A note on language
The downloadable spreadsheet is in Spanish; there is no English version of it yet. Its sheet names, column headers and dropdown values are Spanish, and the tables below keep them exactly as the file shows them, with the English beside them. This page explains what it contains and the reasoning behind each rule, so you can judge whether it fits before translating.
When to use it
Once you have findings and someone has to decide what gets fixed first. It works the same whether they come from a usability test, a heuristic evaluation or interviews with a guided walkthrough: what the matrix sorts is observed problems, not recommendations.
Do not use it to discover. It measures nothing about the product; it orders what you already found. If you are still working out what is going on, the step before this one is the research plan.
Before you use it
Severity is a judgement, not a measurement. What the rubric does is make it a calibrated judgement: that "serious" means the same thing when the researcher says it and when the engineer says it. Without a rubric, the discussion is about adjectives.
Three axes, not one. Frequency, impact and persistence. A problem that happens to everybody but resolves itself is not the same as an isolated one nobody escapes. Frequency and impact weigh 40 % each, persistence 20 %.
Two people score separately. And you only discuss the rows where they differ by more than a point. Scoring as a group from the start lets the first voice anchor everyone else.
Effort is not research's call. The effort column is filled in by whoever will build the fix. That is what turns severity into a decision about what happens now and what goes to the backlog.
Frequency is about the sessions, not the population. "6 of 10 participants" is not "60 % of users", and that translation is the most expensive mistake made with this spreadsheet.
The document
The download is a six-sheet spreadsheet — Instrucciones, Rubrica, Mapa de análisis, Hallazgos, Severidad and Resumen — with the formulas and dropdowns already in place. The severity, level and action columns calculate themselves.
The three middle sheets are a chain. Mapa de análisis is the audit trail from quote to theme, filled in only if there was a thematic analysis. Hallazgos consolidates: a finding enters there whatever its origin — thematic analysis, usability session, heuristic evaluation or analytics — and gets an ID. Severidad pulls it in by that ID and scores it. Writing the finding once is what keeps the chain intact: evidence → code → category → theme → finding → severity.
It ships with five example rows, all marked EJ in the EJ column, which on every sheet is the last of the control columns. That mark is the only valid one, the summary already excludes them from every calculation, and the instructions sheet explains how to delete them in one pass.
Scoring rubric
This is the sheet to read before scoring anything. It feeds the matrix dropdowns. The labels are the ones the spreadsheet shows, in Spanish; the English is beside them.
Frecuencia (frequency) · column D
| Value | Label | In English | Criterion |
|---|---|---|---|
| 4 | Todos | Everyone | Happened to 8 or more of every 10 participants |
| 3 | Mayoría | Most | Between 5 and 7 of every 10 |
| 2 | Algunos | Some | Between 2 and 4 of every 10 |
| 1 | Aislado | Isolated | A single participant, no pattern |
Impacto (impact) · column E
| Value | Label | In English | Criterion |
|---|---|---|---|
| 4 | Bloquea | Blocks | The task cannot be completed; the user gives up |
| 3 | Retrasa | Delays | It gets completed, but with retries or outside help |
| 2 | Molesta | Annoys | Causes doubt or irritation, with no meaningful time cost |
| 1 | Cosmético | Cosmetic | Noticeable, does not affect the task |
Persistencia (persistence) · column F
| Value | Label | In English | Criterion |
|---|---|---|---|
| 4 | No se supera | Never overcome | Happens again on every attempt, even when warned |
| 3 | Se supera con esfuerzo | Overcome with effort | Learned after several attempts |
| 2 | Se supera al segundo intento | Overcome on the second try | Fails once and then not again |
| 1 | Puntual | One-off | Happened once and did not repeat |
Severity scale
The level is calculated on its own, from the weighted severity.
| Range | Level | In English | What it implies |
|---|---|---|---|
| 3.5 – 4.0 | 1 · Crítico | Critical | Fixed before the next release |
| 2.5 – 3.4 | 2 · Mayor | Major | Goes into the quarter's prioritised backlog |
| 1.5 – 2.4 | 3 · Menor | Minor | Grouped with other changes to the same flow |
| 1.0 – 1.4 | 4 · Cosmético | Cosmetic | Recorded, not prioritised |
Each cell carries the number and the word. Colour is reinforcement, never the only channel: anyone who prints the spreadsheet in black and white has to be able to read the level all the same.
Severity is not a simple average: it is (2·Frequency + 2·Impact + Persistence) / 5. Frequency and impact count double.
And an impact of 4 never drops below «2 · Mayor». It is an override on the scale, and the reason is this: the formula already doubles the weight of impact, and even so a finding that blocks the task but happened to a single person scores 2.2 and lands in Minor. With 5 to 10 sessions, a frequency of 1 cannot tell "a rare edge case" apart from "we did not catch it with this sample". The Override column marks the rows where it kicked in, so you can see why they went up.
The matrix columns
One finding per row. You only write in the columns marked as manual.
| Column | In English | What goes in | Filled in by |
|---|---|---|---|
| A · ID | ID | The finding's identifier, copied from the Hallazgos sheet | research |
| B · Hallazgo | Finding | Pulled in by the formula, through the ID | formula |
| C · Paso del flujo | Flow step | Pulled in by the formula, through the ID | formula |
| D · Frecuencia | Frequency | Dropdown 1–4, per the rubric | research |
| E · Impacto | Impact | Dropdown 1–4, per the rubric | research |
| F · Persistencia | Persistence | Dropdown 1–4, per the rubric | research |
| G · Severidad | Severity | Calculated: frequency and impact at 40 %, persistence at 20 % | formula |
| H · Nivel | Level | Calculated from the scale above | formula |
| I · Esfuerzo | Effort | 1 low · 2 medium · 3 high | development |
| J · Acción | Action | Calculated: crosses severity with effort | formula |
| K · EJ | Example mark | The example mark, carried over from the finding | formula |
| L · Vínculo | Link | ok, sin vínculo (not linked) or texto divergente (text differs) | formula |
| M · Override | Override | Marks «sí» (yes) when impact raised the level | formula |
The Acción (action) column is what turns the matrix into a decision: high severity and low effort is Ahora (now); high severity and high effort is Planificar (plan); low severity and low effort is Rellenar (fill in); the rest, Descartar (drop).
The two modes
Linked, which is how it ships: the finding is written once, in Hallazgos, and Severidad pulls it in by its ID. To populate the ID column, copy the ID column from the Hallazgos sheet and paste it as values. It is one action per analysis session, not one per finding, and the column carries no formula on purpose: the sheet gets sorted and filtered.
Standalone: delete the formulas in columns B and C and write the finding directly. The link is cut, and the summary no longer excludes the example rows properly.
What does not work is the halfway point, typing over a formula without deleting it. The Vínculo column in Severidad says so row by row —ok, sin vínculo or texto divergente— and the Puntuado (scored) column in Hallazgos warns about the opposite direction: a consolidated finding that was never copied to Severidad and so was never scored. The Resumen sheet counts all three at the top right. It compares the value, not whether the cell is a formula: if someone overwrites it with the same text there is no real problem; the problem is the text diverging.
What gets shared and what does not
The full spreadsheet is an internal document: it contains the team's rubric and prioritisation criterion. What gets shared with the client is the Resumen sheet, which has the count per level and the average severity per flow step, and which excludes the example rows.
Note on method and limitations
Severity is an expert judgement calibrated with the rubric, not a measurement.
Frequency is estimated over the sessions observed, not over the user population.
Effort is filled in by the development team. A matrix with an empty effort column can sort by seriousness, but it cannot recommend what to do first.
It expires. Review on [date]: a matrix from two releases ago describes a product that no longer exists.
Check before you use it
- Did everyone read the rubric before scoring, or did someone score by feel?
- Is every row an observed problem, or are there recommendations dressed up as findings?
- Did two people score separately?
- Is the effort column complete, or can the matrix still not decide anything?
- Did you delete the
EJrows before sharing? - Is what you send the client the
Resumensheet rather than the whole file? - Was any chart built from the matrix instead of from the summary? It breaks when you sort or filter.
Grounding
The three axes and the idea of weighting them come from heuristic evaluation practice; how one is actually run is covered in Heuristic evaluation: UX audits on your own.
The findings that go in here come out of the sessions the interview guide collects, and the study that produces them is agreed in the research plan.