Your Own Account Is the Best Brief You'll Get This Quarter

You already know which of your last ten posts did the best. You can scroll the grid and see it.

You don't know why. Not in the way a quarterly plan needs.

Was it the product line? The format? The fact that it landed on a Tuesday at 8pm? Was the engagement carried by people who buy from you, or people who scroll past you forever?

Pull your own account. Anna can answer that in one session.

Step 1 — Pull the last 90 days of your own posts

"Pull the last 90 days of posts from @ourbrand on Instagram. Profile and posts. I'll classify them in a minute."

Anna pulls a Profile dataset (your follower count, verified status, baseline engagement) and a Posts dataset (caption, post type, views, likes, comments, shares, post date).

That's the raw material. Untouched, it's a spreadsheet. With one sentence to Anna, it's a brief.

Step 2 — Tell Anna how to classify them

This is where most teams stop. The data is right there, but nobody wants to manually tag 90 posts with category, sentiment, and product mentions. So nobody does. So the report says "engagement is up" and stops.

You don't tag posts. You don't write formulas. You tell Anna what you want and she does it.

"For every post, add four columns. Category — one of Product demo, Educational, Behind-the-scenes, UGC repost, Founder POV, or Promo. Product — the SKU or collection name, or None. Tone — Aspirational, Funny, Informational, Vulnerable, or Transactional. CTA — Yes or No. Save it back to the posts dataset."

Anna reads each caption, decides the answer, and writes the columns into your dataset alongside the originals. Ninety rows resolve in seconds. The columns are there waiting next time you open the spreadsheet.

Now the dataset has answers, not just rows.

Step 3 — Find what's actually working

"Group posts by category and tone. Show engagement rate per group. Flag the categories that beat the median by more than 30%."

This is the bit that rewires planning. The team thought product demos were the engine because they're the most common post. They're not. Behind-the-scenes content engages 75% better, but only made up 12% of the calendar.

The picture gets sharper when you put cadence and engagement on the same chart:

The next quarter writes itself: triple the BTS slot, halve the promo cadence, hold product demos flat.

You couldn't see this without classifying. Classifying by hand takes a week. Asking Anna takes lunch.

Step 4 — Read the comments on the top performers

The post-level read tells you what to make more of. The comment-level read tells you who to talk to.

"Pull comments on the top 10 posts by engagement. Cap at 100 comments per post."

Anna pulls a Comments dataset — comment text, author handle, like count, reply count, posted at. One row per comment.

Same trick. One sentence to Anna.

"For every comment, add three columns. Intent — one of Customer support, Fan praise, Question, Spam, or Creator pitch. Sentiment — Positive, Neutral, or Negative. Issue summary — if it's a support comment, the problem in under 10 words; otherwise blank. Save it to the dataset."

The Intent column is the one that pays for itself.

"Customer support" and "Fan praise" are the two filters of the brief you actually wanted. The 98 in the coral bar is what makes the next two steps non-optional.

Step 5 — Find your top engagers

"From the comments dataset, count comments per author across the 90 days. Surface the top 30 — the people who keep showing up."

The repeat commenters are the brief. Some of them are loyal customers. Some of them are creators with audiences of their own. Some of them are pleading for help in public and you missed it.

You can't tell which is which from a handle. You need their follower counts.

Step 6 — Enrich engagers with their own follower counts

"For each of those 30 handles, fetch the Instagram profile. I want follower count, verified status, and bio in a new column."

Anna fetches each profile in turn. A few seconds later, the comments dataset has a follower-count column.

Now it's not a list of names, it's a roster.

Three groups appear, every time:

The ambassador shortlist (top-right). People with five-figure-plus followings who keep commenting on your content because they actually like it. They are doing free promotion. They are also the people most likely to say yes to a gifted product, an affiliate code, or a paid post — because they're already engaging unprompted. Most teams discover these people through paid creator-search tools. They were already in your comments.

The loyal-fan cluster (top-left). Small accounts, high comment frequency. They aren't going to scale your reach. They are going to renew, advocate to friends, and answer "would you recommend" surveys with a 10. They're worth a personal reply. They are not worth a creator brief.

The support cries (bottom-coral). Negative or neutral sentiment, low follower count, often a specific product issue you didn't see when the comment first dropped. Two of them have been waiting four days for a reply. The issue-summary column Anna wrote already says what each one needs — "shipped wrong size," "can't see verification email," "app freezes on second prompt."

That last group is the one that costs you most when you miss it.

Step 7 — Hand the team three lists

"Make me three tabs. Tab 1: ambassador candidates ranked by followers × comment frequency × positive sentiment. Tab 2: loyal fans for me to reply to personally. Tab 3: open support cases sorted by oldest first, with the issue-summary column visible."

Anna writes the views. Each tab is a board you can share by URL.

The ambassador tab goes to whoever owns partnerships. The loyal-fans tab opens up on your phone and you spend twenty minutes replying. The support tab goes to whoever owns customer success — with the issue already summarised, so they aren't reading every comment from scratch.

Three tabs. Same dataset. One session.

What this replaces

• The agency that charges £2,000/month to manually pull and tag your last 90 posts
• The "social listening" tool that surfaces volume but not intent
• The intern building a spreadsheet by copy-pasting comments one screen at a time
• The DM you would have sent to a creator-search platform asking who would be a fit, when the answer was already in your inbox
• The customer-support escalation that happens when a public complaint gets quoted on Reddit three weeks later

You can also ask

The 90-day classify-and-enrich pattern is the wedge. Same toolset, related questions:

• "Which products in our catalogue get the most positive comments? Group by SKU mentioned in comments × sentiment."
• "Pull the last 30 days of comments. Bucket the negative ones by issue. Send me the top three issue clusters."
• "From this season's UGC reposts, pull the original creators' profiles. Who has grown the most since we featured them? They might be due for a follow-up."
• "Compare engagement on posts that mention our founder by name vs posts that don't. Same period, same length, same product mentions where possible."

Each one is a one-prompt rerun on the same data. The hard work was in step 2.

The take-home

Your own account is a longitudinal study you've been running for free.

Posts tell you what to make more of. Comments tell you who to talk to. The two halves only become a brief once they're classified — and the moment they are, the next quarter is half-written.

Paste your handle. Pull the last 90 days. Tell Anna how to label them. The shortlist is in there.