A weekly comment report for agency clients
Clients get follower and reach numbers from the networks themselves. What they rarely get is a readable account of what people wrote under their posts. This is a report you can build once and refresh in twenty minutes a week, with every number traceable to a row.
Updated Nov 19, 2026 · 7 min read
What the report answers
- How many comments and replies came in, on which posts and networks, against last week.
- How many different people that was.
- How much of it the brand answered, and which questions are still open.
- What the most-liked comments said.
- Who the returning commenters are.
What it does not answer, and should not pretend to: reach, impressions, follower changes, or who liked and shared. A comment export lists who commented and what they wrote. Say so once in the report footer and nobody asks the sheet for more than it holds.
The weekly routine: collect
- List the posts the client published in the week, on every network they use.
- Export each one from the any-platform export page. Wait two or three days after a post goes up, so the comment section has settled.
- On each export page, note comments read and the count the network shows. You will quote both.
- Download the Excel workbook of each export and file it in the client's folder for that week. YouTube exports are deleted from the export page after 30 days and the others on their own schedule, so the downloaded file is your archive.
Free limits per post are YouTube up to 5,000; Instagram up to 1,000; TikTok and Facebook up to 300 comments. A client with bigger posts needs a paid read for those; the pricing page has the options. Budget for it in the retainer instead of reporting on previews.
The Data tab
One spreadsheet per client, with a tab named Data. Row 1 is the header of the Comments sheet (columns A to M, ending in platform) plus three columns you add:
| Column | Header | What goes in it |
|---|---|---|
| N | post | A short label you type, such as Tue reel: autumn menu. |
| O | week | The reporting week, such as 2026-W47. |
| P | question | A formula, below. |
Each week, paste the rows of every new export underneath (without their header rows) and fill N and O for them. Label the week by when the post was published, not by when each comment arrived, so a post stays in one week. The platform column already says which network each row came from.
In P2, filled down, a simple question test that works in Excel and Google Sheets:
=AND(H2="no", ISNUMBER(SEARCH("~?", E2)))
A sharper version with question words and buying intent is in find questions and buying intent with REGEXMATCH.
The Summary tab
Put the week being reported in B1 (for example 2026-W47), the previous week in C1, and the client's own handle in B2 (for example fernhill.market). Every formula below reads those cells, so next week you change B1 and C1 and the report recalculates. Copy each formula to column C for last week's figure.
| Row | Formula in column B |
|---|---|
| Comments and replies | =COUNTIF(Data!$O:$O, B$1) |
| From the audience (not the brand) | =COUNTIFS(Data!$O:$O, B$1, Data!$C:$C, "<>"&$B$2) |
| Brand replies | =COUNTIFS(Data!$O:$O, B$1, Data!$C:$C, $B$2, Data!$H:$H, "yes") |
| Questions asked | =COUNTIFS(Data!$O:$O, B$1, Data!$P:$P, TRUE) |
| Likes on comments | =SUMIFS(Data!$G:$G, Data!$O:$O, B$1) |
| On Instagram (repeat per network) | =COUNTIFS(Data!$O:$O, B$1, Data!$M:$M, "instagram") |
Different people who commented. Google Sheets:
=COUNTUNIQUE(FILTER(Data!$D$2:$D, Data!$O$2:$O = B$1))
Excel (Microsoft 365):
=ROWS(UNIQUE(FILTER(Data!$D$2:$D$20000, Data!$O$2:$O$20000 = B$1)))
Both count user_id, which survives a renamed account. Week-on-week change, in column D of each row:
=IF(C4=0, "", (B4-C4)/C4)
formatted as a percentage. Leave it blank when last week was zero; "up infinity percent" helps nobody.
Open questions: the part clients act on
A question is open when it is a top-level question and no reply under it came from the brand. Add column Q, open, to the Data tab:
=AND(P2, COUNTIFS(I:I, B2, C:C, Summary!$B$2)=0)
Then list them on a tab named Questions. Google Sheets:
=FILTER({Data!N2:N, Data!C2:C, Data!E2:E, Data!G2:G}, Data!O2:O = Summary!B1, Data!Q2:Q = TRUE)
Excel (Microsoft 365):
=FILTER(CHOOSECOLS(Data!A2:Q20000, 14, 3, 5, 7), (Data!O2:O20000 = Summary!B1) * (Data!Q2:Q20000 = TRUE), "none")
Each gives post, person, question and likes. Sort by likes: a question other people liked is one many people have. This list is usually the most valued page of the report, because the client can act on it the same day. How replies are linked to their comments is explained on the page about exporting comments with replies.
On TikTok and Facebook exports, check the Read me sheet of each workbook first. If it has a Replies not read line, some reply threads are missing from the file, and a question may show as open when the brand did answer it. Mark those posts in the report.
Top comments and returning people
The five most-liked audience comments of the week. Google Sheets:
=ARRAY_CONSTRAIN(SORT(FILTER({Data!C2:C, Data!E2:E, Data!G2:G, Data!M2:M}, Data!O2:O = Summary!B1, Data!C2:C <> Summary!B2), 3, FALSE), 5, 4)
Excel (Microsoft 365):
=TAKE(SORT(FILTER(CHOOSECOLS(Data!A2:Q20000, 3, 5, 7, 13), (Data!O2:O20000 = Summary!B1) * (Data!C2:C20000 <> Summary!B2)), 3, -1), 5)
Quote two or three of them in the written findings, word for word.
For returning people, a pivot on the whole Data tab with user_id and username in Rows and a distinct count of week in Values shows who has commented in the most weeks. The method is in how to find your regular commenters across posts. Name the top five in the report; clients like seeing that the same people come back.
Write it up and send it
Numbers alone are not a report. Above the table, write three sentences: what changed, what people asked for, what you recommend. For example: "Comments were up a third on last week, almost all on Tuesday's reel. Eleven people asked about delivery outside the city and six of those questions are still open. We suggest a pinned comment with the delivery area and a reply to each open question by Friday."
In the footer, state the coverage for each post: comments read out of the count the network shows, and the date it was exported. If any post was a preview or had unread replies, say which. The guide on checking an export is complete gives the checks to run before you sign off.
Send a PDF (File, Download, PDF on the Summary tab) or share the sheet as Viewer. Do not give clients edit access to the Data tab. It holds usernames and comments of real people: keep it inside the client's team, and delete it when the engagement ends.
When the client wants charts they can filter themselves, connect the same Data tab to a dashboard: build a comment dashboard in Looker Studio. For tone over time, add the labels from the sentiment guide as one more column.
FAQ
How long does the weekly update take?
About twenty minutes for a client with five to ten posts a week once the sheet exists: export, paste, fill the post and week columns, change the week in B1, read the open questions and write three sentences.
Can the sheet update itself?
Not with new comments. An export is the post as it was read. A live IMPORTDATA link can feed a tab without a download, but it serves that one export and ends when the export is deleted, so paste values into the Data tab for anything you need to keep.
Should brand replies count in the totals?
Report them separately. Audience comments show interest. Brand replies show service. Adding them together hides both.
What if a post has more comments than the free limit?
The export page shows a labelled preview and the options for reading the rest. Never report on a preview as if it were the whole post; state comments read out of the total.