In short
- Sample 100 comments from ten posts, five sponsored and five organic, and code each into one of seven types. Then compare the two halves of the sheet.
- The number that matters most is purchase signal: questions about price, size, where to buy and how to use. Compare it between organic and sponsored posts.
- Read replies as well as top-level comments, and watch what the creator does. Answering questions is a craft signal; hiding them is a red flag.
- Comments cannot prove fraud. Use the sheet with views, saves and growth, and treat a weak sponsored column as a pricing conversation, not a verdict.
To analyse influencer comment quality, take 100 comments from ten recent posts, five sponsored and five organic, code each comment into one of seven types, and compare the mix on the sponsored posts with the mix on the organic ones. The share of comments that carry a purchase signal, meaning a question about price, size, where to buy or how to use, tells you whether this audience answers when a brand speaks.
Our guides to how to spot fake followers and engagement pods explained both use comments as one test among several. This post goes one level deeper on comments alone, with a sampling plan, a coding sheet and a method for reading the result. It is check B3 in the scored sheet in our influencer vetting checklist, with parts of B5 and C4, and it sits under our complete guide to influencer marketing in India. Use the engagement rate calculator for the headline number first, then this sheet for what is inside it.
Draw the sample
A sample that only shows the first hour or only the top-liked comments misleads. Draw it the same way every time.
| Slot | Posts | Comments from each | Why |
|---|---|---|---|
| Sponsored | The last five, same platform | 5 from the top of the list, 5 from further down | Shows how the audience answers a brand |
| Organic | Five of the last 20, same format as the sponsored ones | 5 from the top, 5 from further down | Gives the creator's own baseline |
| Replies | Two comment threads per post | Open the replies | Shows conversation and the creator's behaviour |
Write each comment into a sheet with the post, the type code and a flag. Note two things before coding: whether comments are off or restricted on any post, and whether the visible comments are far fewer than the count shown. Apps let creators filter words or hide comments, and labels change, so a gap is a reason to ask, not proof of anything. A hundred comments is a screen, not a statistical test.
The seven codes
Give each comment one code, and add the flag R if the same account has commented on three or more of the ten posts.
| Code | Type | Looks like | What it suggests |
|---|---|---|---|
| S | Specific | Refers to a detail: "the second shade works on my skin" | The commenter watched and thought |
| Q | Question or purchase signal | "Price?", "which size", "link please", "is it in Jaipur" | A viewer moving towards a purchase |
| T | Tag or share | "@name look at this" | Sharing intent; useful for reach, silent on buying |
| G | Generic | A fire emoji, "nice", "love it" | Passing approval or a template comment |
| C | Conversation | Replies to the creator or to another commenter, over two or more turns | An audience that talks to each other |
| X | Spam or self-promotion | "Follow me", "DM for collab", betting links | Noise, and a brand-safety note if it is heavy |
| N | Sceptical or negative | "Is this paid?", "did not work for me" | Honest friction; read it before the brand does |
A sceptical "is this paid?" under a post that carries a label may mean the label was not noticed. That is an inference, so check where the label sits against ASCI's placement rules in our ASCI influencer guidelines explainer.
Five readings from the sheet
Compute each reading separately for the sponsored half and the organic half.
| Reading | How to compute it | What it suggests |
|---|---|---|
| Purchase-signal share | Q divided by comments in the half | Whether viewers want the product |
| Specific against generic | S compared with G | Whether comments carry information |
| Sponsored drop | Sponsored Q share minus organic Q share | Whether the audience goes quiet for ads |
| Repeat-commenter share | Comments with the R flag divided by comments in the half | A tight fan circle or a pod; see the pod tests |
| Conversation share | C divided by comments in the half | Community depth |
We found no published India benchmark for the mix of comment types, so we do not quote a good figure. Compare shortlisted creators with each other, in the same niche and format, and compare each creator's sponsored posts with their own organic ones. In a 2025 study reported by MediaNews4u, 36% of marketers said content quality is a better performance indicator than reach, and comments are where that quality shows.
Watch what the creator does
Comments are half the picture. The other half is the creator's reply.
| Behaviour | Reading |
|---|---|
| Answers product questions within a day, with specifics | A craft signal; the post works as a sales assistant |
| Pins a useful comment or a frequently asked question | Good habits for a brand post |
| Replies only to praise | Selective; ask how they handle questions about a product |
| Questions vanish from view while the count stays high | Possible hiding; ask |
| Replies "check DM" to every price question | Normal for some; ask what happens in DMs |
Brand teams carry some of this work. Agree in the brief that comments stay on, that the brand's own team answers product questions in the first hours, and that anything about safety or side effects goes to you first. Our guide to social listening for influencer campaigns sets out how to watch comments during a live campaign.
Read in the audience's language
Codes depend on understanding the comment. Hinglish, Tamil, Telugu, Bengali and other regional-language comments are easy to mislabel as generic when the reader does not know the idiom, and a native reader will also spot template comments translated word for word. For regional creators, have someone fluent in the language do the coding, as in the adjustments table in our vetting checklist, and keep a note of any slang that the sheet's codes cannot capture.
Adjust for format
Compare like with like. Reels tend to draw short replies and emojis, while long YouTube videos tend to draw longer comments, so do not mark a Reel audience down against a YouTube one. For a LinkedIn creator, read for the substance of the reply and check for pods, which our pod post covers. Small accounts have few comments, so widen the sample to more posts, and run a paid trial before drawing a firm conclusion.
A worked example
Red flags
- Comments off on sponsored posts only.
- Dozens of comments inside the first minutes, then nothing for hours.
- The same comment text across several posts.
- Commenters whose profiles have no posts, no photo and random handles.
- Comments in a script or language the audience never uses.
- Praise that repeats the brand's phrasing word for word.
- A visible count far above the comments you can scroll to.
What to ask the creator
Ask for evidence the public view cannot give. "Which of your last ten posts drew the most questions, and what were they?" shows what the audience wants. "Could you screenshot the five questions you get most often in direct messages?" shows purchase intent that never reaches the comments. Ask whether any word filters or hidden comments are in use, and for a screenshot of the comments on one past brand post you are proud of. Say you ask every creator the same, and send the wording from our outreach message templates.
How the Gully Score uses comments
Engagement quality carries 25 points of the Gully Score, and buyer fit carries 20. In the sheet, B3 (comments are specific, on-topic and in the audience's language) carries 6 of the 25, B5 (no pod signature) carries 2, and C4 (comments show purchase intent) carries 4 of the 20 buyer-fit points. Those sub-point splits are our suggested split for a manual sheet, and the criterion totals are the Gully Score's own.
The sheet also shows why a creator can score well on engagement rate and badly on engagement quality. Our guide to Instagram engagement rate benchmarks covers the first, and this sheet covers the second. Comment quality is one of the engagement checks in every Gully Score we run. If you want to see it applied to a shortlist for your category, our Instagram influencer marketing service is where we start, and you can send us a brief to get a scored plan within 48 hours of the brief call.
Sources
- India's influencer marketing industry set to grow 25% in 2025: WPP Media and Kantar report, MediaNews4u (11 Jun 2025)
- Indian consumers and influencers: coverage of a 2025 Kantar-backed study of consumers and brands, Campaign India (12 Jun 2025)
Frequently asked questions
How do I analyse an influencer's comment quality?
Take 100 comments from ten recent posts, five sponsored and five organic, five from the top of the list and five from further down on each. Code each into one of seven types, then compare the share of questions, specific comments and generic comments between sponsored and organic posts. Our vetting checklist shows where the result scores.
How many comments should I read?
A hundred is enough to see a pattern and too few to prove one. Read more if the first sample is borderline. We use 100 as a working sample, not a statistical standard, and we compare each creator with others on the shortlist and with their own organic posts.
What is a good share of purchase-intent comments?
We found no reliable published India benchmark, so we do not quote one. Compare creators of the same niche with each other, and compare a creator's sponsored posts with their organic posts. A sharp drop on sponsored posts is the useful finding.
Are comments in emojis a bad sign?
Not on their own. Many real audiences reply in emojis, especially under Reels. They become a concern when emoji and one-word replies make up nearly every comment on paid posts but not on organic ones, or when the same accounts leave them every time. See engagement pods explained.
Can comment analysis prove an influencer has fake followers?
No. It is one signal. Comments from real accounts can still be low quality, and fake accounts can write convincing comments. Combine it with reach against followers, growth history and audience location. Our guide to how to spot fake followers shows how to weigh them together.
What if comments are turned off on a sponsored post?
Ask why. A creator may switch comments off after abuse, or to avoid questions about a product. If comments are off on paid posts only, treat it as a flag. Ask the creator for the questions they received in direct messages, and agree in the brief that comments stay on and that questions get answered.