Start with the 12 most recent eligible posts, then check whether that set includes sponsored work. Record each post's date, format, topic, commercial evidence, fit with your brief, and one concrete observation. Twelve is a suggested review budget for an initial influencer content audit. It is not a validated sample size or a guarantee that you have seen everything relevant.
Use the same selection rule for every creator in a shortlist. Write it down before opening profiles, so an appealing first post cannot change which evidence you collect.
Declare what enters the sample
For an initial review, use this rule:
Review the newest 12 accessible feed posts published within the previous 90 days, ordered by publication date. Include sponsored posts. Count each post once, including a multi-image post. If fewer than three explicitly disclosed third-party collaborations appear, add the newest such posts outside the core sample, looking back no further than 180 days, until there are three in total. Keep those additions separate.
These counts and windows are editorial recommendations. Adjust them before reviewing candidates if your brief requires a different period or format. A seasonal campaign may need a separate same-season review. A video assignment needs evidence of video work.
Apply the rule as follows:
- Use publication dates, rather than screen position, when selecting posts. An older pinned post enters only if its date qualifies.
- If only seven posts qualify, record seven. Do not silently extend the window.
- If the selected period contains no disclosed collaboration, record that gap. Ask the creator for relevant examples.
- Exclude Stories and live broadcasts from this feed-post sample. If those are proposed deliverables, request a separate set of examples and record how the creator selected them.
- Log unavailable posts when you encounter them. Do not describe this as a complete record of deleted or inaccessible work.
This is a recent-work review with a targeted commercial supplement. Statistics Canada's guidance on non-probability sampling explains why non-random selection limits broader conclusions. A declared rule makes the review repeatable; it does not make these posts representative of a creator's whole career.
Modash's profile-review article recommends inspecting regular and sponsored content. The separate supplement helps you examine sponsored execution when the recent set contains little of it. Do not combine that supplement with the core sample to calculate a sponsorship frequency.
Build one row per post
Put the campaign question above the sheet. For example: "Can this creator demonstrate packing a reusable lunch container for an office commute?" That question gives reviewers something observable to assess.
Use these columns:
| Column | What to record |
|---|---|
| Post reference | Platform, profile, direct post URL and sample row ID |
| Dates | Publication date, review date and any date uncertainty |
| Selection | Core or supplement; reason for any exception |
| Format and topic | Video, photo sequence or image; one short topic description |
| Commercial evidence | Exact visible label or a short paraphrase of the disclosure; uncertainty separately |
| Brief fit | Direct, adjacent or unrelated, with a reason |
| Demonstration | What the creator shows happening, rather than a general quality score |
| Delivery | Whether speech, product details and the relevant action are understandable |
| Public response | A bounded comment observation, plus the sort order and number read |
| Metrics and gaps | Visible counts with capture time; unavailable fields marked unavailable |
| Follow-up | One question the evidence leaves open |
Keep observed facts separate from judgments. "The lid closes on camera" is observed. "This could explain our closure mechanism" is your assessment. "The audience buys lunch containers" remains unproven.
For comments, a workable rule is to read the first five displayed comments on each post, or all if fewer appear. Record the displayed order. Those comments can reveal a question worth asking, but cannot establish what the whole audience thinks.
Do not enter zero when a count is hidden. Do not infer audience location or age from appearance or language. TikTok Studio's documentation describes viewer and follower demographics within its analytics tools. A public content review does not establish those demographics. Use the missing-audience-data workflow to decide what to request from the creator.
Code commercial evidence without guessing the deal
Use four codes for this sheet:
- D: An explicit third-party commercial disclosure is visible.
- O: The post explicitly promotes the creator's own business. Keep this separate from third-party collaborations.
- N: No commercial signal was observed in the reviewed post.
- U: A brand mention, offer code or other signal needs clarification.
N means no signal observed. It does not prove the creator had no commercial relationship. D does not establish a cash fee or contract terms.
On TikTok, the content-disclosure help page distinguishes "Paid partnership" for another business from "Promotional content" for the creator's own business. TikTok's commercial-promotion rules cover payment and other incentives. Record the actual label instead of treating every commercial post as a paid third-party collaboration.
If commercial history becomes the deciding question, use the sponsorship-history review for a dedicated record. This sheet only needs enough classification to compare content execution.
Twelve illustrative coded posts
The following dataset is hypothetical. It describes no real creator, posts, metrics or audience. The review date is September 27, 2026. All 12 posts fall within the core window, and three have explicit third-party disclosures, so no supplement is needed.
The brief is the reusable lunch-container demonstration described above. Each row condenses the topic, commercial code, fit and observed execution columns.
| Post | Date | Format and topic | Code | Fit | Illustrative observation |
|---|---|---|---|---|---|
| P01 | Sep 25 | Video: packing office lunch | N | Direct | Shows filling, closing and placing a container in a bag |
| P02 | Sep 22 | Photo sequence: desk lunches | N | Direct | Shows portions clearly; no closure demonstration |
| P03 | Sep 18 | Video: lunch-container collaboration | D | Direct | Partnership label visible; shows opening and washing |
| P04 | Sep 15 | Video: weekend outfit | N | Unrelated | No food or commuting context |
| P05 | Sep 11 | Video: make-ahead grain bowls | N | Adjacent | Explains preparation; storage container stays off camera |
| P06 | Sep 08 | Photo sequence: insulated-bag collaboration | D | Direct | Partnership label visible; shows container placement inside bag |
| P07 | Sep 04 | Video: replies to lunch questions | N | Direct | Answers a question about fitting lunch into a work bag |
| P08 | Aug 30 | Image: cafe discount code | U | Adjacent | Code appears; commercial relationship unclear |
| P09 | Aug 25 | Video: portioning leftovers | N | Direct | Shows portions entering containers and lids closing |
| P10 | Aug 20 | Video: kitchen-tool collaboration | D | Adjacent | Partnership label visible; demonstrates chopping rather than storage |
| P11 | Aug 14 | Photo sequence: holiday scenery | N | Unrelated | No product demonstration |
| P12 | Aug 09 | Video: lunch after a commute | N | Direct | Opens a container on camera; journey itself is not shown |
This sample contains seven direct-fit posts, three adjacent posts and two unrelated posts. The three disclosed collaborations include two direct-fit examples. Those counts describe these 12 rows only.
The strongest follow-up concerns the proposed demonstration. P03 shows handling and cleaning, while P06 shows bag fit. Neither establishes that the creator has demonstrated your particular closure or tested leakage during travel. Ask for a proposed demonstration sequence before deciding on the assignment.
P08 needs a commercial-status question if that relationship matters to the brief. The unresolved code should remain visible in the sheet.
Decide whether to continue the review
Finish with an action and its evidence:
- Advance when relevant execution appears repeatedly and the proposed format has a usable example.
- Request evidence when the creator looks relevant but a required action or format is missing.
- Stop when the declared sample lacks the content your brief needs. Record the mismatch without making a judgment about the creator's overall ability.
If you extend the review, name the unanswered question and the additional selection rule first. Avoid browsing until you find a post that supports your preferred decision.
Create the sheet, write your selection rule at the top, and code the first candidate before opening the next profile.
