Agencies are asked the same question at the end of every Instagram campaign: what did people actually say? The reach numbers come from the platform, the engagement rate is arithmetic, and both are in the client's dashboard already. The comments are the only part of the campaign that contains language — objections, purchase intent, the phrasing people use when they describe the product to each other — and they are the part almost nobody extracts, because extracting them is tedious and the tools look like an engineering project.
They are not. This is the workflow: scope, export, stack, analyse, hand over. It fits inside a normal campaign timeline, costs less than an hour of billable time, and produces the slide clients quote back to you six months later.
Why the API route does not fit agency work
Start here, because it saves a conversation with whoever on the team wants to build something. Meta's Instagram Platform APIs return comments only for media belonging to a Business or Creator account your app is authorised for.
Look at a typical campaign: two brand posts, eight creator posts, three competitor posts you are benchmarking. The API reaches two of thirteen — and only after the client grants your app access to their account, which means Business verification, app review, token management and a person who owns it. Creators will not onboard your Meta app for a two-week collaboration. Competitors obviously will not.
Add the operational reality that rate limits are calculated per app across all your clients, so a busy month for one account degrades every other account you serve on the same app. For an agency, the API is a permanent-dashboard tool, not a campaign tool.
Step 1: scope the campaign properly
Before touching a tool, build a one-tab inventory. Columns: post URL, post type, owner (brand, creator name, or competitor), publish date, current comment count.
Two things fall out of it immediately. First, the largest comment count selects your tier, because caps are per post. Second, the inventory itself is a deliverable — most clients have never seen their campaign laid out with comment volume per creator next to it, and the ranking is frequently a surprise. The creator with the smallest following routinely drives the most comments per thousand followers, and that observation alone has justified the exercise on plenty of accounts.
Include competitor posts deliberately. A campaign report that says "your objection about sizing appeared in 14% of your comments and 31% of theirs" is a different class of deliverable from one that only looks inward, and it costs nothing extra — competitor posts are public URLs like any other.
Step 2: agree the question in writing
The single biggest difference between an export that becomes a retained account and one that becomes a CSV nobody opens is whether somebody wrote down what decision it informs. Four questions cover most agency work:
- Entrant verification. Who entered the giveaway, before the deadline, once each. The answer must be defensible if a losing entrant complains.
- Creator performance beyond reach. Which creators produced comments containing purchase intent rather than emoji, and what did their audiences ask about.
- Objection mapping. Which repeated objections should the next campaign, the pricing page or the FAQ answer first.
- Competitive benchmarking. What do people say in a competitor's comments that they never say in the client's.
Pick one as primary. The others come out of the same dataset for free, but the deck should answer one question well rather than four vaguely.
Step 3: choose the tier, and bill it through
Pricing is per post volume, not per client and not per month:
- Free: first 100 comments of a post, three exports per day, no signup. Use it to check the data shape before you quote — running one client post through the free tier takes a minute and lets you show real columns in the proposal.
- Pay per export: $1 per 100 comments, $2 minimum. Fine for a one-off audit of a single post.
- 3-Day Pass — $14: unlimited posts, up to 10,000 comments each. This clears the majority of campaigns end to end.
- Pro Passes — $49 for 3 days, $349 for 7 days, $1,499 for 30 days: up to 100,000 comments per post, with AI analysis over the exported comments. The 30-day option is the one to reach for during an always-on programme or a heavy campaign month across several accounts.
Every one of these is a one-time purchase that expires by itself — nothing recurs, nothing needs cancelling between clients, and there is no seat to reassign when someone leaves. That makes it clean to bill as a pass-through campaign cost. Because a pass covers unlimited posts for its window, two or three client campaigns running in the same week share one purchase; the only ceiling you have to respect is comments per post.
Step 4: export daily while the campaign is live
For each post in the inventory: paste the URL into the Instagram comment exporter, enable replies, download as CSV or Excel. Every export returns the same columns — author, username, avatar_url, text, likes, replies, created_at, language, is_pinned, id, reply_to_id.
Do it daily, not once at the end. Comments disappear continuously: authors delete them, moderators hide them, and Meta's automated spam enforcement removes them at the scale documented in its Community Standards Enforcement Report. On a giveaway post that enforcement can visibly thin the thread. Whatever was deleted before your single end-of-campaign export is gone, and no tool recovers it.
Daily exports buy a second thing worth more than completeness: a time series. With an export_date column you can show comment velocity by day, attribute spikes to specific creator posts, and pinpoint the hour sentiment turned. That is a chart clients have not seen before, built from files you were pulling anyway.
Enable replies whenever the mechanic touches them. Tag-a-friend, answer-in-the-replies and any thread where the brand responded all push meaningful volume below the top level — on a tag-a-friend giveaway, top-level-only exports routinely miss a third of the entrant list.
Step 5: stack everything into one sheet
Append every file into a single master sheet and add four columns of your own: post, owner (brand / creator name / competitor), campaign, export_date. Because the exporter's column set is identical across posts, platforms and dates, this is a paste rather than a transformation.
Those four columns are what make the analysis fast. Every question in step 2 becomes a pivot: comments per creator, objection counts by owner, entrants inside the date window, velocity by day. Without them you have forty files; with them you have a dataset.
One more habit worth adopting across the agency: use the same column names for TikTok, YouTube and Facebook exports, which arrive in the same shape anyway. A cross-platform master sheet lets you say "this objection appears in 12% of Instagram comments and 34% of TikTok comments" — a finding no single-platform report can produce, and one that usually reallocates budget.
The deliverables clients actually pay for
Five outputs, in rough order of how reliably they renew a contract.
The one-page theme count. Five recurring themes, comment count and combined likes for each, one verbatim quote per theme. An hour of work. Weight by likes as well as count, because nine comments carrying 4,000 likes and ninety carrying 200 are different findings and reporting them identically is the standard mistake. The full method is in how to analyze Instagram comments.
Creator scorecard beyond reach. Pivot the master sheet by owner. For each creator: comments per post, distinct commenters, share of comments containing a question, share containing purchase-intent language ("where can I buy", "does it come in", "how much"). Creators who look identical on reach separate sharply here, and the next brief writes itself. The equivalent TikTok workflow is in bulk TikTok comment export for influencer agencies.
Defensible entrant verification. Filter created_at to the entry window, pivot on username to enforce one entry per person, and keep the file. This is the deliverable that protects the client if a draw is disputed, and it is the one they will remember you for the first time it is needed. The draw mechanics are covered in the Instagram comment picker guide.
UGC and testimonial shortlist. Filter text for experience language — "I bought", "arrived", "been using" — then sort by likes. That is a ranked testimonial list with social proof already attached. Flag clearly in the handover that any use in paid media needs written permission and disclosure; the FTC endorsement guides are unambiguous about repurposed customer content in advertising, and it is the agency's job to say so before the client learns it the hard way.
The competitive language gap. Same theme count run against competitor posts. What their commenters complain about is your client's positioning opportunity; what their commenters praise is the feature comparison the client's page has not made yet.
Compliance and handover
A comment export is a file of personal data, and being public does not remove it from scope — usernames and comment text tied to identifiable people are personal data under GDPR, and the EDPB's guidance applies to publicly available data the same as any other. For an agency, four practices cover it:
- Name the roles. In almost every campaign the client is the controller and the agency the processor. Say so in the contract, along with what you may do with the data and for how long.
- Set a retention period and honour it. "Raw exports deleted 60 days after campaign close, aggregate report retained" is defensible. An undated folder in shared storage is not.
- Control access. A permissioned drive, not a link in a Slack channel three freelancers can still read after the project ends.
- Ship aggregates, hand over raw separately. The deck gets counts, percentages and quotes. The raw CSVs go to the client in one controlled transfer at close, and your copies come off the schedule you agreed.
One thing worth being direct with clients about: do not use exported comment data to build ad-targeting lists or to contact commenters. It is outside what anyone consented to, it is the fastest way to turn a goodwill campaign into a complaint, and platform terms are against you. Analysis, yes. Outreach, no.
Mistakes agencies make with comment data
- Exporting only the brand's own posts. The creator posts carry the honest reactions. The competitor posts carry the positioning. Both are public and both cost nothing extra to include.
- Waiting until the campaign ends. Unrecoverable. Export daily.
- Reporting a single sentiment percentage. Averaging a comment section produces a number describing none of it. Score inside themes and report per theme.
- Presenting volume as success. 60,000 comments where a few thousand accounts posted repeatedly is a mechanic working, not an audience responding. The username pivot answers this before the client asks it.
- Handing over a raw CSV as the deliverable. The file is the input. If the deliverable is a spreadsheet with no interpretation, the client concludes the tool did the work — which is a bad conclusion for a retainer.
- Losing the historical files. Next quarter's campaign is only measurable against this one if the data survived. Retention policies should specify how long to keep, not only when to delete.
Turning it into a retained line item
The one-off version of this is a nice campaign wrap slide. The retained version is a monthly comment intelligence report: export every brand post, every creator post and three competitor posts monthly; append to the master sheet; re-run theme counts; report what moved.
Month one is a baseline and reads as interesting. By month three it reads as a trend line, and trend lines are what get budget renewed. The objection that grew from 4% to 19% of comments over a quarter is a finding no reach report will ever surface, and it is sitting in files that took twenty minutes a month to collect.
Practically: one Pro Pass in a heavy month covers every client campaign running inside its window, and a $14 3-Day Pass covers a lighter one. Against a retainer line, that is close to zero — which is the actual reason comment intelligence is one of the highest-margin deliverables an agency can add.
Related reading
- Bulk exporting Instagram comments — the mechanics and volume ceilings
- How to analyze Instagram comments — the theme-count framework
- The agency guide to exporting TikTok comments — the same workflow on TikTok
- TikTok campaign reporting for agencies
- Exporting Instagram comments to Excel
- Instagram Comment Exporter — 3 free exports a day, no signup