# Reading a report

How to read the common sections of SCL reports, their distributions and segment differences, use persona reasoning, and compare the reports of two studies side by side.

Every number in a report is calculated from the responses of synthetic consumers (personas). All reports follow the same skeleton: summary figures, a main chart, segment breakdowns and, for some methods, persona reasoning. To find, rename and delete reports, see [Reports](/en/docs/results/reports).

## The common structure of a report

From top to bottom, a report page is made up of these parts:

| Part | What it shows |
| --- | --- |
| Breadcrumb | **Reports** › study name |
| Header | A small method label (e.g. **Price Analysis (WTP)**), the product or study name, and a one-line description |
| Buttons | **New Research**, **Refresh**, **Download PDF**, **Download Excel**, **Delete**; not every method shows all of them |
| Summary boxes | The key figures; in most reports, the number of personas included in the analysis (**Participants**) |
| Warnings | Data issues, e.g. **Low Response Rate Warnings** |
| Tabs or sections | The main chart, segment breakdowns, tables |

Hover over a chart to see its values; click a series in the chart legend to hide it. **Refresh** does not ask the personas again: it only reloads the analysis and uses no credits. For PDF and Excel, see [Exporting and downloading](/en/docs/results/exporting).

## Reports by method

Each metric is defined on its method's page. Tabs without data are hidden; in Price Analysis, for example, **Crosstab** appears only if there are at least two breakdown dimensions.

| Method | Main tabs or sections | Persona reasoning |
| --- | --- | --- |
| [Price Analysis (WTP)](/en/docs/research/price-analysis) | **Demand & Revenue**, **Price Table**, **Segment Analysis**, **Crosstab** | No |
| [Feature Price Impact](/en/docs/research/feature-price-impact) (**Feature Value Analysis** in the report) | **Demand Curves**, **Segment Analysis**, **Crosstab** | No |
| [Competitive Analysis (Conjoint)](/en/docs/research/competitive-analysis) | **Demand & Revenue**, **Segment Analysis**, **Price Scenario**, **Crosstab** | No |
| [Priority Ranking (MaxDiff)](/en/docs/research/priority-ranking) | **Best / Worst**, **Segment Analysis**, **Importance Ranking** | No |
| [Multiple-Choice Survey](/en/docs/research/multiple-choice-survey) | **Response Distribution**, **Statistics**, **Insights**, **Demographic Breakdown**, **Crosstab**, **Data Table** | No; choice only |
| [Comparison](/en/docs/research/comparison) | **Summary**, **Product evaluation**, **Themes**, **Segments**, **Usage occasions**, **Responses** | Yes |
| [Real-Time Focus Group](/en/docs/research/focus-group) | **Genel Bakış** (Overview), **Kümeler** (Clusters) | Yes |
| [Ad Testing](/en/docs/research/ad-testing) | **Sahne Sahne** (Scene by scene), **Ayrışma** (Distinctiveness), **Kalıcılık** (Memorability), **Persona Sesleri** (Persona Voices), **Hangi Kitle Ne Düşünüyor?** (What Does Each Audience Think?), **Persona Yorumları** (Persona Comments) | Yes |

## Reading distributions

- **Find the denominator.** Percentages are calculated over the personas included in the analysis. Personas that could not produce a response are removed from the analysis, so the **Participants** count can be lower than the population size.
- **Watch out for conditional and multi-select questions.** In Multiple-Choice Survey, the percentage for a conditional question is calculated only among the personas who saw and answered it. For multi-select questions, the percentages can add up to more than 100%.
- **Account for those who saw no difference.** In Comparison, preference percentages are calculated only over the **Deciders**; personas counted under **No difference** are not part of this share.
- **Use revenue only to rank prices.** In Price Analysis and Competitive Analysis, revenue is the price multiplied by the number of personas who would buy; it is not actual turnover in lira.
- **Check which price it was measured at.** In Competitive Analysis, the **Segment Analysis** bars are the average across all test prices, while **Preference at Optimal Price** is the share at the optimal price only. That is why the two numbers do not match.

## Reading segments and segment size

On the segment tabs, you split the result by a dimension such as age, income group, socio-economic status (SES) or a custom feature; changing the dimension costs no extra credits. Before you interpret a difference, check how many personas the segment has: the **N** column in the lower table on the **Crosstab** tab, the "participants · deciders · saw no difference" line under each group in Comparison, and the **Segment Impact** sheet in the Excel file of most reports all give this number. The [Segmentation](/en/docs/results/segmentation) page covers the dimension selector in each report, which reports hide small segments, and how to choose a dimension based on population size.

## Spotting meaningful differences

Some reports show for themselves whether a difference is reliable:

- **Multiple-Choice Survey:** The **Statistics** tab gives a **p-value** and an **Effect Size** for each pair of question and dimension (except for multi-select questions). Give priority to relationships with p<0.05 and an effect size of "moderate" or "strong"; when you run many tests, some "significant" results turn up by chance. On the **Insights** tab, a segment finding counts as reliable only if the segment has at least 30 personas, differs from the overall result by at least 10 percentage points and has p<0.05; if no finding meets these conditions, the findings are shown with a **Limited Data Warning**.
- **Comparison:** The report gives a 95% confidence interval for the preference share; group rows read "Confidence interval supports the between-group difference" or "Between-group difference is uncertain". A group needs at least 8 deciding personas to be listed as a group that stands apart. The message "No clear preference difference detected" does not mean the products are equivalent.
- **Feature Price Impact:** If the **95% WTP Impact Interval** does not include zero, the direction of the feature's effect on willingness to pay (WTP) is more consistent.
- **Ad Test:** If the **Hangi Kitle Ne Düşünüyor?** section shows the change in intent after the ad, only audiences that rise by more than half a point are green, and only those that fall by more than half a point are red. Compare the **Ayrışma** rate with chance level, not with 100%.

Price Analysis, Competitive Analysis and Priority Ranking give no significance indicator. In these reports, do not treat differences of a few points in small segments as findings. A single number such as the **Optimal Price** can shift from run to run, so look at the overall direction of the curve as well.

## Using persona reasoning

Reasoning shows the why behind a number, and only three methods have it:

| Method | Where | What you see |
| --- | --- | --- |
| Comparison | **Themes** (**Voice of the consumer**), **Responses** (**Persona details**) | Themes and quotes with their persona counts; each persona's **Reasoning** and **Choice reasoning** text |
| Focus group | **Genel Bakış**, **Kümeler** | **Ana Temalar** (Key Themes) and clusters; **Yanıtları gör →** (View responses →) opens the related responses |
| Ad Test | **Sahne Sahne**, **Persona Sesleri**, **Çekenler ve Engelleyenler** (Draws and Barriers), **Persona Yorumları** | "neden zirve" (why it peaked) and "neden dip" (why the dip) quotes for each scene, and the aspects personas liked and disliked |

> [!NOTE]
> Multiple-Choice Survey records only the choice and collects no reasoning, and the reports of the price and feature methods contain no persona quotes either. If you are looking for an answer to "why", see [Choosing the right method](/en/docs/research/choosing-a-method).

1. **Look at the number first, then the quote.** A theme's persona count tells you how widespread the reasoning is; a quote is only an example.
2. **Don't treat extreme examples as representative.** The **Persona Sesleri** cards are deliberately chosen as two personas from each end: those who liked the ad most and those who liked it least.
3. **Search for your own question.** In Comparison, type a word such as "price" in the "Search reasoning…" box and narrow the results by **Archetype**, **Segment** or **Preference**; in Ad Test, use the search in the **Persona Kanıtları** (Persona Evidence) card.
4. **Check summaries against the quotes.** Themes, clusters, **Genel Duygu** (Overall Sentiment) and **Uzlaşı** (Consensus) are a language model's assessment, not a count. If you asked new questions in the focus group, click **Recalculate**; otherwise the new responses are not included in the analysis.

## Comparing studies

If you tested two versions of a product, or two prices, in separate studies, compare their reports like this:

1. **Name your reports.** On the **Reports** page, choose **Rename** from the card's ⋮ menu and give each version a distinctive name.
2. **Open reports side by side.** In separate browser tabs, compare the **Participants** counts first, then the same summary box and the same segment dimension.
3. **Put the Excel files side by side.** For every report except Ad Test, compare the **Executive Summary** and **Segment Impact** sheets of the two Excel files.

To set up the studies you will compare on the same population and to test the difference with a repeat run, see the "Good practices" section of [Accuracy and limitations](/en/docs/getting-started/accuracy-and-limitations). To have the same personas evaluate two products in a single study, use the [Comparison](/en/docs/research/comparison) method.

## Writing up findings

The "Reporting findings inside your organisation" section of [Accuracy and limitations](/en/docs/getting-started/accuracy-and-limitations) uses an example to show how to state the source, the population, the number with its denominator, and the caveats when you share a finding. Argus can also prepare a presentation from the studies it reads in a conversation; see [Strategy reports and presentations](/en/docs/argus/strategy-reports) for details.

## Related pages

- [Reports](/en/docs/results/reports)
- [Segmentation](/en/docs/results/segmentation)
- [Exporting and downloading](/en/docs/results/exporting)
- [Accuracy and limitations](/en/docs/getting-started/accuracy-and-limitations)
- [Choosing the right method](/en/docs/research/choosing-a-method)