FARMRun a weekly TikTok review that improves the next batch
Compare formats, find where viewers lose interest, check publication problems and turn a weekly report into three concrete production decisions.
Finish your weekly review with three decisions: what to make again, what to change, and what to stop spending time on. Review individual posts and formats before judging the whole account.
Keep the meeting short by bringing one table of comparable posts, the audience's recurring questions and any publication problems. The examples below use invented data to show the decisions.
Bring six fields for each post
Start with the account, post link, format and hook, publication date, results at a chosen age, and production time. Choose a consistent review age—seven days, for example—and keep newer posts in a separate group.
The most useful results depend on the question:
| Decision | Look at |
|---|---|
| Is the opening clearer? | Available early retention and average watch |
| Is the explanation useful? | Saves, shares and specific follow-up questions |
| Does the format attract potential customers? | Product questions, qualified visits and later activation or orders |
| Can we keep producing it? | Creative hours, revision time and available new topics |
In PhoneFarm, use populated Analytics account information and Content post snapshots to locate results. Check their update time; collect missing detail from the platform's reporting. Mark an unavailable metric as missing rather than zero.
Compare the typical post with the standout
Suppose four interview-tip posts have 800, 1,200, 1,600 and 16,000 views at seven days. Total views are 19,600, but the median is (1,200 + 1,600) / 2 = 1,400. The standout is worth studying; it should not become the assumed result of every future post.
Put format results beside effort and audience intent:
| Format | Posts | Median views | Relevant product questions | Creative hours | Next decision |
|---|---|---|---|---|---|
| Interview mistake → better answer | 4 | 1,400 | 12 | 3 | Produce three new questions in this structure |
| General motivational text | 4 | 3,000 | 1 | 2 | Test a topic closer to the app's actual use |
| Feature tour | 2 | 600 | 4 | 4 | Shorten to one useful action before making more |
The motivational posts reach more people; the demonstrations generate more specific interest. Your next investment depends on whether the goal is awareness, product discovery or activation. Read the comments themselves before interpreting a count as buying intent.
Diagnose where the next batch needs work
| Pattern | Next week's action |
|---|---|
| Low viewing across several openings | Recheck audience problem and first frame; use the hook test |
| Opening holds, explanation loses viewers | Remove repeated lines and bring the demonstration forward |
| Strong viewing, little product interest | Test a more relevant problem or clearer product reveal |
| Product questions, few visits | Inspect the CTA and profile destination on a phone |
| Visits increase, activation or sales stay weak | Inspect onboarding, offer, price and the promised next step |
| One format weakens while others hold | Refresh that format's substance |
| Unrelated formats drop on one account | Check reporting, actual posts and platform notices before changing infrastructure |
Compare against the account's ordinary recent results, not only its biggest hit. When a change looks promising, repeat it before expanding the production commitment.
Check delivery without turning the review into a support meeting
If 12 posts were planned and 11 appeared, report 11 / 12 = 91.7% completion. If two of those eleven required another attempt, record the extra work as well. One post with two attempts is still one post.
Separate three operational facts: unfinished publications, extra operator time and the recurring reason for rework. Assign the fix through the relevant troubleshooting guide, then return to the creative decisions. Missing posts belong in the delivery report, not as zero-view observations in a format comparison.
Watch for the real bottleneck. If approved videos wait for a free phone, capacity may help. If the editor has no finished exports, adding phones will leave the same queue elsewhere.
Keep rates comparable
Write what each percentage divides. For example, (likes + comments + shares) / views differs from the same interactions divided by reach.
If one post has 10 interactions from 100 views and another has 90 from 9,000, their combined rate is 100 / 9,100 = 1.10%. Averaging their individual rates gives 5.5%, which answers a different question. Keep the same definition when comparing weeks.
For business outcomes, follow the installs and sales guide. Profile visits, installs and orders are separate steps; adding them together produces a number with no useful interpretation.
Leave with the next batch already chosen
A completed review could read:
Make three more mistake-to-better-answer demonstrations. Test showing the recording screen earlier. Pause general motivation posts until we have a product-relevant angle. Resolve the one unfinished publication. Review the new posts at seven days.
Put those actions directly into the content calendar, with the person doing the work and the next review date. Keep the table so next week's decision starts from what you learned.