Every ad you have ever run was one of two things.
It was a claim, which asks the person on the other side to believe you. Or it was a proof, which does not.
That distinction sounds academic until you notice which one you can afford. A claim runs on borrowed trust. When Nike says just do it, forty years of evidence is standing behind the sentence. When your business says premium quality, unbeatable service, nothing is standing behind it except your willingness to have typed it. The words are identical in cost and opposite in value.
Most small business advertising is claims, all the way down. And most of it fails for a reason that has nothing to do with targeting, budget or the algorithm.
The Test That Tells You Which One You Made
Chris Rudy, who runs the ecommerce agency Adacted, has the cleanest version of this test I have come across. Look at your ad and ask:
Could a competitor copy your headline word for word and run it tomorrow?
If yes, you wrote a claim. "Trusted local experts." "Quality you can count on." "The best in the business." Your competitor can paste any of those into their own ad this afternoon and lose nothing, because the words were never attached to anything.
Now consider an ad that consists of your product being dropped off a roof and not breaking. Your competitor cannot copy that. Not because of copyright, but because copying it requires their product to survive the fall.
That is the whole argument for this format, and it is a strategic argument rather than a creative one. A demonstration is the only ad you can run that your competition cannot simply take.
This Is the Oldest Idea in Advertising, and It Is Not "Show, Don't Tell"
Worth clearing something up, because the slogan attached to this idea has drifted a long way from what the people who discovered it actually said.
"Show, don't tell" is routinely attributed to Claude Hopkins, Rosser Reeves or David Ogilvy. It does not appear in their major works, and it is not a line you can source to any of them. It is later creative folklore, condensed and simplified until it lost the useful part.
What Hopkins actually wrote in Scientific Advertising in 1923 is more specific, and it is better advice:
Read the last sentence carefully, because it inverts the slogan. Hopkins is not saying show instead of telling. He is saying tell first, then prove it. The demonstration is not a replacement for the claim. It is the thing that closes the claim.
That distinction matters enormously in practice, and it is where most demonstration ads go wrong. A video of someone using your product with no promise attached is not a demonstration. It is footage. The demonstration only does work when the viewer already knows what you are asserting, so that watching it land creates the small, specific click of oh, that is actually true.
Rosser Reeves built the same idea into a formal theory in Reality in Advertising in 1961, where the unique selling proposition had to be a claim a competitor could not or did not make. Reeves's television work leaned hard on dramatized proof for precisely the reason above: the claim was the point, and the demonstration was what made the claim survive contact with a skeptical viewer.
A hundred years of direct response, and the structure has not moved. Say the thing. Then show the thing being true.
What a Demonstration Cost in 2006
The most famous demonstration campaign of the internet era is worth looking at properly, because the details are more instructive than the legend.
In October 2006, Blendtec's newly hired marketing director, George Wright, noticed that the company's founder, Tom Dickson, had a habit of stress-testing blenders in the research lab by putting wooden boards through them. Wright spent about fifty dollars on props: lab coats, marbles, a garden rake, a bottle of Coca-Cola, a rotisserie chicken and a McDonald's Extra Value Meal. Then he filmed Dickson blending them.
That is the entire production. A man in a lab coat, a blender, and a shopping trip that cost less than dinner.
Be careful with what followed, because this is where the story turns into folklore. You will constantly see a "700% sales increase" attached to Will It Blend?. That figure cannot be traced to any primary financial source. What can be sourced is a 2007 trade-press report quoting Blendtec's own marketing director describing a fivefold sales increase. Still remarkable. Still not the number everyone repeats.
The reason it worked is the reason this whole format works. Blendtec's claim was that their blender was unusually powerful. Every blender company claims that. What Blendtec did was hand you the evidence in a form where believing them was no longer necessary or even relevant. You watched a rake become dust.
What Actually Counts as a Demonstration
The category is much wider than people assume. Most owners hear "demonstration" and picture an unboxing, which as the creative strategist Dara Denney points out is usually the least interesting version. The thing that works is not the box being opened, it is the product in action.
Here is the fuller list, and most businesses have at least three of these available today.
The product doing its job. The most literal version. The thing working, filmed plainly, long enough for the viewer to see it work.
The before and after. The oldest proof format there is, and still one of the strongest when the "before" is something people recognize as their own problem.
The stress test. Blendtec's category. You do something to the product that the viewer expects will break it, and it does not.
The timing. Real-time or clearly compressed. How long the job actually takes, shown rather than asserted. Enormously effective when your competitors are vague about it.
The side by side. Yours next to the alternative, doing the same task. Legally sensitive if you name a competitor, entirely safe if you compare against the generic version of the thing.
The process. What happens between paying and receiving. This is the most underused of all, because businesses assume it is boring. It is not boring to someone deciding whether to trust you with money.
The objection, answered on camera. Take the question you get asked every single week and answer it by showing, not by writing a paragraph on the FAQ page.
Notice that only two of those require a product you can hold.
Your Copy Can Demonstrate Too
The claim-versus-proof split is not only a video problem. It runs through your headlines, and the fix there is the same one.
Compare these two:
Transform your morning routine.
Ready in five minutes.
Both are seven words or fewer. One is a claim you are invited to accept, the other is a fact you can check tomorrow. And the second is not merely more persuasive, it is more memorable, for a reason that is well documented in memory research: concrete, imageable language is recalled far more reliably than abstract language. Read a list containing seamless, transition, musty, leg of lamb and Irishman, and an hour later you will still have the leg of lamb.
Rudy frames the practical rule as writing things a person can picture. Premium skin care solution is a claim wearing a suit. Clear skin in 28 days is a testable assertion with a number in it, and the number is doing the same job the camera does in a demonstration: it takes the sentence out of the realm of things you are asked to accept on faith.
This is also the fastest audit you can run on your existing ads. Go through your headlines and mark every word a customer could not photograph, count or check. If what is left is nothing, that ad is asking to be believed by people who have no reason to.
Services Can Demonstrate, and Almost Nobody Does
This is the biggest open opportunity in the format, and the reason is a failure of imagination rather than a limitation of the medium.
A service business hears "demonstration" and concludes it does not apply, because there is no object to put in front of a camera. But every service produces evidence continuously. It just gets thrown away.
A plumber can show the camera inside the wall before and after. An accountant can show a shoebox of receipts becoming an organized return, with the numbers blurred. A detailer has the most demonstrable business on earth and usually posts a photo of the finished car with no before. A landscaper has a drone and a season. A physiotherapist can show a range-of-motion test at week one and week six with the same patient.
We made the adjacent version of this argument for home service businesses in Home Service Ads Do Not Look Like Ecommerce Ads. The demonstration is the format that answers "will this actually work" rather than "are you a real business," and most service advertisers are only ever answering the second one.
The test for a service is the same as for a product. What do people not believe until they see it? Film that.
What Meta's Own Tests Actually Show
Meta publishes less usable creative research than you would expect, and a great deal of what circulates as "Meta says" cannot be found anywhere Meta actually said it. Two figures on its own Reels ads page are properly sourced, and both are relevant here.
The first: Reels ads built as 9:16 video with audio, with key creative elements inside the safe zone, delivered 34.5% lower cost per result than image ads on Reels. Meta's own source note describes a meta-analysis of fifteen split tests, in ecommerce, retail and consumer packaged goods, including small and medium businesses, with the result holding at 99.9% confidence.
The second: adding native Reels creative built the same way to an otherwise business-as-usual campaign drove on average 5% lower cost per result and an 11% higher conversion rate, from a series of fifteen A/B tests across five countries between May 2022 and April 2023.
Neither of those says "demonstrations win." They say something narrower and still useful: creative built for the placement beats creative resized into it, by a margin large enough to survive fifteen tests. A demonstration filmed vertically on a phone is, by accident, exactly the thing those tests reward. A demonstration cropped out of a landscape brand video is not. We wrote about that specific mistake in Your Story Ad Is Not Your Feed Ad Resized
One correction while we are here, because it cuts against advice you have probably been given. Meta's own documentation states that Reels default to sound on. The habit of designing every video for silent viewing comes from feed behavior and has been generalized into a law it was never entitled to be. Caption your video anyway, because captions help comprehension and accessibility regardless. But do not assume nobody can hear the thing you built an audio demonstration around.
The Cheapest Ad You Will Ever Make
Every other creative format has a floor under its cost. A brand film needs a script, a shoot and an edit. A creator ad needs a creator. A polished static needs a designer, or at least someone with taste and time.
A demonstration needs a phone and the thing you already do every day.
You are not performing. You are not writing. You are recording work that was going to happen anyway, and the version filmed on a phone in a driveway routinely beats the version with a crew, because polish reads as advertising and advertising is the thing people are scrolling to avoid.
This is also why it scales. You are not commissioning creative, you are capturing it, and businesses that get into the habit end up with a library rather than a campaign.
One Shoot Becomes Several Ads
The thing people miss about this format is that it is not one ad. A single recording of the work being done is raw material for a whole set, and this is where the economics get genuinely favorable.
Take the plumber who films one water heater installation properly. That footage is at least four ads.
The proof cut. The install, compressed, showing it done cleanly and completely. This answers "can they actually do this."
The timing cut. The same footage with the clock on it. This answers "how long will my house be a building site."
The objection cut. The thirty seconds where something awkward happens and gets handled. This answers the question your phone gets every week.
The finish cut. The last two minutes, the cleanup, the site left tidy. This answers a question nobody asks aloud and everybody worries about.
Four different arguments, four different ads, one afternoon of recording that was already scheduled. Then each of those becomes several variations as you change the opening line and the on-screen text.
That is what separates a business that runs demonstrations from a business that once posted a video. And it is the practical answer to the volume problem every Meta advertiser eventually hits, which is that the platform will chew through creative faster than you can commission it. Capturing is sustainable at a cadence that commissioning is not. We took the general version of that supply problem apart in Everyone Says Make More Ads. Nobody Says Who Makes Them.
How Demonstrations Fail
Four ways, and they account for most of the disappointing ones.
Demonstrating something nobody doubted. If no reasonable person questions whether your coffee is hot, filming steam proves nothing. The format only pays when it resolves genuine skepticism, which means you have to know what your buyers actually disbelieve. Ask the person who answers your phone. They know.
Demonstrating three things at once. A demonstration is a single argument. The moment it becomes a tour of features, it stops being proof and becomes a brochure with motion.
Burying it. Five seconds of logo, a title card, a slow establishing shot, and the thing you are proving arrives at second eleven to an audience that left at second three. Start on the proof. Everything else is optional. Nobody Reads Your Meta Ad. The Right First Line Makes Them. makes the same case for the copy.
Faking it. The one unrecoverable error. A staged demonstration that a viewer clocks as staged does more damage than no ad at all, because you have now demonstrated something true about yourself instead. The whole value of this format is that it is checkable. Cheating removes the only thing you were buying.
Finding Yours
One question, and it is worth asking out loud with the people who talk to your customers:
What do people refuse to believe until they see it?
Not what you want to say. Not what your website leads with. The specific thing prospects are quietly skeptical about, that would be settled instantly if they could just watch it happen once.
That is your demonstration. There is usually only one that matters, you almost certainly already know what it is, and it has been sitting in your working day this whole time, unrecorded.
Hopkins had it in 1923 and the mechanics have not changed since. Make the promise. Then show the promise being kept, plainly enough that believing you stops being something the customer has to decide to do.
If you would rather someone else worked out which proof to build a campaign around and then built it, that is a fair chunk of what our Facebook and Instagram ads work involves. But this is genuinely the one format where the disadvantage of being small disappears. The camera does not care how big you are. It only cares whether the thing works.
