Instagram Creative Research Shortlist Workflow#

Write one creative-research question, use one visible metric to narrow the currently loaded profile set, and then open every source post. Record visible facts, human observations, do_not_inherit notes, and an original angle separately. The metric selects examples; it does not prove hook strength or creative quality.

INS Sorter temporarily reorders loaded cards on one supported Instagram profile grid in the current Chrome session. It does not read transcripts, download media, export a research library, access private Insights, save the shortlist, or alter the public grid.

Set the question and sample#

The output is a reviewable shortlist of five to eight source-linked posts, not a competitor report. Fix one profile, the loaded-card count, observation time, one sort field, and one narrow question.

Likes, Comments, Views, and Date can locate candidates. They cannot independently establish retention, shares, conversion, originality, or creative quality. For multiple profiles, repeat the workflow separately and retain each boundary.

Copyable source-linked shortlist template#

CopyRESEARCH QUESTION:
PROFILE:
OBSERVED AT:
LOADED SCOPE: ___ currently loaded cards
SORT FIELD / ORDER:
 
Candidate | Source URL | Visible field/value/date | Visible format
 
For each opened source post:
- visible fact:
- human observation:
- visible hook / format / proof:
- do_not_inherit:
- original angle for our audience:
- reviewer status: keep / hold / reject
- rejection reason:

Six-step workflow#

1. Ask one question. Do not research hooks, editing, CTA, and offers at once.
2. Record the loaded-card count and observation time. Leave missing fields missing.
3. Sort one supported profile by one field and take five to eight candidates, plus a middle or low-signal counterexample when useful.
4. Open every source post and record only visible format, copy, scene, or proof.
5. Separate facts, observations, and original angles. Put brand assets, exact wording, people, and unverified claims in do_not_inherit.
6. Keep only rows that answer the question. State missing evidence for hold and a concrete reason for reject.

Verification checklist#

• Every row has an openable source URL.
• Every row uses the disclosed loaded scope, field, and observation time.
• A metric never becomes a creative-quality conclusion.
• Facts, observations, original angles, and prohibited inheritance remain separate.
• The brief copies no exact claim, brand identity, or person.
• Five to eight retained rows answer the same research question.

Limits#

The loaded set is not complete history. Public counts change, hide, or go missing, and do not reveal saves, shares, retention, reach, conversions, or paid distribution. Sorting does not replace opening the source; visible comments and captions may also be incomplete.

Use a separate tool when the task truly requires transcripts, media downloads, bulk export, a shared research library, or owner-only metrics. INS Sorter does not claim those capabilities.

Next step#

Use the INS Sorter product page to narrow one current profile, then open each source and complete the template manually. Other cluster pages are not public yet, so no sibling link is included.

Frequently asked questions#

Should I sort by Views or Comments?#

Choose the field that matches the question. Comments can shortlist visible-discussion candidates; Views are only a visible playback signal. Neither proves quality.

Can I send the highest-ranked posts directly to a creative team?#

No. Open each source first and separate visible evidence from reviewer interpretation.

Why add do_not_inherit?#

It makes the boundary between learning a structure and copying a specific expression reviewable.

Does the tool save my shortlist?#

No. Sorting is a temporary local view; save the template yourself.

Sources#

• BrandSearch organic competitor-research workflow
• FeedRama Instagram competitor-research workflow
• VideoToTextAI Reels creative-research workflow

SKU10 boundaries were checked against the local runtime and product contract on August 31, 2026. No login, real-account collection, deployment, or publication was performed.