If you sell a considered-purchase category on TikTok Shop Indonesia, where the buyer reads reviews and compares options before ordering instead of buying on impulse, the constraint that decided this year was not how much content you could produce. It was whether what the content says can be proved. The shift is measurable rather than a matter of taste. In YouGov's Southeast Asia content study, 60% of Indonesian consumers name honest creator reviews as the content they prefer — the highest of the three markets measured, ahead of Singapore at 55% and Thailand at 50%. After watching, 43% go on to visit the brand's site or social accounts and 34% actively look for reviews or other user content before deciding. That second number is the one that changes your brief. It means a sizeable share of your audience does not buy on the spot: they leave, they check, and they come back only if what they find supports the claim the video made. Indonesian academic work published this year reaches the same conclusion from the other direction — influencer effectiveness tracks credibility, authenticity and interaction quality rather than follower count, which is why brands have been moving toward niche but loyal micro and nano creators. An unboxing that shows the product opening is not evidence. It is a trailer, and your buyer is now going looking for the review.
Put the gap in orders before you decide who to brief. Take one considered-purchase SKU at Rp299,000 with 35% gross margin, so roughly Rp104,650 of gross profit per order. One hundred creators at 50,000 views each is five million views; at a 2% click rate that is 100,000 product-page visits. Now drop in the two conversion rates your own dashboard already reports for this SKU — the one attached to short unboxing clips and the one attached to longer review videos. If those are 1.2% and 3.0%, the same 100,000 visits return 1,200 orders from unboxings and 3,000 from real reviews. The gap is 1,800 orders, which at Rp104,650 each is about Rp188m of gross profit — set against a sample spend of roughly Rp15m for the same hundred creators at Rp150,000 each including shipping. Two days of screening work is not the expensive part of this plan. Briefing the wrong hundred creators is.
What changed in how Indonesian buyers decide
The instinct when conversion softens is to buy more reach, and on this market that instinct is now wrong in a specific way. Indonesian buyers are not scrolling less; they are verifying more. YouGov puts 76% of Indonesian consumers discovering beauty products through social media, but the purchase does not happen at discovery — it happens after a second step in which 43% visit the brand's own channels and 34% hunt for reviews. Meanwhile the platform's own gravity has shifted: TikTok Shop and Tokopedia together sit at 44.6% share against Shopee's 47.5%, with Shopee down 5.7 points from 53.2% a year earlier, and TikTok Shop usage in Indonesia has moved past being a young-person habit into a behaviour spread across age groups. A broader audience means a broader range of purchase decisions, and the expensive ones are decided on evidence. Content that only shows the product exists is now facing a buyer who has already seen fifty such videos and is looking for the one that answers a question.
Two rule changes tighten the same screw from the platform's side. Since June 2026 Indonesia's e-commerce regulation has pushed platforms to prioritise locally registered goods and required sellers to hold a business licence or a business identification number, which raises the floor on who is allowed to sell at all. Then on 14 October 2026 TikTok Shop tightened off-platform redirection across Indonesia, Malaysia, the Philippines, Singapore and Thailand, covering everything from product descriptions to profile bios. The practical effect is that the place where a claim gets checked is now the platform itself. You can no longer route a doubtful buyer to a page you control and talk them down there, which means the video has to carry the evidence on its own.
Step 1 — Measure your own two conversion rates before you brief anyone
Do not take the two rates off a benchmark article. Measure them on your own SKU, because the gap between them is what the whole plan is priced on. In Seller Center, split last ninety days of creator attributed orders by video length, or, if your tagging allows it, by brief type: clips you wrote as unboxing versus clips you wrote as review. Record orders per thousand product-page visits for each bucket. If the two numbers are within about twenty per cent of each other, your category is genuinely impulse-driven and none of what follows will pay for itself — stop here and keep buying reach. If the review bucket is meaningfully higher, write the two numbers down as the pair everything else is measured against, and re-run the split monthly, because it drifts as your price point moves.
One trap to clear before the scoring starts. Longer does not mean better, and on this market the format data actually argues against equating the two: 68% of Indonesian consumers prefer short video under three minutes, while only 27% prefer long video and 29% prefer live. A real review is not defined by runtime but by decision information — how long it was used, what it was compared against, and what is wrong with it. A ninety-second clip that answers those three questions outperforms a six-minute video that only narrates features, and if you screen on duration you will select for the second one and conclude that reviews do not work.
The three signals below reduce to a card with three rows per creator, and the card is what makes the list re-runnable next month instead of a judgement somebody remembers. register with DAMI and ask the team for the three-signal creator scorecard template so the scoring lives in one sheet with one owner rather than in the head of whoever did the outreach that week.
Step 2 — Score the share of real review videos in a creator's last thirty posts
Signal one is the share of a creator's recent output that is a genuine review. Open the last thirty posts and count only the ones that show the product in use after purchase rather than being unpacked on arrival. Write it as a fraction: twelve out of thirty is a different supplier than three out of thirty, even at the same follower count. Set your own floor — a workable starting point is that at least a third of recent posts are reviews rather than arrivals — and treat it as a filter, not a ranking. A creator below the floor does not get a considered-purchase brief at all; they get the impulse-tier commission and the unboxing brief, which is still worth running for reach.
Three phrases make the count quick and keep two people from scoring the same creator differently. Count a post as a review if it contains at least one of: how long the item has been used, a comparison against something the creator used before, or an explicit downside. The downside is the one creators resist and the one buyers trust most, so if you can only ask for one change in the brief, ask for that. A video that says what a product is bad at converts a checking buyer precisely because it is the sentence no trailer would include.
Step 3 — Read the comment section as evidence, not as engagement
Signal two is whether the creator answers questions in their own comments, and it is the signal most sellers skip because it looks like engagement rather than evidence. Scroll the comment section of five recent videos and count how many substantive questions — size, material, battery life, whether it suits a specific situation — received a specific answer from the account itself. Questions answered with a heart emoji do not count. A creator who answers twenty detailed questions a month is doing after-sales work you would otherwise pay for, and that work is exactly what the 34% of Indonesian buyers who go looking for comments are reading before they decide.
Turn the count into a rate and compare it against price rather than against an absolute standard. A nano creator answering fifteen questions a month at Rp150,000 of sample cost is cheaper evidence than a mid-tier creator answering five at ten times the fee, because the questions are where the hesitation gets resolved. Where a creator answers well but rarely posts reviews, pair them: give them the review brief anyway and accept that their value is in the comments rather than in the video. Not every creator in the pool has to produce the same asset.
Step 4 — Look for repeat-purchase feedback under a creator's own videos
Signal three is the hardest to fake and the slowest to accumulate: repeat-purchase feedback appearing under a creator's own videos. Look for comments from viewers who say they bought again, bought a second unit, or bought for someone else, and check whether the same creator has kept promoting the same SKU across more than one month. Both matter. A repeat comment means the first buyers stayed with the product, and a creator who keeps a SKU on their feed for a second month is telling you the first month worked — creators drop products that do not convert because their own income depends on it. This signal cannot be bought, which is exactly why it is worth more than follower count.
Two rules turn the three signals into a decision you can defend. First, make the shortlist a fraction, not a feeling: score every creator you consider on all three signals and keep the top third, so the list is reproducible when a different operator runs it next month. Second, revisit the floor quarterly rather than after every campaign. Signals two and three take months to accumulate, and if you re-screen weekly you will keep selecting for whoever happened to post in the last seven days instead of for the behaviour you actually want to buy.
How DAMI turns three signals into a shortlist you can re-run
DAMI brings an 8-million-plus creator database, competitor and competitor-store creator search, automated outreach tasks, AI-assisted outreach, AI video creation, sample management, campaign management, social media management, store analytics, and multi-store team dashboards into one place. Three actions produce the shortlist and keep it current, and each one ends in something you can open and check:
Source from stores already selling your category, then filter on review share
Use DAMI's competitor and competitor-store creator search to pull the creators already producing for shops that sell what you sell. Run signal one across the resulting list before anyone is contacted, and drop everyone below your floor. The output is a named candidate list where each row carries a review share fraction, not a follower number.
Keep the comment and repeat-buy evidence attached to the creator record
Track the creators you shortlist through DAMI's social media management so the comment answers and the repeat-purchase feedback stay attached to the same record as the brief and the sample. Signals two and three accumulate slowly, and they are worthless if they live in a screenshot folder. The output is a creator record with a three-row scorecard that a new operator can read without asking anyone.
Split commission into two tiers and let the brief decide which one applies
Run the review tier and the unboxing tier as separate campaigns in DAMI's campaign management, with different briefs and different rates rather than one average rate for everyone. The output is a monthly comparison of orders per thousand visits by tier, which is the pair of numbers you started with, now produced on purpose.
Frequently asked questions
Should I stop looking at follower count entirely?
No, but demote it to a floor rather than a ranking. Indonesia's affiliate threshold sits at 600 followers, so almost everyone you can work with clears it, and above that line the research points at credibility and interaction quality rather than reach. Use follower count to exclude accounts too small to sustain output, then rank on the three signals. Ranking on reach is how a review programme quietly turns back into an unboxing programme.
Do real review creators charge more?
Often yes, and the fix is to pay in commission rather than in flat fee. A creator who produces evidence is worth more per order than one who produces a trailer, so express the difference in the rate instead of negotiating a higher upfront cost you cannot attribute. Two commission tiers also let you test the split without committing to it for a quarter.
Does a review video have to be longer than three minutes?
No, and on this market the format data argues against it: 68% of Indonesian consumers prefer short video under three minutes and only 27% prefer long video. Score on decision information — usage time, comparison, and an explicit downside — not on runtime. A ninety-second clip with those three beats a six-minute video without them.
Should I drop the unboxing creators I already work with?
Do not, because they are solving a different problem. Unboxing output is cheap reach and it still feeds the top of the funnel; what it does not do is resolve hesitation for a buyer who leaves to check reviews. Keep both, split the briefs, and let the two-tier comparison tell you each month what the right mix is rather than deciding it once.
Can I score all of this in a spreadsheet?
Yes for one screening round, and it degrades quickly after that. A sheet handles thirty creators and one operator fine. What it does not handle is signals that accumulate over months while the operator changes, a dedupe across stores before every send, and a scorecard that has to be comparable between this month's run and next month's. The failure is not arithmetic — it is that the sheet stops being updated in the week the list is longest, which is the week you needed it.
Conclusion
Authenticity is the one input that decides whether your content is allowed to keep running, and everything above is about making every video hold up on the day somebody checks it. DAMI brings an 8-million-plus creator database, competitor and competitor-store creator search, automated outreach tasks, AI-assisted outreach, AI video creation, sample management, campaign management, social media management, store analytics, and multi-store team dashboards into one place.
Open your last ninety days of creator-attributed orders, split them by brief type, and write down orders per thousand product-page visits for unboxing versus review. Fifteen minutes, and it replaces the assumption this whole plan is priced on with two numbers you measured on your own SKU.
Start with DAMI and put the check where the content is approved rather than where the policy is stored.
