> ## Documentation Index > Fetch the complete documentation index at: https://docs.recepai.ai/llms.txt > Use this file to discover all available pages before exploring further. # Results > What a survey measured — how many answered, where the hotel is weak, and who is still waiting. **Results** answers three questions about one survey: how many guests answered, where the hotel is weak, and which guests are still waiting for you. Every figure on it follows the same rules, written out at the end of this page. *** ## What is on the screen **A period** — *7 days*, *30 days*, *90 days*, *1 year*, or **Pick dates**. A custom range includes both days you pick, read in your hotel's own calendar. **Four cards across the top:** | Card | What it says | | ------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | **Answered so far** | How many guests answered out of how many were invited in this period. Under it, the rate among invitations whose week is over, how many are still within theirs, and how many could not be sent at all. | | **Overall rating** | The average of the answers to this survey's own overall question. Never built from the other questions. Under it, how many answers it rests on — marked *Early signal* while they are few. | | **Gave 4 or 5** | The share of guests who answered the overall question with 4 or 5. Shown beside the average on purpose: an average hides a split room. On a small sample the card adds the range the true share is likely in. | | **Still waiting** | Submissions carrying an answer at or below your follow-up threshold that nobody has marked resolved yet. | **The picture:** * **Overall rating spread** — how many guests gave each score, 1 to 5, with the share that gave 4 or 5 and the share that gave 1 or 2. * **Answers by channel** — which road the answers came back on: WhatsApp, SMS, Email. Invitations that could not be sent are named and counted in no denominator. * **A pick-one question** as a ring, when the survey has one — a share of the picks, never an average. * **Overall rating over time** — by day or by week. A period with no answers is left out, never drawn as zero. **Departments** — a chart, *Where guests are unhappy*, and a table. Each department shows its answers, its average, how many answers fell at or below your follow-up threshold, and how many of those are still open. Departments are ordered by that count — the same threshold that alerts your team. A department that mixes star and 0–10 questions says so and is read on the 1–5 band. **What guests wrote** — written answers grouped by department and subject, with the guest's own words behind every row. [How written answers are read](/front-desk/reviews/surveys-written-answers) explains this block. **Question by question** — every question of the survey, its answers, its average and spread for a rated question, its picks for a pick-one, how many said *Did not use*, and how many answers fell at or below the threshold. A question whose wording changed during the period says so. **Guests still waiting** — the work list, last on purpose. One card per submission with a low answer: the room, the guest, when it came, the overall score, each low answer with the guest's words, and whether it is *Open* or *Resolved*. **Open this feedback →** goes to that entry on the Guest Feedback list; the list itself is where the whole queue lives, filtered by *Needs Attention*. **How these numbers are counted** — expandable, with this period's own figures filled in. *** ## What you do here The average alone hides a split room; the spread and *Gave 4 or 5* show it. The count at or below the threshold is a fact about what guests answered. It is not a claim that the department moved the overall score — that is not measured, and the screen does not say it. The guests' own words are behind it. **Create a work order for this →** opens a new work order in that department with the subject already written in. Results shows the most recent of those still waiting; the full queue is on the [Guest Feedback list](/front-desk/reviews/inbox) under *Needs Attention*, where an entry is resolved. **A percentage that is missing is not a gap.** Under five answers the screen shows the counts instead, because a percentage built on three answers would mislead. *** ## How to read these numbers The **?** beside every survey title opens these rules. They are the method every figure on the survey screens follows. The big number is how many guests answered out of how many were invited in this period. It can only rise while links are still open, so it is never written as a percentage. The rate underneath covers only invitations whose 7-day link has finished — answered or not. That is what makes two periods comparable. An invitation the messaging provider refused never reached the guest. It is named on the screen and enters no denominator. An invitation with no delivery record is read neither as sent nor as failed. Every survey carries one overall question, and that is the only place a rating comes from. We never build a score out of the other answers: a low housekeeping answer beside a high restaurant answer does not average into a verdict on the stay. A guest who never went to the spa is not unhappy with the spa. “Did not use” is counted beside the question and enters no average and no denominator. Under five answers a percentage would mislead, so the counts are shown instead. A figure built on a small number of answers is marked as an early signal — read it as a hint, not a result. Departments are ordered by how many answers fell at or below your follow-up threshold — the same threshold that alerts your team. They are not ordered by “which department dragged the score down”. Saying that would need enough answers to measure how each department moves the overall rating, and we do not claim it until the numbers can carry it. A low answer alerts your team on its own — even under a five-star stay. This is the point of asking during the stay: the moment to fix something, not to report it. Written answers are read a few minutes after they arrive and grouped by department and subject, with the guest’s own words behind every row. A count in that block is over the answers that were read, never over the ones that were written. When the words do not say which department they belong to, they are counted as undetermined rather than guessed into one. **A 0–10 question is reported on its own scale.** Its average and its spread are shown as 0–10, never retold as stars, and it never stands in for the overall rating — that comes only from the survey's star question marked *Overall rating*. A recommendation score is built only from the one 0–10 question you marked as the **Recommendation question** in the editor, and it is shown on [Feedback Analytics](/front-desk/reviews/analytics), not here. *** ## What happens next Nothing changes from this screen. It points at work elsewhere: a work order on [Active Requests](/front-desk/requests/active), an entry to resolve on the [Guest Feedback list](/front-desk/reviews/inbox), or a moment to change under [Guest Feedback settings](/front-desk/settings/reviews).