Why the hidden-coupon Reel went viral: the engineering behind the pause

Field Case 01

A car drives past. In one frame, a code.

story · numbers to follow

The client was a Greek e-shop with an Instagram account and no TV budget. Whatever happened would have to happen inside the feed, among hundreds of videos fighting for two seconds of the same thumb.

Instead of one more discount announcement, we hid the discount inside the video. A car drives along a road, and in one single frame a code appears. At normal speed you barely see it. Paused at the right moment, it reads clearly. The caption said so from the start: there’s a code, go and find it.

People rewound, paused, missed and tried again. The Reel went viral, and the client’s DMs filled up with people who wanted to cash in what they’d caught.

We kept one rule from that campaign, for every mechanism we design: the DM reply is written before the video goes up, not once the messages start arriving.

Decode

Four gears, one motion.

A viral Reel looks like luck. This one was a machine with four parts, each one pushing the next.

1. A gap that begs to be closed

   The viewer knows the code is there and hasn’t seen it. That is curiosity by definition: not ignorance, but knowing that one specific thing is missing. Watching turns into a skill game with a clear target and a clear finish.

   Loewenstein, G. (1994). The Psychology of Curiosity. Psychological Bulletin, 116(1) · academic research

2. Every miss is a replay

   Miss the frame and you rewind. Nobody has to ask you to; the game does. Every rewind adds watch time, and watch time is one of the three main signals Instagram uses to decide who gets shown a Reel, alongside likes and sends.

   Adam Mosseri, via Social Media Today (January 2025) · platform statement

3. A challenge doesn’t stop at a like

   A pretty picture gets a like and a scroll. A challenge gets sent to someone: “did you get it?”. Sends are the third signal, and according to Instagram they matter most for reaching people who don’t follow you yet.

   Adam Mosseri, via Social Media Today (January 2025) · platform statement

4. The DM is the till

   The code was redeemed by message. The path from “I saw it” to “I bought it” never left the platform, and every conversation was started by the customer. That matters: on Meta’s messaging apps, automated replies are only allowed to people who wrote first.

   Meta business messaging policies · platform policy

Your audience isn’t bored. Your format is boring.

Average Instagram Reel engagement was 0.52% of followers in 2025, up from 0.50% in 2024. It’s rising, but it’s still around half a percent: roughly one interaction for every two hundred followers.

Socialinsider, Instagram benchmarks 2025 · vendor figure · “one for every two hundred”: our rounding

One more pretty post is fighting over that half a percent. What breaks through does one of three things, and the code Reel did the first two while keeping the till inside the platform.

• ReplayIt makes people watch again, because something got away.
• SendIt makes people send it, because it’s a challenge, a gift or a question.
• TimeIt pulls time onto your own site, where you do the measuring, not the platform.

Lineage

We didn’t invent it. We moved it.

The reward-in-one-frame idea is twenty years old. The job was to move it to where the audience is now, a Reel rather than a TV spot, on a Greek business budget.

The lesson we’ve kept since 2006: always announce the hunt. Then it’s a game for everyone, not a trick on a few.

The general idea

The pause was the tool. Not the idea.

Keep only “hide a code in a frame” and you’ll make a poor copy. The idea behind the Reel was more general: turn looking into doing, put the reward at one exact point, say so upfront, and have the reply ready before anyone asks.

Every medium has a control people already use without thinking. In the Reel it was the pause button. Elsewhere it’s counting, swiping, a missing word, a guess, a zoom, a live number, or a van that needs filling.

Use cases

Eight trades, eight controls.

We designed them for trades a long way from an e-shop, to show that the engineering travels. We don’t give performance estimates; we write down what we’d measure.

Car hire · Heraklion airportuse case

Count the Goats

Control

Counting

What it forces

Replay

The set-up

Fifteen seconds on the road up to the Lasithi Plateau, filmed from the passenger seat. The caption asks: “How many goats crossed the road? DM us the number. There’s a second secret too; we’ll tell you in 48 hours.”

Pause

What it hides

Forty-eight hours later, a second Reel: “You counted them. But did you see the sign?” A sign in the background had carried an upgrade code all along. Everyone counting goats missed it. A second wave of replays, on the same video.

Why it works

When attention is locked onto a count, the eye misses whatever it isn’t counting. In the famous “gorilla” experiment, about half the viewers counting passes failed to notice a person in a gorilla suit walking through the middle.

Simons, D. J. & Chabris, C. F. (1999). Gorillas in our midst. Perception, 28 · academic research

How it’s redeemed

The right number by DM gets a free second driver on the next booking; the code on the sign, an upgrade. Both work only for bookings made on the company’s own site.

The rule

Filmed by a passenger or on a closed road, never by the driver. Nobody goes near the animals. The second secret is announced in the caption from day one.

What we measure

DMs with a number · watch time on both Reels vs the account median · bookings with a code on the site · share of direct bookings.

The other half

When the DMs fill up with numbers, the AI Front Desk answers them: it checks the number, sends the code, and says it’s an AI in its first sentence. →

Back to the set-up

Furniture workshop · Larissause case

The Code in the Joint

Control

Swiping

What it forces

Replay

The set-up

A five-slide carousel that joins into one panorama: a three-metre oak table, from the log on the saw to the last coat of oil. The caption says only: “The code is in the joint.”

Pause

What it hides

The code is carved into the wood exactly on the seam between slides three and four. You can only read it if you stop mid-swipe. Whoever works that out swipes back and forth until they catch it.

Why it works

The viewer knows the code is there and can’t see it: the same curiosity gap as the Reel. And while searching, they study the craftsman’s work from a hand’s breadth away. Visible effort raises perceived value.

Loewenstein (1994) · Buell, R. W. & Norton, M. I. (2011). The Labor Illusion. Management Science, 57(9) · academic research

How it’s redeemed

The code by DM gets free delivery and assembly anywhere in Thessaly, or initials carved under the tabletop.

The rule

The code isn’t written into the alt text; the alt text says it exists and that it’s sent to anyone who asks by DM, so people who can’t see the image can still play.

What we measure

Carousel saves and sends · DMs with the code · quote requests · workshop visits.

The other half

The same panorama lives on the site too, zoomable down to the grain, where you measure the time, not the platform. →

Back to the set-up

Open-air cinema · Chaniause case

The Seat in the Frame

Control

Pause, for two

What it forces

Send

The set-up

A twenty-second trailer for Friday’s film, cut together with shots of the cinema courtyard: jasmine, screen, rows of chairs. The caption: “Two seats have your name on them. Somewhere in this video.”

Pause

What it hides

In one frame, chalked on the backs of two chairs: “Γ7 · Γ8” (row Gamma, seats 7 and 8).

Why it works

A prize for two needs a second person, so the video gets sent with the only question that matters: “Shall we?”. And every send carries the Reel to someone who may not follow you yet.

Adam Mosseri, via Social Media Today (January 2025) · platform statement

How it’s redeemed

The first DM with both seats wins them for Friday. Everyone who finds them afterwards gets two tickets for the price of one on a weeknight.

The rule

Sending is always optional, never a condition of entry. Automated DMs only answer people who wrote first. The trailer uses only footage the distributor allows for promotion.

What we measure

Sends · DMs · weeknight tickets with the code · weeknight occupancy.

The other half

The decode of the original mechanism, and the game to try it yourself. →

Back to the set-up

Greek language school for foreigners · Rethymnouse case

The Missing Word

Control

Filling the gap

What it forces

Send

The set-up

A grandmother in Rethymno opens the door to her grandson. The subtitles read: «Ίντα κάνεις, ___ μου;». The caption, in English and Greek: “Which word is missing? Answer on Sunday.”

Pause

What it hides

The word is «κοπέλι» (kopéli): Cretan for child, or boy. Locals know it by the first syllable; foreign learners don’t.

Why it works

Whoever knows the answer wants to show it in the comments, and whoever doesn’t sends the video to someone who does. Knowledge that marks you as an insider works as social currency.

Berger, J. (2013). Contagious: Why Things Catch On. Simon & Schuster · book by an academic

How it’s redeemed

DM the word and get a free trial lesson, “Cretan for beginners”, plus the recording of the phrase with an explanation.

The rule

Written consent from grandmother and grandson. The dialect is shown with affection, never as a joke at anyone’s expense. Subtitles in two languages.

What we measure

Comments and sends · DMs with the word · trial lessons · enrolments for the next course.

The other half

“Find the Mistake”: another education scenario, with a deliberate mistake where the missing word would be. →

Back to the set-up

Farm supplies · Almyros plainuse case

Guess the Weight

Control

Estimating

What it forces

Return visits

The set-up

A bale of alfalfa on the shop’s platform scale, its reading hidden behind a piece of cardboard. “How many kilos? Put your guess in the comments. We reveal the reading on Thursday.”

Pause

What it hides

Thursday’s Reel shows two numbers side by side: the real weight and the median of every guess. If history repeats itself they’ll be very close, and that becomes the story.

Why it works

In 1906, at a livestock fair in Plymouth, Francis Galton collected 787 guesses at the weight of an ox. The median was less than 1% off the true weight. The “wisdom of the crowd” makes the reveal worth a second visit.

Galton, F. (1907). Vox Populi. Nature, 75 · academic paper

How it’s redeemed

The closest guess wins a free visit from the shop’s agronomist to their field. Anyone who commented can ask by DM for the season’s sowing guide.

The rule

A contest of skill with written terms; ties go to the earliest comment. No “tag a friend to enter”.

What we measure

Comments with a guess · return views on the reveal Reel · DMs · agronomist visits booked.

The other half

The comments are data. The shop’s sales are better data still: they give you a demand forecast for next season. →

Back to the set-up

Plumbing · Heraklionuse case

Find the Leak

Control

Zooming

What it forces

Time on your own site

The set-up

On the company’s site, a very high-resolution photo of a bathroom you can zoom into down to the grout. A short Reel sends people there: “Water is running somewhere it shouldn’t in this bathroom. Find where.”

Pause

What it hides

Three signs of a hidden leak: one grout line darker than the rest, a skirting board that has swollen, and a water meter turning with every tap shut.

Why it works

Visitors learn the three signs by hunting for them, and what we find for ourselves we remember better than what we read. Next time they spot a dark grout line in their own bathroom, they’ll remember who to call.

Slamecka, N. J. & Graf, P. (1978). The generation effect. Journal of Experimental Psychology: Human Learning and Memory, 4(6) · academic research

How it’s redeemed

Find all three and get a free leak check with a thermal camera, plus a fridge magnet with the three signs.

The rule

The photo is your own or licensed, never from a client’s home without written permission. The three signs are also described in text, for anyone who can’t see the image.

What we measure

Time on page · completions · inspection requests · calls from the page.

The other half

The same engine runs “Human vs Vision AI”: there, the model hunts for defects in a production-line photo, and shows you where it gets things wrong too. →

Back to the set-up

Olive mill · Messara plain, Creteuse case

The Mill Queue

Control

Checking before you set off

What it forces

Return visits

The set-up

At harvest, growers queue outside the mill on their tractors with no idea for how long. A camera at the gate counts the vehicles waiting, and a Viber channel and the website show it live: “Queue: 4 tractors · about 30 min”.

Pause

What it hides

There’s no code hidden here. There’s an hour: the one nobody loses in the queue. Every afternoon, a ten-second Reel of the day’s fresh oil keeps the channel alive beyond the queue.

Why it works

An uncertain wait feels longer than a known one. A live number turns the channel into a tool people open every morning, not an ad they scroll past.

Maister, D. H. (1985). The Psychology of Waiting Lines. In The Service Encounter · academic chapter

How it’s redeemed

No code. Growers who join the channel for the queue are also the first to hear about prices, milling slots and when their oil’s lab results are ready.

The rule

The camera counts vehicles; it doesn’t recognise faces or number plates, and the image isn’t stored. A notice at the gate, as GDPR requires whenever a camera can see people.

What we measure

Channel members · daily views of the queue status · average wait at the gate, before and after · batches milled per day.

The other half

This is machine vision at its simplest: it counts, it doesn’t identify. On a production line, the same technology spots defects. →

Back to the set-up

Removals · Volosuse case

Will It Fit?

Control

Fitting

What it forces

Time on your own site

The set-up

A small game on the site: drag a sofa, a fridge, three bookcases and twenty boxes into a cutaway van. A Reel built on the oldest question of every move, “will the sofa fit?”, sends people there.

Pause

What it hides

The game tells you whether everything fits in one trip. And the volume you’ve just worked out by playing becomes your quote request, already filled in.

Why it works

We value what we build with our own hands more. Here the visitor “builds” their own load, and the form stops feeling like a form.

Norton, M. I., Mochon, D. & Ariely, D. (2012). The IKEA effect. Journal of Consumer Psychology, 22(3) · academic research

How it’s redeemed

No code. There’s a quote that arrives without a phone inventory, and a roll of “Fragile” tape with the first booking.

The rule

The game’s estimate is marked as non-binding. The final price comes from a person, in writing.

What we measure

Game completions · quote requests · time from request to quote · acceptance rate.

The other half

The volume from the game goes straight into a draft quote, which the manager checks and sends. That’s automation, not marketing. →

Back to the set-up

Checklist

Before you post: seven checks.

1. The hunt is announced, in the video or the caption. Announced, it’s a contest; hidden, it would be a subliminal technique.AVMSD, Article 9
2. Every prize has written terms: who can enter, what they win, how many codes exist, how long they last.Greek rules on promotional games · checked by a lawyer
3. If the code is a discount, the lowest price of the previous 30 days is shown next to it.Directive (EU) 2019/2161
4. No “share or tag to enter”. Sending is always optional.Meta Pages policies
5. The code frame never strobes: never more than three flashes a second.WCAG 2.3.1
6. The DM reply is written before the video goes up, and you know who answers at 11pm.Netex rule
7. There’s a route for anyone who can’t play against the clock or can’t see the image: frame by frame, text, or “ask us by DM”.WCAG 2.2.1 · Netex rule

Measurement

We don’t promise viral. We promise measurement.

Nobody serious can promise a video will go viral. What they can promise is a mechanism that gets measured, and tell you upfront what will be measured and against what.

The comparison is always against your own account: the median of your own Reels over recent months, not an industry average that looks nothing like you.

1. Views and accounts reached
2. Average watch time vs the account median
3. Replays, where the platform reports them
4. Sends and saves
5. DMs started by the mechanism
6. Codes redeemed
7. Revenue or bookings from the codes
8. New followers over the period

Source: the account’s Insights and the sales or booking system. Period and baseline are written down before the video goes up.