How to Analyze Hotjar Survey Responses (Now Contentsquare)

On-site surveys collect a particular kind of feedback. A visitor answers one or two short questions on a specific page, at a specific moment in their visit, on a specific device. The answers are brief, often a single sentence, and they make most sense when read together with that context. This guide explains how to collect on-site responses that can be analyzed, how to get them out of Hotjar or Contentsquare and how to turn them into findings about where your website creates friction.

It is written for two groups of readers: teams that have Hotjar survey data in past exports or in an account that is being retired, and teams now running the same kind of surveys in Contentsquare. The analysis is the same in both cases.

Hotjar Is Now Part of Contentsquare

Hotjar merged into the Contentsquare Group on July 1, 2025, and its heatmaps, recordings and surveys are now part of the Contentsquare platform, as Contentsquare's announcement explains. New Hotjar accounts can no longer be created. In Contentsquare, surveys and feedback widgets belong to the Voice of Customer product area.

Existing Hotjar customers are being moved to Contentsquare. According to the migration FAQ, survey and feedback data transfers to Contentsquare, while the Hotjar account becomes read-only for a limited period, during which historical data can still be viewed and exported, and is then deleted. If your team has Hotjar survey responses it wants to keep, export them now.

Types of On-Site Surveys and the Data They Produce

Contentsquare offers six survey types, the same family Hotjar users will recognize. A Popover appears in a bottom corner of the page and suits short questions about that page. A Button and a Bubble stay minimized until the visitor clicks them, which suits feedback visitors give on their own initiative. An Embedded survey sits inside the page content, and a Full Screen survey opens as a modal over the page, which is useful for exit questions but more intrusive. A Link survey lives at its own URL on a separate domain, so its responses cannot be connected to a session replay and behave more like a conventional survey.

The question types include a five-point Reaction scale, a 1–5 or 1–7 Rating scale, a 0–10 Net Promoter Score question, radio buttons, checkboxes and short or long text fields. Each response also carries context: the date, the URL of the page where it was submitted, the device, browser, operating system and country, as Contentsquare's guide to understanding survey results describes. That context is what makes on-site feedback different from a survey sent by email, and the analysis should be built around it.

Writing On-Site Questions That Produce Usable Answers

Ask One Thing About One Page

A visitor is in the middle of doing something else, so on-site surveys need to be short and specific. Start from a decision the team needs to make about a page, such as whether the pricing page explains the plans clearly, and write the question for that page alone. "Is anything on this page unclear?" on the pricing page produces answers you can act on. A general "How is your experience today?" shown across the whole site produces answers that are hard to connect to anything.

Follow a Rating With One Open Question

A rating tells you how many visitors are unhappy; the comment tells you why. Pair a Reaction or Rating question with one open text question, and keep the follow-up neutral, such as "What is the main reason for your score?" One open question is usually the right number. A second or third one lowers the completion rate and splits the explanation across several columns, which makes the analysis harder.

Target by Page and Moment

Targeting decides which visitors see the survey, and therefore what the answers can tell you. Contentsquare's URL targeting can show a survey on specific pages, using rules such as "Starts with" or "Contains", and can exclude pages, for example showing a survey on every checkout step except the confirmation page. Full Screen surveys can appear right after the page loads, after a delay, when the cursor leaves the top of the page or when the visitor scrolls halfway down. Exit-intent timing suits questions about why someone is leaving; a delay suits questions about the page content. Where you can, keep a separate survey for each page, so each file answers one question about one place.

Getting Your Responses Out

Exporting From a Hotjar Account

Teams with a Hotjar account that has not yet closed can still export from the Hotjar interface, following Hotjar's own export instructions. Open the Surveys or Feedback dashboard, apply any filters you need, go to the Responses tab, click the download icon and choose CSV or XLSX. Exports of up to 10,000 responses download directly; larger ones are sent by email as a download link. All times in a Hotjar export are in UTC.

Downloading Responses in Contentsquare

In Contentsquare, responses are listed on a survey's Respondents tab with their context and can be downloaded as an editable file from the icons above the list. Results can be filtered by date, page URL, device, country and other criteria. Contentsquare's help center does not currently have a step-by-step article for the export itself, so check the file format and the columns once you have downloaded it.

Preparing the File

Keep every context column. The page URL, the date, the device and the country are what make on-site responses useful, and they are what any comparison later in the analysis depends on. Check the time zone of the date column before you look at trends: Hotjar exports are in UTC, while Contentsquare displays dates in your local time zone in its interface. If the survey asked for an email address, or if respondents typed names or order numbers into the text field, remove or mask them before sharing the file or uploading it anywhere, in line with GDPR and your own data policies.

If the analysis is about the comments, remove responses with no answer to the open question, along with any test submissions your team made while setting up the survey.

Reading Responses in Context

The same comment can mean different things on different pages. Take a subscription website that runs a short survey on its pricing page and an exit survey on its checkout page. "Too expensive" on the pricing page is usually about the visitor's perception of the plans and what they include. "Too expensive" on the checkout page may point to something more specific, such as a fee or tax that appears only at the last step. Counting both under one theme called "Price" hides the difference; reading them by page shows two separate problems owned by different teams.

Device works the same way. A complaint that a button "does nothing" is a vague comment until you notice that almost all of those responses come from mobile devices on one page, at which point it becomes a likely layout or loading problem. Where a response is linked to a session replay, watching the replay shows what the visitor did before answering, which is often the fastest way to confirm such a problem.

Read the ratings the same way, by page and by device rather than as one site-wide average, and check how many responses sit behind each number. If a page collected twelve responses, one or two unusual answers can move its average a long way.

Analyzing Responses at Scale

Coding Answers by Hand

For a few hundred responses, manual coding works well. Group the responses by page type first, such as product pages, pricing and checkout, because the themes on each are different. Read a sample from each group, write the recurring issues down as short codes with a one-line definition, then assign one or more codes to every response. Count the codes per page and device, and keep two or three representative quotes for each.

Contentsquare's AI Analysis

Contentsquare's survey tools include AI analysis of text answers. Sentiment analysis labels each answer as positive, neutral or negative and lets you filter by sentiment and follow sentiment over time. Automated tags apply tags you define, or choose from AI suggestions, to recurring themes in new responses. Once a survey has at least 20 text responses, an automated summary report gives a summary, findings, quotes and suggested next steps, based on the last 365 days of responses and written in English. These tools are a practical way to keep track of a live survey and to read the main themes of one survey quickly.

Where Built-In Analysis Stops

The limits appear when the question changes from what people are saying to where they say it and whether the difference is real. One sentiment label per answer cannot show that a comment is positive about the product and negative about delivery. Tags that apply only to new responses do not reclassify older ones, and a summary of one survey does not compare it with another survey or an older Hotjar export. Testing whether a theme is really more common on mobile, or on one page than on the rest, means checking every combination of theme, page, device and period. A chat assistant can summarize a few dozen comments, but classifying thousands consistently, testing differences and tracing every figure to its source need a tool built for that work.

Analyzing Survey Responses in Kimola

Kimola is an AI platform that turns customer feedback into enterprise-grade market research. It analyzes survey responses, reviews, support conversations and social discussions the way a research team would. It does not run surveys on your website; it starts from the file you export from Hotjar or Contentsquare.

Uploading the Export

From Home in Kimola, upload the Excel or CSV file. Choose the column that holds the open-ended answers as the text column, the response date as the Date column and, for on-site surveys, the page URL as the URL column. Every other column is kept, including the rating, device and country, and becomes a filter and a comparison group in the Report; a file reduced to a single column of comments loses most of what makes on-site feedback useful. Before creating the Report, you set its output language and see how many queries it will use, as Kimola's guide to uploading a dataset explains. A Report analyzes one text column, so if a survey has more than one open question, create a Report for each, starting with the one closest to your decision.

Pain Points and Experience Stages for Page-Level Friction

Kimola finds themes in the data itself without a predefined list and names them after what visitors say, so a checkout survey would show themes such as an unexpected fee at the last step or a discount code that is not accepted, rather than a broad label like "Pricing". With multi-label classification, a response that mentions two issues gets both themes, each with its own sentiment, which is known as aspect-based sentiment. Responses in 30+ languages can be analyzed together.

Dimensions ask a research question of every response. Pain Points identify what goes wrong or creates difficulty, which for on-site surveys is usually the central question. Experience Stages place each response at a stage of the customer journey, such as Discovery, Purchase Experience or Usage & Support, so comments from browsing, buying and using the product can be separated even when the survey ran across the whole site. Unmet Needs show what visitors expected to find and did not. Dimensions can be added when the Report is created or later, use additional queries and do not change the original dataset.

Signals Across Pages, Devices and Periods

Signals cross every theme and Dimension with the columns in your file and with time, test each combination and keep only the statistically significant results. A Difference might show that a Pain Point is concentrated on mobile devices, on one page URL or among visitors from one country. A Trend shows a theme becoming more or less frequent over time, such as complaints about checkout increasing after a site release. Signals need enough feedback and selected Dimensions. A Report without Signals means nothing stood out, not that the analysis failed.

Action Plans and What Kimola Does Not Do

On the Actions screen, Kimola can write an Action Plan for a Profile that describes your organization. It forms hypotheses from the Report, checks them against the responses, leaves out recommendations the evidence does not support, and shows the findings, Signals and quotes behind each action. Reports can be exported to PowerPoint, PDF and Excel. For a survey that keeps running, Kimola's Enterprise package offers Spaces, where each new file is classified with the same models and Dimensions so results stay comparable, and Signals run again after every upload.

Kimola does not run surveys, record sessions or build heatmaps; Contentsquare does that, and a session replay remains the best way to see what a single visitor did. If a survey receives a few dozen answers a month, reading them by page is often enough. Kimola is most useful with hundreds or thousands of responses and several pages, devices, periods or surveys to compare. Kimola has a free Starter plan with no credit card required, and paid plans start at $49 per month, as listed on the pricing page.

Frequently Asked Questions

Can I still create a Hotjar account?

No. Hotjar is now part of Contentsquare, and new Hotjar accounts can no longer be created. On-site surveys and feedback widgets are available in Contentsquare's Voice of Customer product, which has a free plan that includes a limited number of responses per month.

How do I export Hotjar survey responses?

In a Hotjar account that is still open, go to the Surveys or Feedback dashboard, apply any filters, open the Responses tab, click the download icon and choose CSV or XLSX. Larger exports arrive by email. Times are in UTC. Export anything you want to keep before the account closes, because Hotjar data is deleted after the migration's read-only period.

Can Contentsquare analyze open-ended survey answers?

Yes, within limits. It labels each text answer as positive, neutral or negative, applies tags to recurring themes and, once a survey has at least 20 text responses, writes a summary report in English. For several themes per answer, tested comparisons across pages, devices and periods or analysis across several surveys, export the responses to a dedicated analysis tool.

Can I analyze old Hotjar responses together with new Contentsquare responses?

Yes, if the surveys asked the same question. Put the responses in one file, align the columns so the answer, date, page URL and device line up, convert the dates to one time zone and add a column that names the source. In Kimola, that column becomes a comparison group, and Trends can show how themes changed across the whole period rather than only since the move.

Conclusion

On-site survey responses are short, but each one comes with a page, a device and a moment. Keeping those columns, reading the answers by page and device and testing which differences are real turn a list of one-line comments into specific findings about where a website creates friction.

Whether your responses sit in a Hotjar export or a Contentsquare survey, the same file can go into Kimola. An organization that already collects on-site feedback can upload it as it is and turn it into enterprise-grade market research, without designing a new study. You can create a free account and start with a survey you already have.