instagram-audit

Most founders look at follower count and likes and guess. Both mislead. A post can get many likes from people who already follow and never buy. This audit asks two questions instead: are we reaching new people, and are those people doing something (saving, sending, visiting the profile, messaging, buying)?

It reads numbers the user exports. It never scrapes, logs in, or guesses numbers.

1. Get the numbers

Ask for one of these, in this order of preference:

1. The Insights export from Meta Business Suite (Insights, then Content, then Export) as CSV, for the last 30 to 90 days. Works for business and creator accounts.
2. A pasted table: date, format, reach, likes, comments, shares, saves, follows, profile visits. Whatever the user has.
3. Screenshots of Insights: read the numbers out, and say which ones you could not read.

Also read log.md and old plan.md files for the theme of each post, and the business file for goals (references/business-files.md). An audit without goals can only say what is popular, not what is working.

If the user has no numbers at all, do not invent any. Offer a review you can do now: the profile (/instagram-profile) and the last few captions they paste. Then explain how to export Insights for next month.

2. Run the numbers

python3 scripts/insights.py export.csv --log ~/.founder/social/instagram/<slug>/log.md

For each post it works out shares plus saves per 100 people reached. It groups posts by format and theme, lists the top and bottom posts, and flags a small sample (under 8 posts). It also says which columns the export did not have.

Why shares and saves per reach? People send and save a post when it is worth something to them. Dividing by reach makes a small post and a big post comparable.

3. Read it, honestly

Report only what the numbers support:

• Reaching new people? Reach compared with follower count, and which formats reach furthest. If the export shows followers vs non-followers, use it.
• Worth something? Shares plus saves per 100 reached, by format and theme. Which themes people keep.
• Moving people? Profile visits, follows, link taps, DMs. And, from the user, what matters most: seats sold, trial signups, preorders. Ask for those if not given.
• Top 3 and bottom 3 posts, with a plain guess at why. Label it a guess: "likely the hook", not "because of the hook".

Rules:

• Small samples are hunches. With under 8 posts, or one post far ahead of the rest, say so and keep the advice light.
• No made-up benchmarks. Do not say "a good save rate is 2 percent". Compare the account with itself: this month vs last, this format vs that one. If the user wants an industry number, it needs a source with a link and a date.
• Raw totals are context, not findings. A 10x reach month with no new follows is a finding. A 10x reach month alone is not.
• Reach drops have dull causes first: fewer posts, fewer Reels, a quiet season, a weaker hook. Check those before blaming the algorithm.

4. Keep, stop, try

Three of each, tied to the goal in the business file:

• Keep: what is working, with the number that says so.
• Stop: what costs time and returns little, with its number.
• Try: one test per line, with what you will measure next month.

A keep or stop with no number behind it is a hunch. Move it to try. Fewer than three is fine when the data is thin. Say so instead of filling the slot.

Every number in the scorecard comes from the tool's output or the user. If you work one out yourself, show the sum.

For a local business, also ask three things. Are posts reaching people nearby (location data in Insights, if shown)? Did the offer posts lead to visits or orders (ask the user)? Which collabs helped?

5. Scorecard

End with a small table to beat next month, using only numbers from this audit:

SCORECARD  ·  <month>
posts            <n>       next month aim: <n>
median reach     <n>
shares+saves/100 <n>
follows          <n>
profile visits   <n>
goal metric      <n or {{from you}}>   (e.g. founding seats sold)

6. Output and files

INSTAGRAM AUDIT  ·  <business>  ·  <date range>  ·  <n> posts
data: <file>, columns missing: <list>
sample: <ok / small, read as hunches>

REACH ...
WORTH ...
MOVING ...
TOP 3 / BOTTOM 3 ...
KEEP / STOP / TRY ...
SCORECARD ...

Keep the wording plain; run it through references/before-you-show.md.

Save to ~/.founder/social/instagram/<slug>/audits/<YYYY-MM-DD>.md. Next month, read the last audit first and compare.

This is a read of the user's own data, not financial advice. If the user is deciding on ad spend or hiring based on it, say to sanity check the numbers with whoever does their books.

Reads: the export, log.md, plan.md history, business file, the last audit. Writes: audits/. Next: /instagram-plan for next month, /instagram-profile if visits do not become follows.