---
title: 'What our approach to surveys is based on'
description: 'At its core, the method is continuous improvement: a small change, a check of the result, then the next step. Below, we explain where the questions come from, how Yoseno analyzes responses, and what the method deliberately avoids.'
type: page
locale: en
canonical: 'https://yoseno.com/methodology'
---

# What our approach to surveys is based on

At its core, the method is continuous improvement: a small change, a check of the result, then the next step. Below, we explain where the questions come from, how Yoseno analyzes responses, and what the method deliberately avoids.

The cycle closes: the last step brings you back to the first.

- **improvement cycle:** 4 phases

Ask → See → Discuss → Verify

```mermaid
flowchart LR
  A["Ask"] --> B["See"] --> C["Discuss"] --> D["Verify"]
```

## Why surveys usually change nothing

It's not that teams are reluctant to answer. A typical survey loses the reason, can't handle the volume, and doesn't give the result back to the respondent.

1. **A number evens out differences:** One person's 'three' means chaos in tasks, another's 'three' means not understanding who is responsible for what. The average preserves the level of tension but loses its cause — the only thing you can work with.
2. **A large volume can't be read regularly:** A hundred detailed responses can be manually processed once, but not on every cycle. And without that regularity, the loop never closes, and feedback once again gets reduced to numbers.
3. **Feedback without consequences fades:** When a response leads nowhere, people start answering half-heartedly — not out of fear, but out of futility. Sincerity isn't won once; it needs to be maintained.

## Six ideas behind the method

None of these is new: everyone has encountered each one in their own work at some point. What matters is what each has become in the product.

1. **Improvement is a loop, not an event:** An organization can't be fixed once and for all: new people arrive, new tasks appear, and yesterday's solution stops fitting. The power lies not in a single step but in closing the loop. A measurement without understanding is numbers; understanding without action is talk; action without a follow-up measurement is hope that things got better.
   - **In the product:** A measurement is not a formal record but a turn of the loop. A new round doesn't erase the previous one; it sits alongside it: the team answers the same topics and sees whether the strain has eased where it was strongest.
2. **An honest voice is more important than an averaged number:** When someone is asked to rate their work on a scale, they give a number—and behind that number, all the meaning they intended disappears. One person is worn out by the chaos in tasks; another doesn't understand the boundaries of their role; the number is the same for both.
   - **In the product:** There's always room for words alongside the scale, and for a short or emotional response, the AI facilitator asks one follow-up question. That's usually enough to turn a vague “it’s complicated” into a concrete signal.
3. **Honesty depends on safety:** The truth about work is told where honesty doesn't get punished. This is a matter of environment, not individual courage: in an unsafe environment, even the most direct person will smooth things over.
   - **In the product:** No one sees the responses as they are; only anonymized results are released, and only when a question has at least five responses. The organization itself can work anonymously.
4. **Hundreds of voices come together into shared knowledge:** Knowledge of how work is really done is scattered across people's heads. To use it, you need to bring the pieces together without losing the loud voices or the quiet ones. In the past this hit a wall of volume: you could read a hundred honest responses once, but not on every cycle.
   - **In the product:** The algorithm groups similar responses, summarizes them by question, category, and organization, and connects them to management topics. Scattered comments become a profile: where the team is strained and where all is well.
5. **Small steps are more reliable than rare big ones:** A major restructuring looks like a decisive response, but it's precisely its scale that makes it hard to complete: when everything changes at once, cause and effect become impossible to see. A small step has a low entry cost, a visible 'before' and 'after' difference, and the ability to reverse it if it doesn't work.
   - **In the product:** The result is not a broad diagnosis but a few contained growth points. Each can be tackled on its own at a feasible pace.
6. **Change takes root when the team owns it:** A decision handed down from on high is carried out but not owned — and as soon as possible people revert to old habits. A decision the team arrived at itself sticks, because it is grounded in their own understanding.
   - **In the product:** Results are open to the team rather than going upstairs as a confidential report. Growth points stay open to ideas and discussion even in anonymous organizations; the person in charge takes one on without pulling it out of the shared space.

- [How response protection works](https://yoseno.com/security.md)

## From one answer to the next measurement

The whole approach rests on a single chain, where each link follows from the previous one. The order of links is the same in every survey: questions and themes change, but the path does not.

answer → theme → tension → growth point → practice → next measurement

```mermaid
flowchart LR
  A["answer"] --> B["theme"] --> C["tension"] --> D["growth point"] --> E["practice"] --> F["next measurement"]
```

1. **Answer:** Each question is linked in advance to one to four management themes. The answer gets markers: signal type — complaint, idea, or neutral — plus tone and scope.
2. **Theme:** An answer enriches the profile of the themes the question is tied to. A theme is the unit of analysis, so different surveys speak the same language.
3. **Tension:** When results are summarized, answers close in meaning are grouped, making it clear where tension is higher and where there are gaps in processes.
4. **Growth point:** The theme with the highest tension becomes a growth point — a recurring theme or a working tension confirmed by answers. If there were suggestions alongside the criticism, they are included in the description.
5. **Practice:** A practice from the knowledge base is matched to the growth point's theme; the knowledge base currently contains 118 published cards. The recommendation is derived from the theme rather than paraphrasing the answers.
6. **Next measurement:** The new measurement shows whether the tension on this theme has eased. If not, that's an honest signal that a different approach is needed.

## Why the analysis relies on a shared vocabulary of themes

A theme is a management process used to measure the state of a team. The theme reference is shared across all methodologies and currently contains 61 themes. That's why different surveys draw on one knowledge base, and a repeated measurement compares like with like.

**Example question:** Do you understand who is responsible for what?

- **role\_clarity:** Clarity of roles and boundaries — Understanding your areas of responsibility, role expectations, and the functions of adjacent roles.
- **process\_transparency:** Process transparency — Clarity of work processes, decision-making rules, and success criteria.

One question covers two themes at once, and the answer enriches the profile of each.

Some themes are based on established models: exhaustion and cynicism follow Maslach, and readiness for change follows ADKAR.

If a team raises something that isn't in the vocabulary, a theme is created dynamically — so nothing important is lost in a survey tailored to the team.

## A survey is a facilitated conversation, not a questionnaire

Eight to sixteen questions: enough for depth, and few enough not to tire people out. Questions can be grouped into categories.

What each question carries:

- The wording and focus for the person answering.
- Hidden cues for the AI facilitator: where to probe if an answer is short or emotional.
- A link to one to four management themes.

**Design discipline:** Each methodology is built for a specific goal: it is decided in advance what exactly will be discussed with the team and which themes are deliberately left out. Themes and categories are strictly separated, so no signal can be interpreted two ways. Next to each scale there is an invitation to talk, because valuable context is born in words. And the conclusion is framed as a starting point for discussion, not as a verdict.

How to set up and run a survey in a team — in [documentation](https://yoseno.com/docs/surveys/overview.md).

## AI handles the heavy lifting; decisions stay with people

The ideas behind the method are simple at heart, but in practice they hit the same wall — the volume of manual work. AI isn't here to replace the team's judgment; it's here to make the cycle truly repeatable.

During the survey, it helps turn emotion into a concrete signal. When summarizing results, it groups similar answers, clusters them by topic, shows where tension is strongest, and suggests a ready-made practice to match. People decide what to do with all of it.

## What the method doesn't do

The method is more honest when its limits are stated plainly.

- **Doesn't prove the effect of a single action:** Comparing measurements shows whether tension on a topic has changed. But everything else changes over the same period, so you can't attribute the shift to any one decision.
- **Doesn't diagnose an individual or a team:** A growth point concerns how the work is organized. It's not a basis for evaluating people and not a verdict about anyone personally.
- **Doesn't make decisions for the organization:** A practice from the knowledge base is a management move that can be discussed and adapted. The choice and the responsibility stay with the team.
- **Doesn't guarantee candor:** If the response doesn't lead anywhere, the next round will give less. The method creates the conditions for an honest conversation, but it doesn't replace the intention to listen.
- **Doesn't reveal everything:** An anonymized result appears only with enough responses to a question, so in a small group some breakdowns stay hidden. That's a deliberate trade-off for honest feedback.