The first second is food, not your logo
Open on the food, filled to the edges of the frame. The cheese pull, the sauce going over, the knife cutting through, steam off the grill, the first bite with the hand still in shot. That is second one — no logo, no intro card, no drone shot of the building. Snapchat and TikTok give you about one second before a thumb decides. We test this every week in Kuwait and the pattern does not move: an appetite frame at 0:00 holds two to three times more viewers to the three-second mark than a branded opener does.
Shoot close and light it properly. Most restaurant footage in Kuwait dies under the ceiling lights — orange, flat, and the food comes out grey. One cheap LED panel at 45 degrees off the side of the plate, a window at noon, or the phone torch bounced off white paper is enough. Get the lens 20 to 30 cm from the dish and let it fill the frame. Sound matters more than owners expect: the crunch, the sizzle, the pour recorded on the phone mic beats any music track, because most Snapchat viewing here happens with sound on.
Keep it 6 to 12 seconds. Longer than that and you are paying for views that were never going to walk in. Structure it: appetite frame first, price and dish name on screen by second three, branch and 'order now' by second eight. Vertical 9:16, filmed with the phone held straight — cropped landscape footage with black bars is the fastest way to look like an ad and get skipped. Do not put trending audio over a sizzle. Do not add a voiceover explaining the burger. The food is the argument.
Ready to start your Shopify store?
Start a free trial and try the platform for yourself.
Phone footage of the real dish beats the studio shoot
The phone video of the dish that actually leaves your kitchen out-performs the 400 KD studio shoot in most tests, and we have stopped arguing about it. Two reasons. First, the styled plate is not what arrives at the table, and people in Kuwait have been burned enough times to read polish as a warning. Second, a stylised shoot looks like advertising, and the feed is trained to skip advertising. Rough footage — a hand, the real tray, the real table in Salmiya, the light in the branch — reads as someone showing a friend. That register converts.
Put the price on the screen. Not in the caption, not on the landing page: burned into the video, in KD, inside the first three seconds. A visible 1.750 KD filters out everyone who was never going to buy and pulls cost per order down, usually by a fifth to a third once we add it. Same with the branch. 'Salmiya, behind Al-Fanar' or 'Khaitan, next to the co-op' turns a scroll into a decision, because the biggest reason a hungry person does not act is not knowing whether you are 5 minutes away or 35.
If the ad works, do not send that traffic straight to Talabat. You just paid to acquire a customer and then handed 25 to 30 percent of the ticket to a marketplace that keeps the phone number. Send them to an order page you own instead — a simple Shopify store with KNET at checkout takes an afternoon to set up and covers its own cost at around 40 orders a month. Keep the delivery apps for discovery, but every fils of ad spend should land somewhere you control.
Run five at once and expect to replace them every two weeks
Run five to eight creatives at a time, not one. Same offer, different dish, different first frame, different hook. The platform will find the winner far faster than you will guess it, and the gap between your best and worst creative is usually 3 to 4x on cost per order — bigger than any targeting change you will ever make. Give each one at least 5 KD a day and three full days before you judge it. Below that you are reading noise, not results, and you will kill the one that was about to work.
Food creative dies fast here because the radius is small. A restaurant in Kuwait realistically advertises to 40,000 to 150,000 people inside a sensible driving distance, so at 100 to 200 KD a month the same person sees your video six, eight, ten times within two weeks. Watch frequency: once it crosses about 3.5, cost per order starts climbing, and by 5 it is roughly double. That is not a targeting problem and no agency can optimise around it. The creative is simply used up, and the only fix is new footage.
So build a filming habit, not a campaign. One hour a week, the phone, whatever is coming out of the kitchen that day — 8 to 10 clips banked. Refresh the ad set every 10 to 14 days: keep the two winners running and swap the rest. Ramadan, Eid and Hala February need their own footage shot in advance, because the fatigue clock runs faster when everyone is bidding. And film in July even though the dining room is empty; you will need that bank in September when everybody comes home.