Why a product photo no longer stops the feed
Picture a scene. It is eleven at night and someone is lying in bed, scrolling the feed with their thumb. They give each post well under a second. In that second their mind makes one decision: is this an ad, or something from a person?
A product photo on a white background, the discount in big letters on top, a red sticker in the corner. It takes less than a second to judge this picture. The mind says ad, and the thumb moves on.
The problem is not the quality of the photo. It may be very well shot. The problem is that it looks like an ad, and people do not come to Facebook to see ads. They come to see people.
So the videos that work now look the opposite: shot on a handheld phone, in room light, an ordinary person talking. In the first second and a half the mind says, this is from a person, and the thumb stops. Once it stops, you get the chance to say the rest.
What a UGC ad actually is
UGC means user generated content. In advertising it means a video that does not look like a brand-made ad. It looks like a video an ordinary customer shot on their own phone.
The shape is simple. One person, the product in their hand, the light of a home or a shop, and a few words spoken straight into the camera. No studio, no music-video cuts.
There are three reasons it works:
- The product is visible at real size. Next to a hand, you can tell how big it is, how heavy, and what colour it really is
- A person is taking responsibility. When someone shows their face and says something, it sounds more believable than the text on a poster
- It looks like the rest of the feed. Everyone in the feed is watching phone videos, so if yours looks like theirs, it does not jump out as an ad
One thing needs saying clearly. UGC does not mean fake reviews. If the person speaking is not really a customer, making them say "I used it, it is great" is cheating. Showing the product, saying what it is for and saying the price is fine.
How fifteen seconds are divided
A video that works is not random. It has a structure. Split the fifteen seconds into four parts:
| Time | What happens on screen | The job |
| First 3 seconds | Face and product together, a question or a problem | Stop the thumb |
| Next 6 seconds | One point, turning the product in the hand | Explain |
| Next 3 seconds | The price or offer, stated clearly | Get a decision |
| Last 3 seconds | One next step | Get action |
Each of the four parts has one job. If one part tries to do two jobs, both come out weak.
Look at the second part separately. It says "one point". This is the rule people break most. People want to fit everything into one video: fabric quality, stitching, colour, price, delivery, the return policy, the shop address. But a viewer does not remember more than one thing in fifteen seconds.
Which point should you pick? Ask yourself which question customers ask most when they come to the shop. That is your one point. If people keep asking "will the colour run in the wash?", let the whole video answer that one question. Keep the last step to one as well: a message in the inbox, or a word in the comments, or a tap on the link. Show three paths and people take none of them.
The first three seconds are the whole game
It does not matter how good the other twelve seconds are if nobody stops in the first three.
What the three seconds need:
- A face, from the very first frame. Do not start with a logo, do not start with a name
- The product in the same frame. The viewer should see at a glance what this is about
- The first sentence is a question or a problem. Something direct like "Do your lips crack in winter?", not "Assalamu alaikum, today we will talk about"
The most common mistake is the introduction. People stand in front of the camera and first greet, then introduce themselves, then say the shop name, and only then get to the point. By then seven seconds have gone and the viewer has too. Give the introduction at the end, when the viewer is already hooked.
One more small trick. Let the hand move in the first frame: lift the product, or open the box. Movement stops a thumb better than a still face.
The mistakes that happen most
- Too long. A one-minute video does not survive in the feed. Keep it to fifteen to twenty seconds
- Too polished. Studio light and perfect editing work against you here, because then it feels like an ad again
- No face. A video with only hands and a product works much less, because trust comes from the face
- Sounds like reading a leaflet. If you memorise a script and recite it, your voice gives it away. Speak the way you talk, and a small stumble does no harm
- The product is hardly visible. If it is held small and far away, the viewer cannot tell what you are selling
- No instruction at the end. The video ends, the viewer is impressed, and then the thumb moves on
- Bad sound. People put up with a blurry picture, but if they cannot hear the words they leave at once
The last mistake is the most neglected. Phone cameras are good enough now, but a phone microphone does not pick up well from far away. Stand about a metre from the phone, and switch the fan off.
How to take the product photo on your own phone
Whichever path you take, you need one clean product photo. A phone takes good photos, if you follow a few rules:
- Shoot in window light, not midday sun. Put the product on a table by the window and stand facing the window
- Flash off. Direct flash changes the colour of the product and makes hard shadows
- A white or plain-colour background. A sheet of white paper or a plain cloth is enough
- Take it from three angles. The front, the side, and one close up where the stitching or the printed text can be read
- Keep one photo next to a hand. It shows the size, and returns go down
- Hold the phone steady. Rest both elbows on the table. A tripod is not required
The clearer the photo, the easier the next step, whether you shoot it yourself or give it to someone else.
How long the script should be
In fifteen seconds Bangla usually fits 35 to 40 words. Write more and you either have to rush, or the video runs long.
Here is a four-line mould that works for almost any product:
- Line 1: A question or a problem your customer suffers from every day
- Line 2: Show the product in your hand, and say one thing
- Line 3: The price or offer, stated clearly
- Line 4: One next step
The easiest way to write a script is to say it out loud first and record it on your phone before you write it down. What feels awkward to say will sound awkward in the video too.
Three ways to make it
| Path | What you need | Where it gets stuck |
| Shoot it yourself | A phone, light and some time | Talking in front of the camera |
| Have a creator do it | Budget, coordination and waiting | Settling the rate and redoing it |
| Make it with AI | A product photo and a script | Bangla text on the screen |
All three are valid paths, and they cost three different things: the first costs your time, the second costs money, the third costs Token. Each is explained separately below.
What you need to shoot it yourself
The cheapest path and often the most believable, because when the shop owner says it, nobody can fake that.
What you need: a phone, window light and a quiet room. If you have no tripod, lean the phone against a box.
Three habits change the result a lot. One, assume the first take will not be good: shoot five and keep the best. Two, instead of memorising the script, write the four lines on paper and stick it next to the camera. Three, if you do not want to stand in front of the camera yourself, put the shop staff member who talks most easily there, but with their permission.
What to find out before hiring a creator
Dhaka now has a separate market for this work, and rates there vary with many things: who is doing it, how many followers they have, whether the video goes on their own page or is only given to you, and how long you will run it as an ad. With so much variation, quoting one number would only confuse, so get rates from a few people and compare.
When you ask for a rate, clear up these five points first:
- Will only you use the video, or will they post it on their own page too
- Until when can you run it as an ad
- How many versions you get, and how many times they will fix it if you do not like it
- Whether you must send the product to them, and whether it will come back
- Who writes the script, you or them
Remember the last point: being clear on the terms of use matters more than haggling over the price. If you cannot run the video after three months, even a cheap price turns out expensive.
What AI gives you today, and what it does not
The third path is new, and honesty matters here, because this is where the promises are biggest.
What you get today: give one photo of the product and you see a person holding it and speaking Bangla, in a fifteen-second video. The person can be you, using your own selfie, or one of the ready-made presenters. AI writes the script and you edit it. The voice can be of two kinds: an AI voice, or your own voice, where you speak for a few seconds into the microphone and the presenter's lips match your voice. A video takes two to four minutes to make.
What you do not get today, and this is the most important part:
- Bangla text on the screen is not drawn inside the video. If you want to put a price or an offer on it, you have to add it yourself later in an editor such as CapCut
- You cannot use another person's face without their permission, and never a photo of a well-known person. When you give your own selfie, you have to tick "the person in the photo is me"
- Some claims cannot be made. Curing disease, doubling money and political ads are blocked
- It is not right to show the product as something it is not. If the colour or size is changed, the customer will be disappointed when they get it, and returns and complaints both go up
So the practical use looks like this: AI makes the person and the speech in the video, and you add the on-screen text afterwards. At the image-making step, a picture with the product placed in the presenter's hand is shown first. Only when you say "this is right" does the video start, so there is no risk of paying for a wrong picture.
What it costs in Token
I will give the number directly, because working it out by guess goes wrong.
In BanglaCodes, a 15-second UGC ad with music costs 85,000 Token in the economy quality and 163,000 Token in the better quality. The price is shown as a number before making starts, and if no video is made, the Token come back.
Check it against the packs: the ৳499 Starter pack has 100,000 Token, which is one economy ad. The ৳999 Builder pack has 300,000 Token, which is three.
Two things to add honestly. One, scripts and ordinary images are free for everyone, but making a video is possible only on an account that has bought Token. Two, if you use your own voice, the AI voice Token are not needed. The four steps of Reel Studio are walked through in the post on making your first reel.
How to start
- One video this week, not three. One product, one point
- Write four lines on paper, following the mould above
- Take the product photo, from three angles, in window light
- First shoot once on your own phone. If you do not like it, then think about another path
- Run it on a small budget, and after a week see how many people got past three seconds
- Make another version of the one that worked, not something new
The last step is the one most often skipped. When one video works, people sit down next time to make something completely new, but small changes to a working video are the cheapest gain.
Questions and answers
Can I make a UGC ad without showing my face? Yes, and in many cases it works well. Hands, the product and a voice can make the whole video: opening the packet, using it, saying the price. You need to show a face only when trust is the main point.
How long can one video be run? As long as it works. If people see the same video again and again, the result falls, so when it starts to fall, make another version of that same video rather than something new: the same point, a new hook or a new first three seconds.
If I pay a creator, who owns the video? Paying does not automatically give you ownership. It has to be written down. Before the work starts, settle where you can run the video, for how long, and whether you may run it as a paid ad.
How do I add the Bangla text on the screen? Do not try to have AI draw that text inside the video, because Bangla conjunct letters still break. Download the video, put the text on top in any phone editor, and the spelling stays in your control too.
What budget should I start with? A small one. The first video's job is not to sell but to teach you: to see how many people get past three seconds. Once you have that number, raising the spend rests on data instead of a guess.
How will I know if the video is working? First see how many got past three seconds, then how many stayed to the end, and last how many tapped the link. If the first number is low, the hook is weak. If the last is low, the offer is not clear. How to write the hook is in the post on reel hooks.
In short
- A staged product photo no longer stops the feed, because it is clear in one second that it is an ad
- A video that works has one person holding the product and talking in ordinary light. Studio polish works against you here
- The structure has four parts: face and product in the first three seconds, then one point, then the price or offer, then one next step
- The two most common mistakes are a long introduction at the start and putting every point into one video
- There are three ways to make it: shoot it yourself, have a creator do it, or make it with AI, and the cost is time, money and Token respectively
- In AI-made videos the Bangla text on the screen is still not drawn inside, so you add it yourself afterwards
- In BanglaCodes a 15-second UGC ad costs 85,000 Token in economy and 163,000 Token in better quality, and the ৳499 pack holds 100,000 Token