We ran 6,519 Instagram comments and DMs from a real brand account through our MintSense AI classification engine to understand what's actually happening in brand comment sections — how much is noise, how much is a real buyer, and whether brands can catch those buyers fast enough.
The most striking numbers: only 2.9% of comments showed genuine buying intent (192 comments), and 83% of those buyer comments went unanswered beyond 24 hours. They're buried in a comment section where 49% is general engagement, 23% is open questions, and 25% is spam. Without intent labels, buyer comments don't visually stand out from the noise around them.
We also measured classification speed (median: ~1.1 seconds per comment) and tested reply generation quality on a sample of real messages. Here's the full breakdown.
What's actually in a comment section
| Category | Share of comments (n=6,519) |
|---|---|
| General engagement (praise, chat, reactions) | 49.4% (3,219) |
| Questions | 22.7% (1,481) |
| Spam | 25.0% (1,627) |
| Buyer intent | 2.9% (192) |
Nearly half of all comments are general engagement — praise, emojis, casual chat. Questions make up about 1 in 4 comments, covering sizing, shipping, product details, and support. True buy intent stayed small — about 3 in 100 — which is why Instagram buyer intent is easy to miss when you're skimming a feed by eye, especially when 25% of the volume is spam noise.
What spam looked like in this mix
Among the 1,627 comments classified as spam, the most common patterns were:
- Emoji-only noise — strings of emojis with no substantive content
- Foreign-language generic praise — generic compliments in languages unrelated to the brand or post
- Celebrity worship — off-topic fan comments unrelated to the product
- Engagement bait — "check my page", "DM me", follow-for-follow requests
At 25% of total volume, spam is not just a minor annoyance — it's one of the largest buckets in a brand's Instagram inbox, often drowning out the 3% of comments that represent real sales opportunities. For more on why ad threads especially fill with junk, see why spam comments show up on Instagram ads.
What buyer comments look like
The 192 buyer-intent comments looked like what sellers already recognize: price checks, size and fit questions with purchase energy, shipping to another country, and "how do I order" style asks. They sit next to a much larger pile of general praise (49%), open questions (23%), and spam (25%) — so without intent labels, they don't visually stand out.
And 83% of those buyer comments went unanswered beyond 24 hours — not because the brand wasn't responsive, but because buyer comments blend into the noise and are easy to miss when you're manually scanning a busy comment section.
Want the practical checklist for spotting these in your own feed? How to tell if an Instagram comment is a buyer.
Multilingual coverage (and a gap)
The comment volume was predominantly English with a significant Russian minority:
| Language | Share of comments (n=6,519) |
|---|---|
| English | 87.3% (5,690) |
| Russian | 11.1% (724) |
| Other languages | <1% (105) |
This distribution reflects the brand's audience — a US-based account with an international Instagram following. The MintSense classifier handled both English and Russian comments with the same median ~1.1s classification time, and reply generation (tested on a sample of 15 messages) produced context-appropriate responses in both languages.
How fast can this actually be caught?
Classification speed matters if the goal is catching a buyer before they lose interest. Across all 6,519 comments, MintSense Fusion — our AI comment classification engine — labeled each comment with a median of ~1.1 seconds (p95 1.8s) — about a second from text in to knowing whether it's a buyer, a question, spam, or general engagement.
That speed is critical when 83% of buyer comments are going unanswered beyond 24 hours. Sub-second classification means you can surface buyer intent essentially the moment a comment arrives, before the customer loses interest and moves on to a competitor's post.
What replies looked like
We also ran a bounded reply sample on 15 real messages (5 buyer-intent, 5 questions, 5 spam) to test reply quality. The reply engine correctly:
- Generated replies for 10 messages (buyer and question intents) with brand-specific pricing, sizing, and shipping CTAs
- Skipped all 5 spam messages — no wasted AI calls or suggested replies for junk comments
- Matched input language for both English and Russian messages where substantive text was present
This is a qualitative check on reply appropriateness and spam handling, not a scored production benchmark. The key finding: the system knows when not to reply, which matters when 25% of your inbox is spam.
What this suggests
The pattern is clear: buyer intent is rare (2.9% of all comments),spam is abundant (25%), and general engagement dominates (49%). Without automated intent classification, buyer comments blend into the noise — they're short, direct questions ("how much?", "still available?") that don't look different from casual engagement or spam at a glance.
The cost of missing these comments is real: 83% of buyer-intent comments went unanswered beyond 24 hours. That's not a failure of effort — it's a signal-to-noise problem. When you're manually scanning hundreds of comments, and only 3% are buyers, they get lost.
The speed matters too: sub-second classification means you can surface buyer intent essentially the moment a comment arrives, before the customer loses interest and moves on to a competitor's post.
That's the case for intent-aware AI comment moderation for brands: automatically surface Instagram buyer intent, hide spam, and reply in the customer's language while the window is still open — before the sale is lost.
Methodology note
This analysis is based on 6,519 Instagram comments and DMs from a single active brand account, classified by our production MintSense Fusion engine over one week (August 1–8, 2026). All data shown here is aggregated and de-identified in compliance with our Privacy Policy: we report pattern categories, percentages, and performance metrics — never workspace names, customer names, business niche details, or verbatim message text.
Classification was performed in real time via webhook as comments arrived. Reply generation testing was performed on a bounded sample of 15 real messages (5 buyer-intent, 5 questions, 5 spam) to validate quality and spam-handling behavior. Language distribution (87.3% English, 11.1% Russian) reflects the brand's actual audience demographics.
This is one brand's Instagram account, over one week — not a cross-brand benchmark. The patterns we see here (low buyer-intent rates, high spam volume, and fast classification times) match what we observe across other production customer accounts, but your mileage may vary based on account size, content type, and audience.
Surface buyers. Skip the spam.
ReplyMint's MintSense classifies Instagram comments in about a second — surfacing buyer intent, matching language, and skipping low-value spam so your team replies while the sale is still open.
Start free — no card requiredFAQ
What percentage of Instagram comments are spam?
In our analysis of 6,519 Instagram comments, 25% were classified as spam — emoji-only noise, foreign-language generic praise, celebrity worship, and engagement bait. Real brand accounts typically see spam rates between 15-35% depending on follower size, content type, and whether they run paid ads. Larger accounts and those running ads see more spam.
How many Instagram comments actually show buying intent?
In our dataset, 2.9% of comments (192 out of 6,519) showed clear buying intent — asking about price, availability, how to purchase, or expressing direct interest in buying. This matches the pattern we see across D2C brand accounts: buyer comments are rare (typically 2-5% of total volume) but high-value, and easy to miss when buried in 25% spam and 49% casual engagement.
Do most brands reply to buyer comments on Instagram?
In our analysis, 83% of buyer-intent comments older than 24 hours went unanswered. Most brands struggle to catch buyer comments in real time because they look like regular short comments at a glance. A comment like "price?" or "still available?" blends into 25% spam and 49% casual engagement. Without intent classification, these high-value comments often get missed entirely — even by brands actively trying to respond.
How fast can AI classify an Instagram comment's intent?
Modern AI comment classifiers like MintSense Fusion can analyze intent in roughly 1 second (median 1.1s across 6,519 classifications). That means buyer comments can be surfaced to your team essentially the moment they arrive, before the customer loses interest or moves to a competitor — critical when 83% of buyer comments would otherwise go unanswered beyond 24 hours.
Can AI reply to Instagram comments in multiple languages?
Yes. The comment dataset included 87.3% English, 11.1% Russian, and smaller volumes of other languages. Modern AI reply systems can detect input language and generate context-appropriate responses (pricing, sizing, shipping) in that same language — critical for international D2C brands with global Instagram followings.