How to Build Customer Personas from Feedback

min read - Updated on Nov 15, 2026

You can build customer personas from feedback in Kimola by turning the reviews, survey answers and support conversations your customers have already written into a Report that shows who they are, how many customers each type accounts for and what each type talks about. Kimola asks two research Dimensions of every piece of feedback: Personas, which asks "Who is experiencing this?", and Social Context, which asks "With whom does this occur?" It groups the same type of person across all of the records, counts the records behind each type and keeps every persona tied to the quotations that support it. You then cross the personas with the Report's themes in Tables to see what each type of customer talks about and how it feels about it. This guide follows three steps: add the two Dimensions, read the personas and cross them with themes.

Getting Ready

Reports, the Personas and Social Context Dimensions and Tables are available in every package. Sign in to your Kimola account and have your customer feedback ready: a spreadsheet export or the review page links of your product, with feedback in which customers describe themselves or their situation.

The Goal

This guide answers the question that marketing, product and customer research teams ask before they decide what to build and whom to speak to: who are our customers, and how do the different types of them differ. Most teams work from personas that they designed in a workshop or drew from a handful of interviews. Customers also describe themselves in their own words in every review, survey answer and support conversation, and almost none of it is counted. Kimola reads all of it, groups the same type of person across thousands of records and counts the records behind each type, so a persona rests on what many customers wrote.

A persona in Kimola is a type of person or a role that Kimola infers from the text of the feedback, such as a parent shopping for the family or a shift worker who uses the product on a break. It needs no name, account or profile. A persona groups everyone whose feedback suggests the same role or situation, which sets it apart from a customer in Customers, available in the Enterprise package, where Kimola recognizes each actual customer by the value in a Space's Customer Column. Social Context adds the people around the persona: whether the experience is lived alone or with a partner, children, friends, colleagues or a team.

The method works best on feedback in which customers tell what happened to them and who they are: product reviews, app store reviews, open-ended survey answers and support conversations. Feedback that says nothing about its writer, such as "Arrived broken.", gets no Personas answer. The share of records that do answer Personas tells you how much of your feedback describes its writer at all.

Example

A mobile network operator wants to know who its customers are before it redesigns its plans. It puts its app store reviews and open survey answers into one Report. The Personas analysis might list remote workers who rely on home internet, parents managing a family plan and students on a tight budget as the largest types, each with the number of records behind it and quotations in customers' own words. Social Context might show whether these customers write about their plan on their own or together with their families.

Step 1: Add Personas and Social Context

Create one Report from the feedback you want to study and add the two Dimensions while you create it. Start from Reports or Home, depending on where your feedback is. Upload a spreadsheet (.xls, .xlsx, .csv or .tsv, one file of up to 30 MB) and choose its text column, paste the link of a product or business page, or connect a helpdesk or a spreadsheet in Google Sheets. Every way leads to the same classifier, structure and Dimensions steps. The full methods are in Create a Report by Uploading a Custom Dataset, Create a Report by Adding Links and Create a Report Using Integrations.

In the classifiers step, choose the Automatic Classifier, which discovers the themes of your data, and the Sentiment Classifier. Step 3 uses the themes to show what each persona talks about and the sentiment to show how it feels. In the Structure step, turn on Enable Multi-label Classification, so that a review that raises several topics receives a label for each. See Multi-label vs Single-label Classification.

In the Dimensions step, select Personas and Social Context. Personas is listed under All and under Actors, and Social Context is listed under All, under Actors and under Context. Click each one to add it to your selection. The selected Dimensions are listed under My List, so you can review them before you go on. Personas names the types of people who write, and Social Context names the people they share the experience with, so the two together describe a type of customer and its setting.

[Screenshot: The Dimensions step with Personas and Social Context selected and My List showing 2]

Tip

Review your selection under My List before you create the Report.

You can add other Dimensions to the same Report. Pain Points, Motivations and Unmet Needs give each persona its problems, its reasons for choosing you and what it still misses.

On the Review screen, check the Report Title and the Report Language, the language in which Kimola writes the analysis. Required Query shows the queries the Report will use. Every Dimension you select adds to them: queries = rows × models × Dimensions. See How to Calculate Queries for an example. Create the Report. When it is ready, click View Report.

Note

Each Dimension adds to the queries of the Report, so check Required Query before you create it.

If your Report is already analyzed, you can add the Dimensions to it without creating it again. Open its Dimensions page, click the Personas tab, which has no check mark yet, and click Generate Now. Kimola classifies the Report's existing records with Personas and writes its analysis. Do the same for Social Context. See Add Dimensions to a Report.

Step 2: Read the Personas

Open the Report and click Dimensions in its menu. A row of tabs runs across the top of the page: All, then one tab for each Dimension. A Dimension with a check mark next to its name has been analyzed in this Report. All shows a card for each analyzed Dimension, with a ring for the share of the Report's records that answer it. Click the card of Personas, or its tab, to read it. See Dimensions in Reports for every part of the page.

Read the Summary

The Personas tab opens with the summary card. It gives the number of records that answer Personas and their share of total records, for example "600 records" and "20% of total records", followed by a headline and a few paragraphs on who your customers are, with key terms in bold. The share tells you how much of your feedback describes its writer at all. Kimola classifies the records and counts the personas first. An AI model writes the summary afterwards from numbers that are already computed, so the text never decides which persona is larger.

A persona is inferred from what the feedback says, so read each one together with its quotations. They show how customers describe themselves, and you can judge the persona from them.

Read the Personas as Clusters

Below the summary, each persona is a cluster card, largest first. A card shows the persona's name and a headline, its number of records and its share, a summary of what its records say and quotations: direct excerpts from the dataset in customers' own words. Click Show more >> to read a long summary in full. Click the eye icon next to a quotation to open the whole record.

[Screenshot: A Personas cluster card with its ring, record count, summary and Quotations]

On the right, a panel lists every persona with its number of records and a bar for its share. Cluster Scale at the top of the panel shows each persona's share of the records that answer Personas, and Dataset Scale shows its share of all the records in the Report. A persona with 150 records reads 30% on Cluster Scale when 500 records answer Personas, and 6% on Dataset Scale in a Report of 2,500 records. One record can point to more than one persona, for example a parent who also works from home, so the shares on Cluster Scale can add up to more than 100%.

Note

One record can point to more than one persona, so the shares on Cluster Scale can add up to more than 100%.

Kimola writes a summary only for a persona with enough records to support one. When some personas are too small, the page lists them under "Summaries are not available for the following topics due to small cluster sizes:", and their record counts are still shown.

Example

For a mobile network operator, the card of Remote worker who relies on home internet might carry a summary about video calls that drop during the workday and a quotation from a customer who describes working from a spare room. The card of Parent managing a family plan might center on bills that change when a line is added.

Add the Setting with Social Context

Open the Social Context tab to read with whom customers live the experience. It reads the same way as Personas, with its summary, its clusters, their record counts and their quotations. Its clusters are the settings customers describe, such as Alone, With a partner or With children. Read the two tabs together: Personas tells you who your customers are, and Social Context tells you whom they share the experience with. A student who writes about sharing a plan with friends can then be told apart from a student who writes about a plan used alone.

Look Inside a Persona

Every cluster card ends with two buttons that show what lies behind a persona. Columns lists the columns of your data, such as a country, a platform or a plan. Dimensions lists the Report's other Dimensions, each with the share of this persona's records that also answer it. Drag a column or a Dimension into the box that says "Drop columns and dimensions to create a table" to see its values inside the persona. Social Context shows the settings that go with the persona, and a column shows, for example, the platforms it writes from. When the Report has a sentiment classifier, each row also shows its sentiment. A table on a cluster card is a quick look and is not saved. Step 3 shows how to keep one.

Step 3: Cross Personas with Themes

A list of personas tells you who your customers are. Crossing it with the themes of the Report tells you what each type talks about and how it feels, and that turns the list into personas you can act on. You do this in Tables, the page of a Report where you drag classifiers, columns and Dimensions into a table and Kimola groups every record by them. Tables reads the analysis the Report already has, so building a table starts no new analysis and never changes the Report. See Tables in Reports for every part of the page.

Build the Table

Open Tables in the menu of the Report. Drag Personas from the Dimensions card into the table area, which says Drop columns to create tables until you add a metric. The table lists every persona as a row with its name, the number of records in brackets and a bar with its share. Then drag the content classifier from the Classifiers card, such as Themes, into the table. Its labels now appear under each persona. A row under a persona shows its share of that persona's records: if a persona has 100 records and 40 of them carry the theme Billing, the Billing row under it shows 40%.

[Screenshot: A table grouped by Personas, then by Themes, with the two metric chips above it]

The bar on the right of each row shows sentiment: green for positive feedback, red for negative feedback and grey for feedback that is neither. The label above it gives the share of positive feedback, such as "81% Positive", or the share of negative feedback, such as "100% Negative", when the group has no positive feedback at all. A persona row tells you how that type of customer feels overall, and a theme row under it tells you how that type feels about that theme.

The order of the metrics decides how the table reads. Personas first, then Themes, shows the themes of each persona. Themes first, then Personas, shows how each theme splits across personas.

Tip

Put Personas first to read the themes of each persona, and put Themes first to read the personas behind each theme.

Because a piece of feedback can carry several labels, in a multi-label Report the shares of a metric can add up to more than 100%.

Example

For the mobile network operator, a table grouped by Personas and then by Themes might show that Parent managing a family plan talks mostly about Billing, with a bar that is mostly red, and that Remote worker who relies on home internet talks mostly about Data Speed and Coverage. Student on a budget might hardly mention Customer Service at all. Grouped the other way round, the Billing theme lists the personas that raise it.

Change the Angle

To look at the personas from another angle, click the cross on the Themes chip and drag another metric into the table. Social Context shows the settings of each persona. A Dimension you added to the Report, such as Pain Points, shows what goes wrong for each persona. A column of your data, such as a platform or a country, shows where each persona writes from.

Filters narrow the records a table reads, and the table is built again from those records only. Click Add in the Filters card and choose Columns to read the personas of one platform or country, or click the dates at the top right of the page to choose a period.

Read the Records Behind a Persona

A table shows how much of the feedback each persona accounts for and where it sits. The records show what customers wrote. Open Content in the menu of the Report, click the Dimensions button, tick the persona you want and click Apply. The page shows only that persona's records, and the chart, the most frequent words and the record count follow the filter. Add a theme from Themes to read what that persona wrote about it. See Filter Records in Reports.

To see which themes hurt most for one persona, open Performance and choose the persona under Dimensions. The Dissatisfaction Score of every theme is calculated again for that persona's records only. See Performance in Reports.

To set personas side by side, click Save as New Filter on Content once for each persona, then open Compare and choose Compare by Filters. Compare shows the sentiment, the themes and the Dimension answers of each saved filter in one table, for up to 5 filters at once. See Save Filters in a Report and Compare in Reports.

Save the Table

Click Save at the top right of Tables to keep the table. In the Save Table window, type a name that says what the table shows, such as "Personas by Theme", and click Save. Saved tables are listed under Choose a Table, and opening one puts its filters and metrics back. A saved table can also be pinned to the Overview of the Report, where the TABLES card shows it with an Open Table link.

Tip

A name such as "Personas by Theme" makes the table easy to find under Choose a Table.

What You Get

A Report built this way gives you the following, all from your customers' own feedback.

Personas in your customers' terms. The types of people who write to you, each with the number of records behind it, its share of the feedback and a summary, on the Dimensions page of the Report.

Evidence for every persona. Quotations in customers' own words under each persona, and the whole record behind each quotation.

The setting of each persona. The social settings customers describe, such as alone, with a partner or with children, from Social Context.

A persona-by-theme table. The themes of every persona with their shares and their sentiment, kept in Tables and pinned to the Overview if you want it on the first page.

The records behind each persona. The feedback of one persona on Content, and the themes ranked by Dissatisfaction Score for that persona on Performance.

To share the findings with colleagues who don't use Kimola, export the Report as a PDF file or a PowerPoint file.

Next Steps

A Report describes the customers of one moment. To follow whether new types of customers start writing after a launch or a campaign, a Space in the Enterprise package gets Personas from a Dimension model in its Funnel and marks a persona whose share changed with the last upload. See What is the Personas Dimension?.

Personas become more useful next to the problems and reasons behind them. How to Find Customer Pain Points shows how to read what goes wrong, and How to Understand Why Customers Choose You shows how to read why customers pick you and what they replace. To turn what you learn into work, How to Turn Customer Feedback into an Action Plan takes a Report to rated actions. In the plan's The situation field, you can describe the type of customer you plan for. For how teams use persona research, see Personas.

Was this article helpful?

Tell us about your thoughts and experiences regarding the article.