TL;DR
- Video used to be advertised around RM1,000 a piece. Since AI, offers as low as RM20 a video show up in the feed — usually inside a large bundle, so the order total is still in the thousands.
- AI did not invent this. The same bundles existed before: 10 videos for RM8,000–10,000, or 15 for about the same. The structure is identical. Only the decimal moved.
- The extremes are readable; the middle is not. At RM20 you know what you are getting. Between RM100 and RM500 there is often no way to tell which is better — that band is where the real problem lives.
- Buying cheap on purpose is legitimate. Three rules: know that is what you are doing, go cheap but not floor-price — ask whether they can still make money — and use it for direct-response ads, not branding.
- One caveat we would apply to ourselves: testing ads with material you already know is weak gives you weak test data, and ad budget is expensive enough already.
- There is more than one market. One runs on constant advertising and one-off jobs; one runs on referral and reputation — we have worked in the second since 2018 without buying advertising. Both are real, and as a buyer you are choosing which one you are shopping in.
Scroll far enough through the feed and the numbers stop making sense. Video production advertised at a thousand ringgit a piece. Then five hundred. Then, more recently, twenty ringgit a video — inside a big enough bundle that the order is still worth a few thousand to whoever is selling it.
It is tempting to read that as AI having broken the market. That is not quite what happened, and the more useful version is less dramatic.
Why can someone sell a video for RM20?
Only inside volume.
A twenty-ringgit unit price does not exist on its own. It exists at thirty or fifty videos at a time, where the total order is still worth thousands, and where each individual piece therefore carries about twenty ringgit of attention — iterations, art direction, someone deciding a version is not good enough and starting again.
That figure is not an insult. It is a specification. It tells you precisely what the work can be, and it cannot be more than that no matter who is selling it or what they intend.
Did AI make video cheaper, or just cheaper-looking?
This is the part worth slowing down for, because the shape of these offers is not new.
Before AI, the same bundles were already being sold:
| Era | Typical bundle | Roughly per video |
|---|---|---|
| Before AI | 10 videos for RM8,000–10,000 · 15 videos for about the same | RM500 – RM1,000 |
| With AI | 30–50 videos in one package | as low as RM20 |
The packaging logic is identical. Sell volume, quote a headline number, let the buyer do the division later — if at all. What AI changed was not the model. It changed how far the unit price could fall before the model stopped working.
So this is not a new problem arriving with a new technology. It is an old problem with a new decimal point.
Why is it hard to tell a good video studio from a cheap one?
Here is the observation that matters most, and it runs against instinct: the extremes are the easy part.
At twenty ringgit a video, nobody is confused. You know what that is. At the top of the market, the work usually speaks for itself.
The trouble is the middle. Between roughly RM100 and RM500 a video, there is frequently no way for a buyer to tell which is better — and quite often, there is no difference to find. Same models, same platforms, same largely automatic process. The gap between those two numbers can be entirely premium, buying nothing.
That is the band where most buying decisions actually happen, and it is the band where price carries the least information. We wrote about the mechanism behind that — economists call it a market for lemons — in what AI video costs.
When is buying the cheap option the right call?
Often. This is worth saying plainly, because there is a real use case and pretending otherwise would be dishonest.
If you run performance advertising at any scale — a hundred or more pieces of creative a month across social — then one of your production lines can reasonably be handed to a volume operation. That content is not carrying your brand story. It is direct-response: a call to action, a UGC-style piece, one of many variants competing for a click. Craft is not what the job requires, so paying for craft is waste.
How should I buy it if I am going cheap on purpose?
Three rules.
- Know that is what you are doing. Almost everything that goes wrong with a cheap bundle starts with a buyer who half-expected it to be better than it can be. Decide in advance, in writing if it helps, that you are buying volume rather than craft.
- Go cheap, but not to the floor. Put yourself in the seller's position and ask whether they can still make money at that number. If the answer is clearly no, the price is being paid somewhere — usually in the parts you cannot inspect until after delivery. The best value tends to sit just above the bottom, not at it.
- Point it at the right job. Direct response, CTA creative, UGC-style variants. Not the brand film. Not the piece that has to make someone trust you.
Should I use cheap AI video to test my ads?
This is the one place we would push back on our own advice above, and it is a caveat rather than a rule.
Testing is the most common justification for buying cheap creative — it is only a test, so it does not need to be good. The difficulty is that an ad test measures the creative as much as the offer. If you already know the material is weak, a poor result does not tell you whether the idea failed or the execution did. You have spent real media budget to produce ambiguous data.
Media spend is expensive. Cheap creative on top of it does not make the test cheaper; it makes the conclusion less usable.
How do I judge a studio if I cannot judge the work in advance?
Ask for something you can actually verify. Two things work.
For AI work, ask for a short test piece — around fifteen seconds — before you commit anything. This is how we run it: we produce a brief trial, and only sign once you have seen it. It turns an unverifiable claim into a sample sitting in front of you, and it costs you nothing to ask. Any studio confident in its process can do this.
For conventional production, a test piece is not possible — you cannot shoot fifteen seconds of a film speculatively. What replaces it is the conversation, so assess that instead:
- Do they try to understand what the video is actually for before quoting?
- Does the discussion move toward your goal, or toward a bigger number?
- Will they tell you when they are not the right team?
That last one is the most informative, because it is the only question whose honest answer runs against the seller's own interest.
We refer people elsewhere regularly. If what we do would be a poor use of someone's money — or if it is clear they are shopping in the volume market — we do not try to talk them round. We point them toward someone cheaper and leave it there.
The reason is self-interested as much as principled. The worst version of this business is taking a job at a number that is already our floor, from a client who still thinks it is expensive. Nobody wins that one: we cannot go lower, and they never stop feeling overcharged. It is better to say no at the start than to spend three months proving it.
There is more than one market
The volume market is real and it is not going away. But it is worth knowing that it is not the only market, and that it never was.
One market runs on constant advertising and largely one-off jobs — which is precisely why the advertising has to be constant. The other runs on referral and reputation, and moves more slowly because it has to.
We have worked in the second one since 2018 — eight years, without spending a cent on advertising. Work arrives through referral and through people finding what we publish. That is a choice with real costs: it does not scale quickly, and it means turning down work that does not fit. Growth for us has meant getting more value out of the capacity we already have, rather than adding headcount — we are not trying to become a larger company, and we do not measure ourselves by the size of the team.
Neither market is the honest one and neither is the trick. They serve different needs, and most confusion about video pricing comes from shopping in one while expecting the other's outcome.
The number that started this — twenty ringgit a video — is not evidence that the market broke. It is evidence that a very old packaging model finally met a technology that could take its unit cost close to zero. The offers were already there. The question worth asking has not changed either: what does this piece have to do, and does the price attached to it make that possible?
Thinking about a project like this?
Tell us what it needs to do — we'll take it from there.
Frequently asked questions
- How can someone sell a video for RM20?
- Only inside volume. A RM20 unit price appears in large bundles — thirty or fifty videos at once — so the order is still worth thousands to the seller, and each individual piece carries about twenty ringgit of attention. That is the whole specification. It is a real product, but it is not the same product as a video priced on its own.
- Did AI make video production cheaper?
- It moved the number, not the structure. Before AI, bundle offers ran at roughly 10 videos for RM8,000–10,000, or 15 for about the same, with a per-video figure around RM500–1,000. Now the same bundle shape holds 30 to 50 videos. The packaging logic is identical; AI mostly changed how far the unit price could fall.
- Is cheap AI video worth it?
- It can be, if you know that is what you are buying and you are using it where craft is not the job — direct-response ads, call-to-action creative, UGC-style content. It is the wrong buy when the piece has to carry your brand, and it is a poor choice for testing ads, because weak creative produces weak test data on top of ad spend you have already committed.
- How do I choose between two cheap video quotes?
- Avoid the absolute floor. Put yourself in the seller's position and ask whether they can still make money at that number. A price nobody can profit from is paid for somewhere — usually in the parts you cannot see before delivery. Go cheap deliberately, but not to the very bottom.
- How can I judge a video studio if I cannot judge the work in advance?
- Ask for something you can actually check. For AI work, ask for a short test piece — around 15 seconds — before you commit; it converts a claim into a sample. For conventional production a test piece is not possible, so the thing to assess is the conversation: whether they try to understand what the video is for before they quote, and whether they will tell you when they are not the right team.
Keep reading
- What AI Video Costs in Malaysia: Per-Second Rates, Packages, and What We ChargeAI video is quoted per second, per package and per project, and the same words hide wildly different products. Here is how the pricing actually works, the ratio we quote against real production, the point where AI stops being cheaper at all, and the kind of brief where we are honestly not your best option.
- Usage Rights, Talent and Music Licensing: The Part Everyone ForgetsThe line brands overlook until it's a problem. Here's how usage rights, talent fees and music licensing actually work in video production — where the cost really sits, what a studio does and doesn't charge for, and the questions to ask before you sign.
- "We Already Have Footage — Just Recut It": Whether That Works Is Decided by the Footage, Not the BudgetWhen the brief becomes "don't shoot this year, recut the old material," you're being asked to bet on the physical state of a batch of files. Before you say yes, spend fifteen minutes on what you actually have — because a finished film isn't raw footage, 1080p doesn't become 4K, and some footage has quietly gone out of date.