Designed to tackle broad challenges when defining and creating transparent documentation, adapt these to your context.
Incorporate broad perspectives on transparency into your Data Card.
Whether you’re an organization, a team, or collective, invite a wide variety of stakeholders from across the dataset lifecycle into your process to create truly actionable transparency.
Define dataset transparency with your team. Develop a deep and shared understanding through a series of critical discussions and reflections that will guide your data efforts.
Define the scope for your Data Card efforts. Outline the goals and vision, then define successes and how they may help your data transparency goals.
Review your dataset’s lifecycle to identify your stakeholders. Take note of who may interact with the dataset or its documentation. Consider how stakeholders may contribute to the Data Card(s).
Use meaningful criteria to drill down on your highest priority audience groups using 2x2 matrices.
Create personas and archetypes that describe your agents’ proficiency, as well as uses for datasets and Data Card(s) to reference when creating Data Card(s). Use our worksheets to conduct semi-structured interviews with individuals who represent your agents.
Treat your Data Card like a product and create user journeys for it. Determine what’s essential to convey to someone using the Data Card and set them up for success.
Identify the more obvious and expected themes in your Data Card(s) by voting on commonly documented lenses. Run a discussion on the long tail of voting results to unpack both agreement and disagreement.
A more nuanced version of the Agent Information Journey activity. Co-create lenses that you can take forward in the scopes brainstorm with stakeholders from across your dataset lifecycle for your Data Card(s) for thoughtful Agent Information Journeys.
An advanced and systematic way to create lenses. Brainstorm and check how well your lenses cover the lifecycle of the dataset(s) using the OFTEn Framework. Refine your lenses to address gaps and redundancies.
An advanced version of the Questions with Optics activity. Discover the variety of questions that agents want your Data Card(s) to answer. Systematically deconstruct lenses into scopes—questions at different fidelities that become the building blocks of your Data Card(s).
Rephrase lenses as scopes—questions at different fidelities that help agents establish accurate mental models of the dataset. Combine Telescopic, Periscopic, and Microscopic questions to emphasize structure and relationships in your Data Card template.
Organize your lenses and scopes into a Data Card template that can be inspected, tested, and implemented. Refine this template to capture stakeholder needs across your dataset’s lifecycle and support both producer and reader journeys.
Using accepted heuristics, introduce early telemetry to identify critical next steps to improve the usability of your Data Card template.
Use OFTEn to calibrate your template against data lifecycles. Identified action items towards a more utilitarian and representative template.
Take stock of your Data Card template and evaluate it on five dimensions for transparency. Uncover directional insights to improve your template.
Using your map of stakeholders in your dataset(s) lifecycle, add or remove questions, decide on trade-offs, and discuss how to write answers that agents can easily understand.
Use friction logs to test your template on representative datasete. Anticipate challenges and opportunities in your Data Cards scaling strategy.
Determine which answers can be automated, and which require human input or oversight.
Introduce guidelines and structures to help producers answer challenging, open-ended questions.
Review this checklist with your team before you fill out your Data Card to organize your documentation efforts, elist the help of stakeholders, and anticipate bottlenecks.
Use a table to determine the framing and nuance of your Data Card for your reader. Refer to this heuristically as you fill out your Data Card to stay on track or course-correct.
Use a set of heuristics designed to identify opportunities to improve the overall experience of your Data Card. Use these to identify common pitfalls and critical action items.
Systematically work through the benefits, limitations, and related applications of your dataset using attributes that are intrinsic to the dataset and contextual to the domain of application.
Use a fishbone diagram to work through different factors that can cause undesirable outcomes when your dataset is used.
Use this checklist to think through and plan for a variety of scenarios in which you will need to update your Data Card.
Use the five information types in this activity to articulate the specific value that your Data Card adds by comparing a completed Data Card to existing documentation and the Data Card template.
A checklist to ensure that your implementation addresses basic needs that can drive the adoption and use of your Data Card and the corresponding dataset.
This worksheet lists six common assumptions and their corresponding evaluation gaps. Use this table to audit a Data Card for possible gaps and remediative actions.
A template to track points where you encounter an issue or generally get stuck and feel frustrated.
Uncover directional insights using qualitative approximations and find immediate opportunities to start refining your Data Card or template.