The Question Behind “We Got 40,000 Views”
A multi-store seller running home goods out of Shenzhen sent me a screenshot of two videos. Both had done roughly 40,000 views in seven days. One had produced 214 orders. The other had produced 6. He wanted to know what was different about the second video.
Nothing about the second video was different. What was different was where the views came from. The first video drew 83% of its views from search — people typing a product phrase and landing on it. The second drew 91% from the For You feed, which is reach without intent. Same view count, opposite commercial meaning, and the second video was not broken. It was simply answering a different question than the seller was asking of it.
This article is for the operator running TikTok Shop in more than one market — usually a US shop plus two or three Southeast Asian ones. You carry 50 to 200 SKUs, and your team posts its own product videos alongside creator content. If you report “views” upward as one number, you are reporting a number that hides the only thing worth knowing.
It is a diagnostic, not a growth playbook. We are not going to argue that you should chase more views. We are going to argue that the split behind the views tells you which of three problems you actually have. Treating all three the same way is why some teams keep re-editing a video that was never the problem.
Why One View Count Hides Three Problems
A view is a view regardless of how the viewer arrived. But the arrival route determines what the view can reasonably be expected to produce, and the three main routes behave differently enough that a blended number is close to useless.
Search traffic arrives with intent already formed. The viewer typed or tapped a query, which means they were already looking for something in your category. These views convert at a higher rate and, more usefully, they keep converting for weeks because search demand does not decay the way feed distribution does.
For You traffic arrives with intent unformed. The platform decided to show it. The viewer may have zero interest in your category and scroll past in a second. These views are cheap and plentiful and convert poorly per view — but they are the only route that can put you in front of someone who was not looking for you.
Following and profile traffic arrives with intent already won. Someone came back to your account or a creator’s account to look for you. The volume is small, the conversion rate is high, and it is the slowest of the three to build.
The Three Routes Side by Side
| Route | Intent on arrival | Typical conversion behavior | Reliable after |
|---|---|---|---|
| Search | Already formed | Converts early and keeps converting for weeks | 7 days |
| For You | Not formed | Cheap reach, low conversion per view | 2–3 days for reach, 7+ for orders |
| Following / profile | Already won | Highest per view, smallest volume | Months of consistent posting |
The failure mode is reading a 40,000-view video as “working” or “not working” without knowing which of the three it was. A search-led video with 40,000 views is a compounding asset. A feed-led video with 40,000 views is a lottery ticket that already paid out or did not.
The Diagnostic: Read the Split Before You Touch the Video
The order of operations matters. Most teams edit the video first and read the traffic second, which means they fix the wrong thing and then conclude the channel does not work. Read first, edit second.
Pull the traffic split for every video in the last 30 days that crossed 5,000 views. Put them in a sheet with four columns: views, orders, the traffic split, and the category. Then sort by conversion rate rather than by views. What you are looking for is not the best video — it is the pattern across your best-converting videos.
| Traffic mix | Usually means | What to do next | When this reading is wrong |
|---|---|---|---|
| Search >50% | Demand exists; you are capturing it | Make more videos on the same query phrases | If the category is seasonal, the demand may vanish next month |
| For You >70% | Reach is working, intent is not | Fix the product-link moment, not the hook | If the video is <7 days old, the mix has not settled yet |
| Following >30% | Audience is forming | Post on a fixed cadence to keep it | If one creator drives it, you have a creator dependency, not an audience |
| Flat across all three | No route has traction | Test three different opening frames, not three edits of one | If the account is <30 days old, the sample is too small to read |
The important caveat: a video younger than seven days has an unstable mix. Feed distribution often front-loads and then decays while search accumulates slowly. Reading a two-day-old video’s traffic split is like reading a stock chart after one hour of trading. Wait a week.
Search-Led Videos: The Asset Most Sellers Under-Invest In
A search-led video is the closest thing TikTok Shop has to a durable asset. It keeps producing orders after you stop promoting it, and it tells you a query phrase that real buyers use.
The Shenzhen seller’s 214-order video was search-led on a phrase his team had never deliberately targeted. Once they saw it, they made four more videos using that phrase in the spoken hook and the on-screen text. Three of the four landed in the 60–70% search band. That is the whole play: find the phrase the market already uses, then say it out loud in the video.
The limitation is real, though. Search-led volume is capped by how many people search that phrase. You cannot scale a search-led video by spending more; you scale it by finding more phrases. And phrases go stale — a seasonal product’s search demand falls off a cliff when the season ends, and the video’s orders go with it.
“We treated every video like a lottery ticket for eight months,” a Guangzhou operator running a 150-SKU home organization shop told me. “Now we sort by search share before anything else. Half the catalog videos stopped getting edited entirely. Total orders still went up, because we stopped touching the ones that were quietly doing their job.”
This is also where most seller-made video loses to creator-made video, and it is worth being honest about why. A creator saying the phrase reads as a recommendation. A brand account saying the same phrase reads as an advertisement, and buyers scroll past advertisements even in search results.
Feed-Led Videos: Reach Without Intent
A feed-led video with high views and near-zero orders is not usually a content problem. It is usually a moment problem — the point in the video where the viewer is asked to do something.
The diagnostic here is narrow: find the second where attention drops and check whether the product link appeared before or after it. If the link appeared after the drop, the viewers who would have tapped never saw it. If the link appeared before the drop, then the drop happened for a content reason and the video really does need a re-edit.
The second thing to check is whether the product in the video is the product in the link. Feed viewers arrive with no intent, so the gap between what they saw and what they would have to buy has to be close to zero. A video that showcases a whole range and links one item asks the viewer to do the work of choosing, and feed viewers do not do that work.
One honest limitation: fixing the link moment rarely converts a feed-led video into a strong seller. It typically moves a video from 6 orders to 15 or 20. If you need 200 orders, the answer is usually not more editing — it is a search-led video on a phrase buyers actually type.
Following-Led Videos: The Slowest and Most Durable
Following and profile traffic is the smallest slice for most sellers and the easiest to misread, because it looks like a failure when it is actually the beginning of something.
A video drawing 30% or more from following means people came looking for you. That is worth more per view than the other two routes, and it is the route that survives algorithm changes, because it does not depend on distribution at all.
The tradeoff is that it takes months. There is no tactic that produces a returning audience in two weeks, and any playbook claiming otherwise is selling something. The only reliable input is a posting cadence your team can actually hold — three videos a week for six months beats fifteen videos a week for one month, because the audience forms from consistency, not volume.
The DAMI store data view is where this shows up as a pattern over time rather than per video. It is one of the few places a multi-store operator can see whether the following share is rising across shops or only in one.
What to Change, by Market
The mix is not the same across markets, and this is where a single global playbook breaks.
In the US shops, search-led video tends to carry more of the volume because buyers there are further along in using TikTok as a product search surface. In Southeast Asian shops — particularly Indonesia and the Philippines — feed-led volume dominates, and the same video structure produces very different splits.
The practical implication is that you cannot copy a US video structure into a Philippines shop and expect the same conversion rate, even at the same view count. What you can copy is the diagnostic: read the split in each market, then build the structure that market rewards.
There is a limit to how far this generalises. Market-level patterns describe averages across many sellers, and any single catalog can run against the average. A category with strong branded search demand in Indonesia will look like the US pattern regardless of where the shop sits. Read your own split before you trust the market rule.
Frequently Asked Questions
What percentage of search traffic is “good”? There is no universal good. Above 50% search on a video past its first week usually means durable demand, but a video at 20% search with a 4% conversion rate is healthier than one at 60% search converting at 0.3%. Read the split together with the conversion rate, never alone.
How old does a video need to be before the traffic mix is reliable? Seven days is a workable minimum. Before that, feed distribution is still front-loaded and search has not accumulated. A 48-hour-old video’s split will usually overstate the feed share substantially.
Does the mix change if I promote the video? Yes, and it changes in the direction of whatever the promotion buys. Paid distribution typically floods the feed share and dilutes the search share, which is why a promoted video can show more views and a worse split at the same time. Read organic and paid separately.
Can a creator video be search-led too? It can, and it often is when the creator’s audience searches their handle. But creator videos skew feed-led because that is where the creator’s own distribution lives. Judge creator content on its own mix, not against your brand-account videos.
Should I delete videos with a bad mix? Generally no. A feed-led video with low orders is still reach, and reach is the input to the following share. Deleting videos resets whatever audience signal they accumulated. Leave them up and fix the next one.
Read the Split Before You Re-Edit
The Shenzhen seller’s second video was never broken. It was a feed-led video being judged on a search-led expectation, and his team spent two weeks re-editing a video that was doing exactly what a feed-led video does. The re-edit changed the order count from 6 to 9.
The fix that actually worked was boring: read the split first, find the phrase the market searches, and make videos that say that phrase out loud. If your team is reporting views as one number, that is the number to split before anything else. For the wider account-health signals that sit behind distribution, the diagnostics in this guide cover the view-to-conversion path in more detail. See how the DAMI workspace breaks video performance out by shop on the product page, or create your account to run the traffic-mix read on your last 30 days of video.
