# Comparison

Explains the Comparison method, which shows two products with their images to the same personas and measures which one is preferred, why, and in which groups the preference diverges.

Comparison shows two products, with their images, to the same synthetic consumers (personas) and measures which one is preferred, why, and in which groups the preference diverges. An example question: "Is our new-season linen shirt or a competitor's similar style preferred more, and why?"

Every persona in the population sees the details and images of both products. Each persona is given a situation in which they might buy or use the product, called a usage moment. The persona first scores each product separately and writes down their reasoning, then chooses A or B, or says they see no difference between the two products. Which product is shown as "A" is decided at random for each persona; in the report, the products appear in the order you entered them.

## When to use it

- When you're deciding between two designs, colours or product versions.
- When you're comparing your own product visually with a competitor's.
- When you want to see in which consumer groups each product stands out.

If you want to measure how price affects demand, use the [Price Analysis (WTP)](/en/docs/research/price-analysis) method.

Comparison is available in two verticals:

| Vertical | Products | Groups in the report |
| --- | --- | --- |
| **Fashion** | Clothing | Archetype, age, gender, income group |
| **General Research** | All other product categories | Market segment, age, gender, income group |

In Fashion, each persona is automatically assigned one of 12 fashion shopper types (an archetype); archetypes are described on the [Verticals](/en/docs/research/verticals) page. Personas also keep in mind a monthly clothing budget set by their income group.

## Before you start

- **Population:** You need a population in the vertical where you'll run the study. Populations aren't shared between verticals; a Fashion population can't be used in General Research. See [Creating a population](/en/docs/populations/creating-a-population).
- **Product information:** For each product, have the product page link ready, or the brand, product name, category, a short description and at least one photo. If you know the price, add that too.
- **Credits:** Comparison costs 10 credits per persona. In the **New Research** flow, you need at least 500 credits in your balance to create a General Research study, and at least 50 credits to create a new population in Fashion.

## Setting up the study

### In Fashion

1. Click **New Research** in the sidebar and choose the **Fashion** card.
2. Choose the **Comparison** card. Studies you haven't started are listed under **Drafts** on the same screen.
3. On the **Choose a population** screen, select a population with **Use existing population** and click **Continue with selected population**, or define a new population with **Create a new population**.
4. On the **New Fashion Study** screen, check the badge next to the title: **Popülasyon hazır** (Population ready), **Popülasyon oluşturuluyor** (Creating population) or **Popülasyon oluşturulamadı** (Population could not be created). You can enter the products while the population is being created.
5. If you like, fill in the **Study name (optional)** field, then add the products on the **From URL (Quick)** or **Detailed Entry** tab (see below).
6. Click **Next Step**. Until the population is ready, the button reads **Waiting...** and can't be clicked. No credits are spent at this step.
7. On the **Research Preview** screen, check the **Person Count** and **Credit Usage** values on the **Experiment Summary** card. **Credit Usage** is shown as "balance/estimate".
8. Click **Start**, then click **Continue** in the dialog that opens.

### In General Research

1. Choose **New Research** › **General Research** › the **Comparison** card.
2. On the **Choose Population Mode** screen, select a population with **Select Existing Population** or define a new population with **Create New Population**.
3. On the **New Comparison Study** screen, add the products on the **From URL (Quick)** or **Detailed Entry** tab. The **Estimated cost: … credits** badge next to the title shows what the study will cost.
4. Click **Next Step**. The study starts immediately and the progress screen opens.

> [!WARNING]
> General Research has no preview step; **Next Step** starts the study without opening a confirmation dialog. Before you click, check both products and the **Estimated cost** badge.

### Adding products by link

On the **From URL (Quick)** tab, paste the product page links into the **Product URL** fields on the **Product A** and **Product B** cards. Both links are required and must be different from each other. Shortly afterwards, **Reading product…** appears below the field. If the page is read successfully, a green preview card appears showing the image, brand, product name and price. If the page can't be read, you see a warning and a **Use manual entry →** link that takes you to the **Detailed Entry** tab.

In Fashion, the screen has a **Works with:** row, and a hint below each link tells you the site's status:

| On-screen hint | Sites |
| --- | --- |
| **This site is supported.** | Zara, Massimo Dutti, Bershka, Pull&Bear, Stradivarius, Oysho, Zara Home, Trendyol, Boyner, Koton |
| **This site is supported on a best-effort basis — switch to Detailed Entry if scraping fails.** | Beymen, İpekyol, LC Waikiki, Mango, Vakko, Network |
| **This site blocks bot traffic — please use the Detailed Entry tab.** | H&M, DeFacto |
| **We do not recognize this site — we will try, but it may fail.** | Other sites |

In General Research, there's no site list or hint; every product page is tried, and the preview card works the same way.

### Adding products manually

1. Open the **Detailed Entry** tab.
2. On the **Product A** and **Product B** cards, fill in the **Brand**, **Product name**, **Category** and **Description** fields. In Fashion, you choose the category from a list (e.g. **Women Dress**, **Men Shirt**, **Other / General**); in General Research, you type the category yourself.
3. If you like, fill in the **Color** and **Price (TL)** fields, and also **Fabric composition** in Fashion or **Key specs** in General Research.
4. In the **Images** field, drag and drop 1 or 2 photos for each product, or click the field and choose the files.

If any information is missing, you see the warning **Brand, product name, category, and description are all required.** or **At least 1 image is required.**

### While the study is running

The progress screen moves through the **PREPARING**, **RUNNING** and **SAVING** stages. If you leave the page, the study keeps running in the background. If you stay on the page, the report opens on its own within 3 seconds of the study finishing. For details, see [Starting a new study](/en/docs/research/new-study).

If the study ends with the **Experiment Failed** screen, in Fashion you can restart it with the same products by clicking **Retry**. In General Research, this screen has no **Retry** button; write to the SCL team through **Support** and include the study's name (see [Support and troubleshooting](/en/docs/account/support-and-troubleshooting)). If the image of either product can't be read, the study fails; in a new study, enter that product with **Detailed Entry** and upload an image.

## Settings and limits

| Setting | Limit |
| --- | --- |
| Number of products | Exactly two: **Product A** and **Product B** |
| Links | Both required, valid and different from each other |
| Required fields for manual entry | **Brand**, **Product name**, **Category**, **Description** and at least 1 image |
| Images | Up to 2 per product; JPG, PNG or WebP; up to 5 MB each |
| Description | Up to 2,000 characters |
| Study name | Optional; if you leave it blank, the study is named after the products in the form "Product A name vs Product B name" |
| Number of personas | The size of the population you choose (10–2,001 personas) |

## Credit calculation

Credits = number of personas × 10. For example, a population of 50 personas costs 50 × 10 = 500 credits, and a population of 200 personas costs 200 × 10 = 2,000 credits.

- **Fashion:** The cost appears in the **Credit Usage** value on the preview. If your balance is too low, you can't click **Start**; the **Buy Credits** button in the info icon's tooltip opens the purchase page.
- **General Research:** The cost appears in the **Estimated cost** badge. If your balance is too low, the form shows the warning **Not enough credits to start the survey.**

Credits are deducted only when the study completes successfully; a failed study costs no credits. Opening the report and downloading the PDF or Excel file don't cost credits either. For details, see [Credits](/en/docs/account/credits).

## Reading the results

The completed study is listed on the **Reports** page under the **Comparison** filter. The **Report sections** bar at the top of the report takes you to the **Summary**, **Product evaluation**, **Themes**, **Segments**, **Usage occasions** and **Responses** sections; a section with no data isn't shown. In the report, Product A is shown in blue and Product B in green.

> [!NOTE]
> Preference percentages are calculated only over the personas who chose A or B (the deciders). Personas counted under **No difference** aren't included in these percentages.

### Summary

At the top are the cards for the two products (image, brand, name, price). The **Study verdict** heading gives one of these results:

| Heading | What it means |
| --- | --- |
| **Product A stands out in this panel** (or Product B) | More than half of the personas chose one of the two products, and the 95% confidence interval for the winning product's preference share lies entirely above 50%. |
| **Preferences differ in some groups** | There's no clear difference across all personas, but at least one group's preference clearly differs from the rest of the population. Open these groups with **Explore diverging groups**. |
| **No clear preference difference detected** | The data doesn't support a clear difference in preference. This doesn't mean the products are equivalent. |

Next to it, the **Preference among deciders** bar shows the A and B percentages, and the line below shows the **Participants**, **Deciders** and **No difference** counts. The confidence interval is the range in which, based on this study's data, the preference share can reasonably be expected to lie. Headings seen in older reports, such as **Product A was chosen more often in this panel**, report only an observed difference that the confidence interval doesn't support.

### Product evaluation

The **Commercial diagnostic** table puts the two products side by side. In the **Difference** column, B's value is subtracted from A's: a positive value means A is higher, and a negative value means B is higher.

| Metric | What it shows |
| --- | --- |
| **Willing to buy** | Personas who report a purchase likelihood of 60% or higher, as a share of all participants |
| **Purchase probability** | The average purchase likelihood (0–100%) reported by the personas |
| **Attraction** | Visual appeal, scored 1–10; given independently of price |
| **Perceived value** | Whether the product seems worth its price, scored 1–10 |
| **Price objection** | The share of personas who raise a concern about price in their reasoning |

**Explore distributions** opens the distributions of the purchase likelihood, appeal and perceived value scores.

### Themes

**Voice of the consumer** shows two lists for each product: **Why other personas didn't pick this** and **Why advocates chose this**. Themes are created by grouping the personas' reasoning with AI. Each theme shows a short phrase, how many personas mention it and its direction (**positive**, **negative**, **mixed**); click a theme to open example quotes. The reasoning of no-difference personas isn't included in these lists.

### Usage occasions

**Fit by occasion** shows, for each AI-generated usage moment, the number of personas given that usage moment and the two products' average fit score (1–10). If a product has a **poor fit** marker, it was liked but doesn't suit that usage moment. Usage moments given to fewer than 3 personas aren't shown.

### Responses

**Persona details** is collapsed at first; expand it to see each persona's choice, scores and reasoning. You can search with the **Search reasoning…** field and narrow the list with the **Preference** filter, plus the **Archetype** filter in Fashion or the **Segment** filter in General Research.

## Analysing by segment

The **Segments** section has two cards.

**Preference patterns by segment** lists the groups whose preference clearly differs from the rest of the population, in the **Groups leaning more towards A** and **Groups leaning more towards B** columns. Each row shows the group's **Preference share**, sample counts and **Share of panel**. Only groups with at least 8 deciders and a difference supported by the confidence interval are listed; if there are no such groups, the card isn't shown. A group leaning more towards A is one that chose A more often than the rest of the population did; this doesn't mean that A has the majority within the group.

**Audience profile** shows the preference share in four breakdowns: **By archetype** in Fashion or **By segment** in General Research, plus **By age group**, **By gender** and **By income group**. Each breakdown first shows the 6 largest groups; open the rest with **Show all**. Click a row to open the 95% confidence intervals and the personas in that group.

Older reports may also show a **Group with the closest preference shares** box: the group, at the intersection of two dimensions, where the A and B shares are closest to each other. For the concepts of breakdowns and segments, see [Segmentation](/en/docs/results/segmentation).

## Exporting

The **Download PDF** and **Download Excel** buttons in the report header prepare the file in the background; while it's being prepared, the button reads **Generating PDF...** or **Downloading Excel...**. The PDF report summarises the page's sections in a slide layout. The Excel file contains these sheets:

| Sheet | Contents |
| --- | --- |
| Meta | Study name, number of personas, category, the brand, name and price of both products, export date |
| Executive Summary | Preference shares, 95% confidence interval, average scores, decider and no-difference counts, the interpretation of the result and the recommended next step |
| Segment Impact | Each group's preference shares and how they differ from the overall result |
| Themes | Themes, their frequency and example quotes |
| Scenarios | Usage moments and fit scores |
| Raw | One row per persona: demographics, usage moment, choice, scores and the reasoning behind the choice |

For file formats and downloading, see [Exporting and downloading](/en/docs/results/exporting).

## Tips and common mistakes

- **Check the image on the preview card.** If the card has no product image, enter the product with **Detailed Entry**.
- **Add the price.** Personas judge purchase likelihood and perceived value against the price. If the preview card shows no price, fill in the **Price (TL)** field on **Detailed Entry**.
- **Check the number of deciders.** If most personas saw no difference, the report doesn't declare a winner, even if all the deciders chose the same product.
- **Read price objections correctly.** This metric counts the personas who mention price in their reasoning; it doesn't measure price sensitivity.
- **Account for the brand effect.** Personas see the brand name, so brand awareness affects the result.
- **In General Research, wait for the population to be ready.** The study can't start until the personas have been created, and the form shows an error message; wait a while, refresh the page and try again.
- **Test the results with real consumers.** Validate diverging groups and important decisions with a small-scale test on real customers. See [Accuracy and limitations](/en/docs/getting-started/accuracy-and-limitations).

## Related pages

- [Starting a new study](/en/docs/research/new-study)
- [Verticals: General Research, Fashion, Ad Testing](/en/docs/research/verticals)
- [Reading a report](/en/docs/results/reading-a-report)
- [Segmentation](/en/docs/results/segmentation)
- [Credits](/en/docs/account/credits)