What should I do when two options score almost the same?

Call it a near-tie before trying to break it

A tiny difference between totals usually means the scoring exercise has not established a meaningful preference. Keep the totals, but stop treating the larger number as automatically decisive. First confirm that both options still meet the objective and every real constraint. An option that violates a non-negotiable does not become viable because other preferences compensate for it.

The University of Minnesota Extension criteria-grid guide says the highest total is probably the best choice, not certainly the best. When choices are close, it suggests weighting or ranking criteria; when options tie, it also suggests considering whether they can be used together or combined. Those are prompts to inspect the choice, not evidence that every tie should be forced into a single numerical winner.

Run the NARROW check

NARROW is a Forever Free Compass framework, not an external research finding. Use it on the existing comparison rather than rebuilding the scorecard until one option wins.

1. Non-negotiables: verify that each option meets every hard constraint, permission, duty, affordability limit, and safety boundary that applies.
2. Assumptions: circle the one uncertain score, forecast, or criterion weight most likely to affect the ranking.
3. Range test: move that input through a plausible low and high value while holding the rest of the comparison steady.
4. Read the trade-offs: set the totals aside and state where each option is meaningfully stronger or weaker at the criterion level.
5. Optionality: ask whether the options can be sequenced, combined, sampled, delayed for one decisive fact, or tested through a reversible step.
6. Write the next move: record a robust choice, one bounded test, or an explicit judgment call and the reason for it.

Check constraints before preferences

Separate pass-or-fail boundaries from preferences that can trade off. A legal requirement, another person's consent, a firm budget ceiling, a required qualification, or a material safety condition should not be averaged against convenience or appeal. If one option fails a genuine boundary, record that failure and remove it from this low-stakes comparison.

If both remain viable, verify that the criteria describe differences that actually matter to this decision. HM Treasury's multi-criteria guidance says criteria should cover key performance aspects and be phrased so the relative merits of options can be considered. Adding more criteria simply to manufacture separation increases effort and can hide the real trade-off.

Stress-test one uncertain input

Choose the weakest decision-relevant input: perhaps an uncertain outcome estimate, a subjective performance score, or a criterion weight that could reasonably differ. Change only that input across a plausible range and watch whether the ordering changes. Do not vary several values at once or choose an extreme value merely to produce the answer you prefer.

HM Treasury's 2026 Green Book says sensitivity analysis tests how changes in key assumptions affect results and uses switching values to identify when one option becomes less attractive than another. Its supplementary multi-criteria guidance says the purpose is to test whether the top-ranked options remain stable under different scoring or weighting assumptions. For an everyday choice, a simple range test can reveal whether the near-tie is robust or fragile without pretending to be a formal appraisal.

Read the trade-off that the total compressed

Put the two rows side by side and finish two sentences: “Option A is stronger when ___ matters more,” and “Option B is stronger when ___ matters more.” Then name any meaningful consideration the grid did not capture, such as an effect on another person, uncertainty about access, or a value that would be distorted by a numerical score.

The Green Book says an initial ranking by a summary metric is not sufficient on its own and that balanced judgment also considers objectives, unmonetised effects, risk, uncertainty, financial impact, and distributional effects. This page adapts that principle narrowly: the total is a summary of your stated inputs, not a complete or objective measure of what the option is worth.

Choose the next move that matches the result

If the same option leads across plausible input values and the criterion-level difference matters, choose it and record why. If the ranking flips easily, keep both options viable. You can gather one decisive fact, run a bounded reversible test, choose the option with a genuinely relevant reversibility or learning advantage, sequence the options, or make an explicit judgment based on the trade-off. Do not quietly add reversibility as a tie-breaker unless it belongs in this decision.

If two options can be combined or used in sequence without crossing a boundary, that may be better than forcing mutual exclusivity. If they cannot, a judgment call is still a valid outcome. Record which value, trade-off, or practical condition decided it so the choice can be reviewed honestly when new evidence appears.

Raise the standard when the stakes are high

NARROW is for low-stakes personal reflection. It is not a substitute for formal financial analysis, clinical decision support, legal advice, workplace authorization, engineering review, procurement appraisal, or another person's consent. A simple scorecard may omit dependencies, unequal effects, correlated risks, and obligations that require specialist methods.

When the choice could materially affect health, legal rights, regulated money, employment, confidentiality, immediate safety, or another person's welfare, pause the informal tie-break and obtain the required permission or qualified guidance. A small difference between totals does not make a high-stakes decision low-stakes.