How responses are read

Usertick reads responses with AI in the background, on every plan, with nothing to set up. The inbox filters by these readings, the badges on each response show them, and the reading on your Overview is written from them.

​

Which responses are read

A response is read when it has a written answer of 20 characters or more, or two or more answers of any kind. What a person writes beside a screenshot or a file counts as a written answer. A link or an email address alone does not. When a response has two or more answers and none of them is a written answer of 20 characters, only the response as a whole is read, from its numbers and picks. A response that is not read shows Not analyzed in the inbox, and the reason is on hover:

• Nothing to read: the response holds too little to read, such as one score, one short answer, or only a link or an address.
• Analysis pending: the response is waiting to be read.
• Analysis unavailable: the reading could not be completed.

​

What each response gets

Every written answer of 20 characters or more is read on its own for:

• Category: the part of your product or service it is about, such as billing, delivery or the mobile app.
• Theme: the specific topic inside that category, such as a late delivery or a double charge.
• Type: what the person is doing. One of Bug, Idea, Complaint, Question or Praise.
• Sentiment: how the person sounds. Negative, Neutral or Positive.

The response as a whole, numbers included, is read for:

• Urgency: how soon someone should act. High, Medium or Low.
• Customer risk: Healthy, At risk, or About to leave when the person says or clearly implies that they are leaving.
• Spam: whether the text is gibberish, test input or an advertisement. A response with nothing written is never marked as spam.
• Satisfaction, Effort and Churn signal: how pleased the person sounds, how hard they had to work and how likely they are to leave. These three show on the response and can be used in workflows. They describe what the person said, so they can’t be changed by hand.

A reading appears only when the AI is confident enough. A response that could not be placed clearly shows No category instead of a guess.

​

Where the categories come from

When you create your organization, you select your sector, and your categories and themes start from the set made for it, up to your plan’s limits. Each sector has its own set: software, e-commerce, agency, finance, education, health and media. An online store starts with categories such as Shipping and delivery, Returns and refunds and Checkout and payment, and a finance product with Payments and transfers, Cards and Identity checks. When the set has more categories than the plan allows, the organization starts with the first ones, and the rest are offered as suggestions. If you type your own sector under Other, the organization starts from a general set that fits most products, with categories such as Ease of use, Speed and reliability and Pricing and value. The first time you open Settings → Analysis, Check your categories shows a set Usertick drafted from what you typed. Keeping the categories that fit and selecting Use applies them, and Keep the general set keeps the general one. Draft again writes a new draft, a limited number of times. Drafting and using a set is open on every plan. When more categories are kept than the plan allows, Use stays unavailable until enough are turned off. Settings → Analysis lists every category and its themes. Each category has a name and criteria, the description the AI reads when it decides. The page and every change on it need the Manage surveys permission.

​

Plan limits

Each plan allows a number of categories, and a number of themes in each category: 5 and 3 on Free, 10 and 5 on Starter, 20 and 8 on Pro, and 30 and 8 on Business. A deleted category or theme doesn’t count. Within these limits, every plan can do everything below. Settings → Analysis shows how many categories are in use beside Add category, and how many themes the open category has beside Add theme. When a limit is reached, adding is unavailable and the page says how to make room or upgrade. After a move to a plan with lower limits, every category and theme stays, and responses keep being filed under them. They can still be edited and deleted, and a new one can be added once the count is under the limit. See Plans and billing.

​

Change the categories

• Add categories your sector’s set offers but did not start with. They are listed under the categories as Suggested for your sector, or as Suggested categories for the general set, and selecting one adds it with its themes, up to the plan’s limit of themes.
• Add your own category with Add category. After you describe what it should catch, or select one of the ideas for your sector, Draft it drafts the name, the criteria and a few themes. Write it yourself skips the draft.
• See what a new category would catch before adding it. As soon as the draft is written, Usertick tries it on up to 100 of your written answers from the last 90 days, and again after you change its name or criteria. When it would catch some of them, Tried on your answers shows how many and where those answers are filed today. When most of them already sit in one category, Usertick offers Add as a theme under that category instead.
• Add a theme under any category with Add theme.
• Rename a category, or edit its criteria when it keeps catching the wrong answers. Criteria work best when they describe the situation instead of the words people use.
• Delete a category with Delete category, or a theme with Delete beside it. New responses are not filed under it, and responses already filed under it keep it. Before deleting, Usertick lists the workflows, saved views, webhooks and API subscriptions that use it, since they will stop matching, and asks you to type its name.

​

Correct a reading

A reading can be changed on the response:

• In the Details panel, the type, urgency, category and theme each have a menu. Choosing another value corrects the reading, No type (or no category or theme) removes it, and Back to automatic returns it to Usertick. Urgency offers High, Medium and Low.
• An answer that reads differently from the rest shows its own reading, and its popover corrects that answer’s category, theme, type or sentiment. Signals opens the readings for the whole response, such as customer risk and spam. In these popovers, Let the model decide returns a value to Usertick.

A corrected reading shows a small person mark. The correction applies everywhere: in filters, counts, workflows, exports and the API.

​

Where the readings show up

• The inbox filters and groups by category, theme, type, sentiment and urgency. See The inbox.
• The reading on the Overview is written from them: urgent responses taking a larger share, people ready to leave, a kind of feedback growing or spam reaching a survey. See Results.
• Workflows can act on them, for example by sending every response that reads as about to leave to a channel.