TokSift
← Guides

How to Analyze TikTok Comments in Excel: Filters, Themes, and PivotTables

· 3 min read

A comment export is most useful when it answers a specific question. If you need to spot repeated questions, compare reactions to two videos, or decide what to answer next, you can do the first pass in Excel without treating an AI score as the whole story. This workflow uses the CSV or Excel file from TokSift's comment exporter and leaves a clear trail from each finding back to a comment row.

1. Export a public video's available comments

Copy the public TikTok video link, paste it into TokSift, and download the Excel file (or CSV if that fits your workflow). Start with one video and write down your question: for example, “Which setup problems appear more than once?” The file can include comment text, comment and parent IDs, timestamps, likes, reply counts, and language where available. TikTok's displayed count can be higher than the number actually delivered; use the exported rows as your denominator and see the comment-count troubleshooting guide before reporting coverage.

2. Keep a raw sheet and make a working copy

Save the original file unchanged. Duplicate its data into a working sheet and keep the source video URL, export date, and delivered row count beside your notes. If you open a CSV, check that emoji and non-Latin text display correctly and that ID columns remain text before editing or sharing. The supplied Excel file is a convenient starting point when you want to avoid CSV import choices.

3. Filter for the question you actually have

Turn the range into an Excel table, then filter rather than reading every row in sequence. For unanswered questions, search the text column for question wording and review the replies in context. For high-visibility concerns, sort by like_count or reply_count, but do not assume the most-liked row represents the whole audience. Use parent_id to distinguish replies from top-level comments, and use created_at to compare reactions over time. Keep the filter criteria in your notes so someone else can repeat the pass.

4. Add a small, reviewable theme column

Create a new column named theme. Read a manageable set of rows and apply a few labels tied to your decision, such as question, objection, request, praise, or other. Add a second column for action (answer, investigate, test content, or none). Do not label a comment solely from a keyword: sarcasm, replies, and mixed opinions need context. Review examples from every label before you summarize them. If you share the file outside the team, remove unnecessary personal fields and handle public commenter data responsibly.

5. Build a PivotTable, then return to the examples

Select the working table and use Insert > PivotTable. Put theme in Rows and a nonblank field such as comment_id in Values as Count. This shows how many delivered rows you assigned to each label. You can add language or a video identifier as a filter if you have comparable exports. A pivot is a map of your coded file, not a representative survey of all TikTok viewers. Open several underlying rows for each large category and record the exact questions or objections people used.

6. Make one decision you can check later

Write a short finding with its evidence and next step: “Several exported comments ask how to set up the product; test a 30-second setup video and check its follow-up questions.” For content planning, use the comment-to-content workflow. If you want to compare tone, the sentiment interpretation guide explains why a positive/negative percentage is not a substitute for reading the themes. Keep the source URL, exported row count, filters, label definitions, and a few row examples with the finding.

Quick checklist

  • One research question and a public video URL
  • Original export preserved; working copy labeled
  • Delivered row count recorded, not assumed from the TikTok counter
  • Filters and theme definitions written down
  • Pivot counts checked against actual comment examples
  • One next action with a date for review

FAQ

Can Excel automatically tell me what every comment means?

No. Sorting and pivots summarize the rows and labels you provide. Human review is important for context, replies, sarcasm, and ambiguous wording.

Should I remove duplicate commenters?

Only if your question is about unique people. A person can leave several different comments; deleting those rows would distort a question about comment themes. Keep your original data and state which unit you counted.

Why do my PivotTable totals differ from TikTok's displayed comment count?

The export may contain fewer retrievable public comments than the app shows. Filters, blanks, or your coding choices can reduce the pivot count further. Report the delivered row count and the filtered count separately.

Try it on your own video

Paste a TikTok link and get the comments as CSV or Excel. No TikTok login, and the first batch on every video is free.

Export comments