# Monthly reputation watch sheet **Version:** 2.0 **Date:** 2026-08-22 **Type:** Instrument --- ## 1. What this is A one-page monthly record of the property's public reputation, kept by the property, filled by hand from each channel's own public page. One row per channel, one sheet per month. It sits beside whatever dashboard a vendor supplies. It is not that. A dashboard shows the state now. This keeps a series, and each sheet is chained to the one before it. ## 2. The sheet **Month recorded:** ____________ **Previous sheet's month:** ____________ **Checking window:** within ______ days of month end (set once, before first use) **Sheet closed by:** ____________ **Date closed:** ____________ | Channel | Type (booking / review) | Out of | Score as displayed | Count this month | Count carried from last sheet | Net change | Checked (date) | By | |---|---|---|---|---|---|---|---|---| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Per channel, carried forward each month: | Channel | Rank | Market size | Brand-required minimum (with scale) | Reviews needed to move the displayed score one place | |---|---|---|---|---| | | | | | | | | | | | | | | | | | | Add rows for channels. The columns are fixed. The closing identity depends on them. ## 3. How the sheet closes Every row must satisfy: > **Count this month − Count carried from last sheet = Net change** **Count carried from last sheet** is copied from the previous sheet, not recalculated. **Count this month** is read off the channel. The difference is written in the third cell. The row does not close until the three agree. That makes the sheet checkable against a second document. Lay this sheet beside the previous one: every carried figure must equal that sheet's **Count this month** for the same channel, and the **Previous sheet's month** in the header must be the month immediately before this one. A skipped month shows as a gap in the header of the next sheet ever filled. A fabricated count shows as a carried figure that matches no prior sheet. Neither requires anyone to remember anything. No cell has a blank state. Where a channel publishes nothing for a field, write "unpublished". A blank therefore means the channel was not checked; it can never be read as zero, because a zero is written as `0` and a blank leaves the row's identity unresolved, so the sheet cannot be closed with it. **Checked (date)** must fall inside the window written in the header. A date outside it is a row carried over, not a row checked. **Net change** is a net figure and may be negative. Counts fall: channels remove, filter and re-file reviews. A negative is a normal reading, not an error, and it is recorded with the minus sign. ## 4. Scale and score **Out of** is filled first, on every row, every month. Channels do not share a scale. Scores written side by side with no scale attached stop meaning anything within a few months. Record the score **exactly as displayed, at the precision displayed.** Do not round, do not add a decimal place the channel does not show. Rounding by different hands in different months destroys the series. There is no line for a combined score, deliberately. Averaging across scales produces a number with no referent. Converting each score to a percentage of its own scale and weighting by review count is a convention, not a solution: it assumes the scales share a zero and behave alike across their range, and several do not, because in practice they are used only near the top. If a single figure is wanted anyway, it goes on another page with that assumption written beside it. ## 5. Reading the count The count moves before the score does. Where a channel computes a simple unweighted mean over every review it holds, a month's new reviews can move the displayed score by less than its displayed precision, and the score sits still while the experience underneath it changes. Most channels do not publish their method — some weight recency, some show a trailing window, some filter — so for those the figure in the last column of the second table cannot be derived and is written "method undisclosed". Where it can be derived, derive it. It tells you how blind that score is. The sheet supports readings, not verdicts. Count up and score level, count up and score down, count level, count down — each is a fact about the month, and the cause is not on this page. Type matters here: on a booking channel a zero is a statement about volume; on a review site it is a statement about whether anyone wrote. That is why the type column exists. ## 6. Rank and market size Rank moves without the property changing, because it is a position among others. Record **Market size** beside it every month. The channel defines the market, can redraw its boundary without notice, and does not have to say it did, so rank is comparable month to month only as far as that holds. Record both figures and let whoever reads them see the movement. ## 7. Numbers only This sheet records numbers. No reviewer name, no review text, no quotation, paraphrase or summary, no reply text, no guest identifier direct or indirect. Copying a review into an internal file makes the property the keeper of a record about a named person, held for its own purposes — with the questions about accuracy, retention and access that follow from that. Reading it where it was posted carries none of them. The form follows the rule: every field is labelled and numeric or a fixed word. There is no free-text column and nowhere to write one. A review that needs handling is handled on the channel, under whatever process the property already uses. ## 8. Benchmarks and conventions Where the brand requires a minimum score, that requirement is binding. Write it in the second table with its scale beside it, because a minimum without a scale is not a number. Everything else — a category average, a target lifted from a deck, a "good score is" figure — is an external convention to calibrate against this property's own measured series, not a fact to plan against. The house has a standing example. A par level built on a room count taken from a status report rather than from the room list made every par on seven floors wrong in the same direction at once, and nothing in the process showed it. ## 9. Questions to answer before first use - Which channels go on the sheet, and who decides when one is added or dropped? - What is each channel's scale and displayed precision, confirmed by looking? - What is today's count on each channel — the figure the first sheet carries? - How many days after month end is the checking window, and does the closing day fall inside it? - For which channels is the method published well enough to derive the blindness figure? - Who fills the sheet, and who lays it beside last month's before it is closed? - Where does a review needing a reply go, since it does not come here? ## 10. What this instrument does not do It does not tell you why anything moved. It does not read reviews, and a single serious complaint sitting inside a large count is invisible here by design — a real limitation, not a feature. It does not cover channels the property cannot see, and it will not catch a duplicated or misfiled listing except through a count behaving oddly. It cannot verify what a channel publishes; ranks especially are computed by a method the channel does not disclose, so that column records what was displayed and nothing more. The carry-forward makes a skipped month visible on the next sheet filled, but it cannot make anyone fill the next sheet. And two people filling it differently will still produce two different series, which is why the precision rule is in section 4 and not in a footnote. The sheet publishes blank. Every property's filled sheets are its own.