Custom features
A custom feature is a characteristic, written in free text, that you add to a set percentage of the synthetic consumers (personas) in a population. It lets you add to the population a habit, possession or attitude that the built-in filters don’t offer, and compare results between the personas who carry it and those who don’t. For example, in a coffee subscription study you can give 30% of personas the feature “Brews filter coffee at home every day” and compare this group’s price response with everyone else’s in the report.
Adding a custom feature to a population
Two custom features are defined in the population form, with their ratios and Smart Distribution set; then a completed report shows the Custom Feature breakdown and the Excel column.
When to use it
- When you want to add something to the population that the built-in filters don’t cover: a product the persona uses, something they own, a habit or an attitude.
- When you want to test a hypothesis: “Will pet owners pay more for this pet food?”
- When you want to split the report’s results into groups you define yourself.
For built-in options such as age, gender, socio-economic status (SES), city or work status, use a filter or custom ratios instead of a custom feature. The Filters and attributes page covers these options.
Before you start
Custom features can only be added while a population is being created. You reach the population form through the Create New Population card in the New Research wizard (Create a new population in Fashion and Ad Testing). The Creating a population page covers the whole form. The Custom Feature tab is available in all three forms: General Research, Fashion and Ad Testing.
Adding a custom feature
- Open the Custom Feature tab from the Population features list on the left.
- In the Feature Description field, describe the feature in one short sentence (e.g. “Brews filter coffee at home every day”).
- In the Ratio field, enter the percentage of personas that should get the feature. The first row starts at 50%; if you leave it unchanged, half of the personas get the feature.
- If you want the feature to go to personas whose profile fits it rather than to random personas, turn on the Smart Distribution toggle.
- To add another feature, click Add Feature and repeat steps 2–4 in the new row.
- Check the Total Usage badge in the tab header and the line below the list. This line shows how many personas will be left without a feature (such as “12 persons (%30) will have no feature”) or says “All personas will have a feature”.
- In the Population summary panel on the right, check that the features are listed in the format “(%30) 1. …”.
- Click Create population.
You can delete a row with its bin icon (Remove feature). On this tab, the Clear All Filters button deletes every custom feature row and leaves a single empty row set to 50%. If you set up the population with Argus, Argus can also enter the custom features in the form for you.
Smart Distribution
When Smart Distribution is off, the feature goes to the share of personas you set, chosen at random. This keeps the personas with and without the feature demographically similar.
When the toggle is on, the system interprets the feature text with a language model and picks the personas that fit it by looking at their profiles (age, gender, education, work status, income group, lifestyle, car and home ownership, and so on). For example, an income-related feature such as “Uses an iPhone” goes more often to higher-income personas. If not enough personas fit the feature, it may go to fewer personas than the ratio you asked for.
If you want to measure the feature’s own effect, leave the toggle off. If a realistic spread of the feature across the population matters more, turn it on; in that case, part of the difference between the two groups in the report may come from demographic differences such as income or age.
Limits
| Item | Rule |
|---|---|
| Feature text | Up to 1,000 characters in General Research populations. Rows with empty text are ignored. |
| Ratio | Between 0 and 100 in each row. |
| Total ratio | The rows can add up to at most 100%. Once the total reaches 100%, Add Feature is disabled. If it goes above 100%, the badge turns red, the “Total exceeds 100%” warning appears and Create population doesn’t work. |
| Per persona | Each persona carries at most one custom feature. |
| Number of personas | Population size × ratio, rounded down. 15% of a 50-persona population is 7 personas, not 7.5. |
| Credits | Adding custom features costs no extra credits. |
How personas take on the feature
The feature is added to the persona’s profile text. The language model that plays the persona reads it as a fixed part of the persona’s identity. The persona’s opinions, the examples they give and their purchasing behaviour are shaped by it. The profiles of personas without the feature don’t contain this information.
The feature’s effect on results is a language model’s interpretation; it shows direction and magnitude but doesn’t replace measuring real consumers. For details, see Accuracy and limitations.
Where the feature appears
| Where | What you see |
|---|---|
| Population detail panel | When you’re choosing an existing population and click View details on its card, the feature texts are listed under the Custom Features heading. |
| Focus group Participant Selection | A Custom Feature row on the profile card that opens when you hover over a participant card. |
| Reports | A Custom Feature dimension in the breakdown list. |
| Odak Grubu Analizi (Focus Group Analysis) | A Custom Feature tab and a Custom Feature Breakdown section; participants without a feature are grouped under “No custom feature assigned”. |
| The report’s Excel file | A separate column on the Population sheet, which lists the personas, and a breakdown by this dimension on the Segment Impact sheet. The column header is Custom Feature: in most reports it is always in English, and in the focus group file it follows the interface language. |
The population page in the Populations library, and the file you get from it with Download Excel, don’t show the custom feature. To see each persona’s feature, use the study report’s Excel file. Because this column is filled in when the file is downloaded, it also appears in the files of studies completed earlier.
Custom Feature breakdown in reports
The Custom Feature dimension appears in the breakdown list of Price Analysis (WTP), Feature Price Impact (Feature Value Analysis on the report screen), Priority Ranking (MaxDiff), the Choice-Based Conjoint (CBC) type of Competitive Analysis, and Multiple-Choice Survey reports. In Price Analysis, Feature Price Impact and Choice-Based Conjoint (CBC) reports, the groups are formed as follows:
- If a single feature was given to fewer than 90% of personas, there are two groups: Özel Özellikli (feature text) (With Custom Feature) and Özel Özelliksiz (Without Custom Feature).
- If a single feature was given to 90% or more of personas, the personas with the feature form one group and those without it form the Standart (Standard) group.
- If there are two or more features, each feature forms its own group; personas without a feature are left out of this breakdown.
These three reports don’t show groups of fewer than 5 personas in their charts. In Comparison, Ad Test and Fashion Multiple Choice reports, breakdowns use their own dimension lists, and Custom Feature isn’t one of them. The Segmentation page explains how to read breakdowns.
Good and bad examples
| Feature | Assessment |
|---|---|
| “Brews filter coffee at home every day” | Good: a single, concrete, observable habit. |
| “Prefers eco-friendly products” | Good: a clear attitude in one sentence. |
| “Has a cat at home” | Good: a situation the built-in filters don’t cover. |
| “Uses an iPhone” (Smart Distribution on) | Good: a possession linked to the profile; it makes sense to let the profile drive the distribution. |
| “Aged 25–34, lives in Istanbul” | Bad: already in the built-in filters; use the Age and City filters. |
| “Loves our brand and will buy it again” | Bad: presupposes the very result you want to measure. |
| “Price-sensitive, sporty, two children, into technology” | Bad: four features in one row; you can’t tell which one makes the group behave differently. |
| “Shops sometimes” | Bad: vague and true of almost everyone. |
| “Spends weekends with their grandchildren” in a population where only the 18–24 age group is selected | Bad: contradicts the population’s filters. |
Tips and common mistakes
- Leave a comparison group. If you give a single feature to 100% of personas, no group is left without it and you can’t compare. Keep the total below 100%.
- Define mutually exclusive options. Because each persona carries at most one feature, the rows form groups that are alternatives to each other, e.g. “Brews filter coffee at home every day” and “Mostly drinks coffee in cafés”.
- Work out the group size. 10% of a 40-persona population is 4 personas; that group won’t appear in Price Analysis, Feature Price Impact or Choice-Based Conjoint (CBC) charts. For a sturdier comparison, enlarge the population or raise the ratio: 30% of a 100-persona population is 30 personas.
- Describe the persona, not the result. The feature should say who the persona is or what they do, not how they will respond to your product.