How to Review Buying Questions in Instagram Comments

1. Define what you will review

Choose known public product posts owned by the brand. Write down the post links and a short review period before you start. State whether you will review comments once or return for a later check.

A top-level comment sits directly under a post. A reply sits under another comment. This method uses public top-level comments. A person can inspect visible replies for context, but the collection input below does not export their text.

Write a short scope note with these items:

1. The public product posts you chose.
2. The reason you chose those posts.
3. The review period.
4. The maximum comments to collect per post.
5. The person responsible for checking product facts.
6. The normal support process for personal order details.

The reader question behind this guide asks about “qualifying leads from Instagram DMs and comments.” Here, qualifying means checking what a question needs before a useful reply. DMs are direct messages. They are outside this collection method. The supplied question comes from one vendor-affiliated public discussion. It shows that someone asked about this work. It does not show demand, product use, or willingness to pay. Reader question, published July 2, 2026; one vendor-affiliated observation.

Do not use likes, account names, or profile details to guess who will buy. Keep your decision tied to the words of the question and the product context.

2. Collect a small sample only when needed

An Actor is a hosted program that collects data. The documented fit of the Instagram Comments Scraper is to save public top-level comments from known public posts or Reels. It does not return DMs, reply text, purchase intent, or sales. These limits come from the supplied public contract check for build 0.1.21. No paid run was made for this guide. Public default build, checked October 25, 2026; contract inspection only.

Use this small input as a starting shape:

{
  "collectAllAvailable": false,
  "commentSortMode": "default",
  "maxCommentsPerPost": 10,
  "postUrls": ["https://www.instagram.com/reel/DIwcCZgxVZP/"]
}

Replace the sample Reel URL with a known public product post owned by your brand. The sample URL is not evidence about your product. This input passed a check against the public input rules. It was not run for this article.

`postUrls` holds the post links. `maxCommentsPerPost` sets a maximum of 10 unique top-level comments per post. Ten is a limit, not a promised result count.

`commentSortMode` set to `default` keeps the source’s default order. It does not promise the newest comments or a random sample. The other documented choices sort the collected sample by date.

`collectAllAvailable` set to false keeps the numeric limit. Setting it to true removes that limit and may cost much more. The run can still stop when the public feed ends, time runs out, or the source fails. Removing the limit does not prove complete coverage.

The Actor also accepts optional date fields, `commentsFrom` and `commentsUntil`. This example does not use them. If you add dates, save those settings with your review note.

A Task is a saved input preset for an Actor. The public Instagram Customer Feedback Comment Dataset Task, checked October 25, 2026; presence only can provide a starting page. Its public presence was checked. Its saved input was not read through an authenticated account, and it was not run. Inspect its settings and replace sample values before use.

3. Keep source fields separate from review decisions

Save the original export before you add labels. A field is one named value in a saved row. One row represents one collected top-level comment.

Keep these documented source fields:

These are Actor fields. Your question labels and approval states are separate manual fields. The public output contract lists the source fields above. Public default build, checked October 25, 2026; no runtime test.

Keep identifiers as text when you use a spreadsheet. An identifier names a record; it is not a number for calculation. For file handling, use the Instagram comment export guide. Keep this review focused on the question and the facts needed to answer it.

Remove repeated imports by `commentId`. Keep two comments with different IDs even when their words match. If an ID is missing, keep that row unresolved. Do not merge rows because both have a blank ID or the same text.

Preserve missing values. A null value means the value was unavailable. Zero is an observed numeric value. A null `repliesCount` does not mean zero replies. A zero reply count does not supply reply text. Neither value tells you what a useful product reply should say.

4. Read the post before you label the question

Open the original product post. Read its caption and inspect the product shown. Then check the current product page, price terms, stock information, or shipping terms needed for the question.

A short question can have several possible meanings. “How much?” might refer to one item, a bundle, or a variant shown in the same post. “Will this fit?” needs a clear product and a clear use. “Do you ship here?” does not name a destination.

For each comment, record what you checked and when. Include a source link where possible. If the post shows several products, write which product the question appears to concern and why. If you cannot tell, leave that choice unresolved.

Read visible thread context manually when it helps. A comment may refer to an earlier reply that is absent from the export. Do not invent that context. If the post or relevant reply is unavailable, record the gap and assign a manual check.

Use current product terms for the proposed answer. The comment date and collection time serve different purposes. The first tells you when the question appeared. The second tells you when the tool saved it. Neither date proves that today’s price or stock matches an earlier offer.

5. Label the request and name the next fact

Use a small set of labels. A label describes the information requested. It does not rate the person’s value or predict a purchase.

These are proposed analyst labels. An analyst is the person who reviews the rows. The Actor does not supply these labels.

One comment can need several labels. “How much is the blue one, and can you send it here?” needs both price and shipping checks. Keep one comment row and record both labels. Do not turn it into two buyers.

Write the next fact as a specific task. “Check price for the blue variant” gives the owner a clear job. “Follow up” does not.

Keep shipping destinations, order numbers, and other personal details out of a shared public-question table. If the work requires those details, use the brand’s normal approved support process. The public table can record that a private check is needed without copying the details.

An existing-order complaint also needs care. The comment reports an order problem. It does not independently verify the order or the person’s buyer status.

6. Build the manual reply-review table

Add review columns beside the unchanged source fields. Use enough detail for another person to understand the decision.

Use states such as needs context, needs product fact, ready for approval, approved, and action recorded. Also allow no product reply needed. These are your manual states. They are not Actor output fields.

A clear question can deserve a reply before you know whether the person will buy. Approve the answer on its accuracy and usefulness.

Give each comment a traceable decision. The giveaway review guide uses this same record-keeping principle for a different purpose. Here, the decision concerns the next reply action. Keep unresolved questions visible rather than forcing an approval.

7. Worked example: six comments, no confirmed buyer total

Original fiction: All comments, IDs, context, and review decisions in this example are invented for teaching. They are not Actor results or records from a real brand.

In this fictional case, a product post shows a standard item and a blue variant. The post does not state a price. The reviewer checks the fictional post but has not yet checked current prices or a commenter’s delivery destination.

The proposed actions remain separate from completed actions:

No posted reply is claimed in this example. A proposed action is not an approval, and approval is not proof that someone posted it.

The arithmetic counts comment rows:

• Product-question rows = 2 price rows + 1 shipping row = 3 rows.
• Existing-order service rows = 1 row.
• No-action compliment rows = 1 row.
• Unresolved rows = 1 row.
• Total = 3 + 1 + 1 + 1 = 6 comment rows.

The product-question share is 3 product-question rows divided by 6 reviewed comment rows. That equals 50% of this fictional sample’s comment rows. It is not a buyer rate or a purchase rate.

No confirmed buyer count is established. Do not write zero buyers: buyer status was not verified. Do not write three buyers: three rows ask product questions. The example also does not establish how many people wrote the six comments.

C3 still needs a destination check. The words “Do you ship here?” do not prove delivery eligibility.

8. Check gaps before you approve an action

Review each row for these failure states:

1. Missing source identity: Keep a row without a usable comment ID unresolved. Do not guess its identity from repeated words.
2. Unavailable context: Record a missing post, unclear variant, or unavailable reply. Assign a manual check or clarification.
3. Missing product fact: Hold a price, stock, delivery, or fit claim until the responsible owner checks a current source.
4. Personal service details required: Record only the approved support action in the shared table. Keep personal details in the normal support process.
5. Collection limit reached or coverage uncertain: Describe the saved sample only. Do not claim that it contains every product question.
6. No rows saved: Do not conclude that the post had no questions. Check the source manually before deciding what the empty result means.
7. Action status unknown: Leave the action unconfirmed. A draft or approval does not prove a reply was posted.

Before posting, the owner should open the comment again. Confirm the product context, the current answer source, and whether the proposed reply still fits. This is an action check, not a claim that the export contains all replies.

Keep later purchases and service outcomes in owner-only records with their own evidence. A public price question alone cannot support a sale attribution. Do not add a revenue score to make an incomplete record look certain.