. Proc. Priv. Enhancing Technol., 2024, 635–656. https://www.semanticscholar.org/paper/Tailoring-Digital-Privacy-Education-Interventions-A-Aly-Liu/1f250fc3a07334dfd2a09146edaf27f4ff1f66a9?utm_source=direct_link.
K. C. Chang, R. Nokhbeh Zaeem and K. S. Barber, “Is Your Phone You? How Privacy Policies of Mobile Apps Allow the Use of Your Personally Identifiable Information,” 2020 Second IEEE International Conference on Trust, Privacy and Security in Intelligent Systems and Applications (TPS-ISA), Atlanta, GA, USA, 2020, pp. 256–262, doi: 10.1109/TPS-ISA50397.2020.00041. https://ieeexplore.ieee.org/document/9325370.
Zhang, Shikun, Yuanyuan Feng, Yaxing Yao, Lorrie Faith Cranor, and Norman Sadeh. “How Usable Are iOS App Privacy Labels?” Proceedings on Privacy Enhancing Technologies, vol. 2022, no. 4, 2022, pp. 204–228. De Gruyter, https://doi.org/10.56553/popets-2022-0106.
Lo, Kyle, Joseph Chee Chang, Andrew Head, Jonathan Bragg, Amy X. Zhang, Cassidy Trier, Chloe Anastasiades, et al. “The Semantic Reader Project: Augmenting Scholarly Documents through AI-Powered Interactive Reading Interfaces.” arXiv preprint, arXiv:2303.14334, 2023. arXiv, https://arxiv.org/abs/2303.14334.
