Skip to content

Act

Actionable Recourse

What can I do to change this outcome?

Turns the factors behind an adverse decision into specific, realistic steps a person can actually take.

  • Evidence: research-based (Ustun et al. 2019, Wachter et al. 2017)
  • Tested: concept only, mock data

Overview

Why it matters

An explanation without a way forward is just a verdict.

People who understand why they were declined, but can do nothing about it, are no better off than people who were simply told no. This pattern turns the factors behind an adverse decision into realistic steps, and separates what a person can change from what they can’t.

The technically correct answer isn’t always a useful one. “Reduce your monthly debt by €400” only helps if that’s actually possible for this person.

What does it look like?

Loan applicant seeking a path forward

  1. Applicant receives a declined loan decision.
  2. Attribution panel shows top factors.
  3. Below attribution: “What you can do” section.
  4. “High impact: Provide income verification: your income could not be verified from the documents submitted. [Upload documents]”
  5. “Medium impact: Reduce monthly debt obligations below €550. Your current level is €680. This may take 3–6 months.”
  6. “Cannot change: The credit default recorded in 2021 is fixed and cannot be removed.”
  7. Applicant uploads income documents. Case re-evaluated in 3 days.

Employment screening challenge

  1. Applicant is flagged in a pre-screening process.
  2. Recourse section: “The following items were flagged and can be addressed: an employment date that could not be verified, and a certification that has expired. Upload a current employment confirmation and the renewed certificate to resolve them.”
  3. Applicant submits updated materials.
  4. Case moves to human review with updated information.

No actionable recourse available

  1. Applicant receives a declined decision.
  2. System determines that all contributing factors are fixed (credit history, existing defaults).
  3. Recourse section shows honestly: “At this time, the factors that affected this decision cannot be changed. The earliest you could reapply with a meaningfully different profile is [date range].”
  4. Applicant is not given false hope, but is given a timeline and contact for further questions.

When does it fit?

Use it when

  • The decision was adverse and the person wants a different outcome
  • Some factors are within the person’s control
  • The process allows re-application or re-evaluation

Skip it when

  • Nothing behind the decision can be changed: then recourse can’t exist
  • Showing the path would enable gaming without real improvement
  • The suggested path is so unlikely that it creates false hope

Design

How do I apply it?

  • Separate what can change from what can’t.

    A mixed list of factors is not recourse. It’s confusion.

  • Be specific.

    “Improve your credit score” is not a step. “Keep monthly debt below €600 for three months” is.

  • Be honest about impact.

    If one step alone won’t change the outcome, say so.

  • End every step with an action.

    A link to upload, resubmit or re-apply, not “go somewhere else”.

  • Don’t invite gaming.

    Point to factors that genuinely matter, not features that can be tricked.

For your technical team

The counterfactual a model produces (the smallest change that crosses the decision boundary) is not automatically actionable recourse. Filter suggestions for features the person can realistically change, and test that the suggested combination is feasible, not just mathematically sufficient.

How can I show it?

Lighter Richer
Content Block / List

Recourse action list

What you can change, what you can’t, what we can do.

Content Block / Section

Cannot change section

Fixed factors, shown honestly.

Inline Signal / Indicator

Impact indicator

How much each step would really help.

Inline Signal / Indicator

Time estimate

How long each step realistically takes.

Navigation & Flow / Flow

Document request flow

Upload and submit right where the step is shown.

Data Visualization / Timeline

Recourse timeline

The path from now to the next decision.

Start with the action list and the honest “cannot change” section. Add impact, time and flow as the decision allows.

What does it cost?

Transparency vs. gaming

Precise steps reveal thresholds. Where gaming is a risk, show the direction rather than the exact number.

Helpfulness vs. false promise

“Likely to improve your assessment”, not “will be approved”. Policies and models change.

Specificity vs. disclosure

Precise recourse means sharing scoring logic. Decide how much is right for your context.

Connections

Which patterns work with it?

Counterfactual Explanation

What would have changed the outcome?

Counterfactuals describe the decision boundary. Recourse translates it into action. Use counterfactuals as the explanation; recourse as the action plan.

Contestability & Appeal

How can I challenge this decision?

Contestability challenges the past decision. Recourse addresses the future decision. Both are forms of agency: they address different time horizons.

Attribution

Why did the system make this decision?

Recourse is built on attribution: the factors that drove the decision are the factors that recourse can address. Attribution is the input; recourse is the output.

Human-in-the-Loop

Who can review or override this decision?

Uploading documents or correcting data as part of recourse should trigger re-evaluation: either automated or via human review. Recourse and HITL should be connected.

Sources (3)

Ustun et al. (2019) — Actionable Recourse in Linear Classification

proposes recourse as a measurable, designable property of classification systems. Introduces the distinction between counterfactual changes and actionable recourse: changes that are actually controllable and realistic

Wachter et al. (2017) — Counterfactual Explanations without Opening the Black Box

the foundational paper connecting counterfactual explanation to legal rights under GDPR. Establishes that counterfactuals/recourse are a legal as well as a design requirement for consequential automated decisions

Karimi et al. (2021) — Algorithmic Recourse: from Counterfactual Explanations to Interventions

makes the critical argument that counterfactual explanations are necessary but not sufficient for genuine recourse. Recourse must also account for causal feasibility and user agency. The theoretical basis for the controllable/non-controllable distinction in this pattern

Sounds familiar?

You explain your decisions, but people still don’t know what to do next.

Moving from “we explained it” to “we showed a way forward” is where this pattern starts. Let’s look at your interface together.

Work with me