Long form video ads: when a 3 to 20 minute ad beats a 15-second one, and how to build and measure it

Long form video ads run from a few minutes to about twenty and sell inside the ad, like a VSL. They earn their length when a product needs a demo or a mechanism explained to cold viewers and the platform lets uninterested viewers skip. Build the first five seconds around the skip button, follow the VSL arc, and judge the ad by cost per sale, not cost per view.

When do long form video ads beat short ads?

A 15-second ad can show a product and ask for a click. It can't explain why a problem keeps coming back, demonstrate the fix, answer three objections and present a bundle offer. A long form video ad can, and on a skippable placement the viewers who sit through it have selected themselves: most people who weren't interested left at five seconds.

That self-selection also changes what you pay for. On YouTube, with cost-per-view bidding, a skippable in-stream ad is charged when a viewer watches 30 seconds (or the whole ad, if it's shorter) or interacts with it, whichever comes first (Google Ads video formats). Viewers who skip early don't cost a view.

Use a long ad when most of these are true:

• The product needs a demonstration or a mechanism explained before the price makes sense.
• The traffic is cold: viewers know the problem but not your solution.
• The order value, with bundles, can carry a higher cost per sale.
• You have a story, a demo or a presenter that can hold attention past the first minute.

Short ads win for simple products, impulse prices, audiences that already know you, and testing angles quickly. You can run both: the long ad for cold viewers and short cuts for retargeting the people who watched part of it. The two side by side:

Short ad vs long form ad

6–60 secondsShort ad

• Job: stop the scroll and earn a click
• The selling happens on the landing page
• Fits simple products, impulse prices and retargeting
• Judge by cost per sale; click-through rate is a diagnostic
• Risk: cheap clicks from viewers who are still cold

A few minutes to about 20Long form ad

• Job: make most of the sale inside the ad
• The landing page only has to close
• Fits products that need a demo or a mechanism explained
• Judge by cost per sale; view rate and retention are diagnostics
• Risk: a weak opening wastes the whole edit

Four long form video ad formats

The VSL as the ad is the cheapest to test, because the asset exists. Recut only the first 30 seconds: an opening that hooks a visitor who clicked through to watch isn't necessarily one that holds a viewer reaching for the skip button.

The mini-documentary works when the market has heard every claim and needs to see the problem investigated. It has to stay recognizable as an ad: the FTC's native advertising guidance says an ad shouldn't suggest or imply that it's anything other than an ad (FTC: Native Advertising guide). No fake news anchors, no invented experts. The advertorial guide draws the same line for articles.

The founder story sells trust, so keep it true and specific: the problem the founder had, what they tried, why they built this. Leave out promises the founder can't back up, especially about health or income.

The demo walkthrough is the long version of a product demo: more uses, harder tests, the questions buyers ask in the comments. The product demo video post covers how to keep it honest.

Should the ad sell, or send viewers to a landing page?

There are two models, and they need different pages.

Sell in the ad. The ad is the whole VSL, offer included. A viewer who clicks has already heard the pitch, so don't make them sit through it again: send them to a page where the offer is visible at once, or straight to the order page. This model asks the most of the ad and the least of the page.

Warm up in the ad, sell on the page. The ad is a shorter story that stops before the offer, and the click goes to a VSL page that continues the argument and makes the offer. This suits offers that already have a page that converts; the mini-VSL guide covers the shorter structure.

Test the two models on the same audience and judge by cost per sale. Whichever you choose, carry the campaign data to the page, or you won't know which ad earned the sale. If the landing page is a RevvyScales VSL page, add the rvs parameter to the Final URL suffix in Google Ads, and each sale is attributed to its campaign, ad group and ad (ad tracking with rvs).

How to structure a long form video ad around the skip button

The skip button appears after 5 seconds, and that fact sets the structure. The first five seconds aren't an introduction; they're the reason to stay. Then the ad has until 30 seconds, the point where a skippable view counts, to make a promise worth ten more minutes.

What goes into the first five seconds:

• Name the viewer or the problem, in words and on screen, in the language your buyers use.
• Show the most surprising true thing you have: the result of the demo, the unexpected cause, the before.
• No logo and no "Hi, I'm...": a brand intro spends the only seconds everyone sees.

After the hook, follow the VSL arc with two changes for an ad. Mention the call to action earlier and more than once, since a viewer can click at any point. And keep every claim one you can back up: Google's Misrepresentation policy doesn't allow claims that present an improbable result as the likely outcome (Google Ads Misrepresentation policy). The video ad script post has beat-by-beat templates, and the guide to how long a VSL should be covers sizing the arc to your price and audience.

Here is the arc with rough timings for a ten-minute ad:

The arc of a ten-minute ad

Rough timings; your retention data sets the real ones

1. 1Hook (0:00–0:05)The problem or the most surprising true result, before the skip button
2. 2Promise (0:05–0:30)What the viewer will see if they keep watching
3. 3Problem and mechanism (0:30–3:00)Why the usual fixes fail, and what works instead
4. 4Demo and proof (3:00–6:00)The product working and real customers, with the first call to action
5. 5Offer (6:00–8:30)Bundles, price, guarantee and the call to action
6. 6Close (8:30–10:00)A recap for the viewers still watching, and a last call to action

Which YouTube campaign types run long form video ads?

Google's video ad formats, as its help page describes them:

Skippable in-stream ads run in several campaign types, including Demand Gen, Video view and Video reach campaigns and Performance Max (Google Ads video formats). For a direct-response advertiser, the change that matters is that Video Action Campaigns no longer exist. Google removed the option to create new ones in April 2025, began upgrading existing ones to Demand Gen in July 2025, and set April 2026 for the last upgrades (Google Ads Help). Demand Gen supports conversion bidding, including Maximize conversions, Target CPA and Target ROAS, alongside click-based strategies, and Google says it can reach up to 3 billion monthly active users (Demand Gen FAQ).

Google's own session on Demand Gen covers inventory, product feeds, targeting and bidding. It's a recorded livestream and a long one, so use the chapters:

In Meta's feeds the filter is the scroll, not a skip button, so the opening has to stop the thumb rather than beat a countdown; the hook rate guide shows how to measure that. Check Meta's Ads Guide for the current length limits of each placement before you export a long cut.

How to measure long form video ads

Google counts a view on a skippable in-stream ad when someone watches 30 seconds, or to the end if the ad is shorter, or clicks on part of the ad, and it defines view rate as views divided by impressions (About YouTube ads and view metrics). It also reports how many viewers reached 25%, 50%, 75% and 100% of the video, and notes that those quartiles are less precise than the view count. Use them to diagnose the edit, not to judge the ad. The shape to expect from a long skippable ad is a cliff before the view counts, then a long, shallow tail:

Share of impressions still watching a 10-minute skippable ad

Illustrative curve in 30-second steps

0%25%50%75%100%0:002:004:006:008:0010:00View countsFirst CTAOffer6.3%

Time into the ad

Show the numbers

Time into the ad	Still watching
0:00	100%
0:30	21%
1:00	15%
1:30	12.5%
2:00	11%
2:30	10%
3:00	9.2%
3:30	8.6%
4:00	8.1%
4:30	7.6%
5:00	7.2%
5:30	6.7%
6:00	6.3%
6:30	5.9%
7:00	5.6%
7:30	5.3%
8:00	5.1%
8:30	4.9%
9:00	4.7%
9:30	4.6%
10:00	4.5%

In this illustrative curve, 790 of every 1,000 people who see the ad skip before it counts a view. Of the 210 who reach 0:30, 63 are still watching when the offer starts at 6:00, which is 30% of counted viewers. If the curve falls steeply between 0:30 and 1:00, rework the promise; if it falls steeply just before the offer, the proof runs too long. The retention graph guide reads each shape.

Then the number that pays the bills. Say you spend $3,000 on a ten-minute ad and get 60,000 impressions, 12,600 views (a 21% view rate) and 40 sales: $0.24 per view and $75 per sale. A 15-second version of the same offer buys cheaper views, $0.05 each, or 60,000 views for the same $3,000, but only 31 sales: $96.77 per sale. The long ad's views cost almost five times as much, and its sales cost 22.5% less.

Whether $75 is good depends on your break-even cost per sale. With a $120 average order, $22 of product cost, $9 of shipping, 4% in payment fees and 5% in refunds, each order leaves $78.20 before ads, so $75 a sale is a thin profit and $96.77 is a loss. Put in your own numbers:

Calculator

Break-even ROAS and CPA calculator

Your numbers per order. Below the break-even ROAS, every order loses money before fixed costs.

Average order value$Product cost per order$Shipping and fulfillment per order$Payment fees (% of the order)%Refunds and chargebacks (% of revenue)%Other costs per order$

Break-even ROAS

1.53x

Profit per order before ads

$78.20

Contribution margin

65.2%

Break-even cost per order (CPA)

$78.20

ROAS = revenue ÷ ad spend. Refunds are counted as revenue you give back; the product cost and shipping of a refunded order are already spent.

Reset to the example

Two cautions. Count the sales you can tie to a click first, and treat view-through and engaged-view conversions, which Google reports separately, as a softer signal. And compare what Google reports with the orders in your own store before you scale.

A long form video ad test plan

1. Check the fit: a product that needs a demo or a mechanism, and an order value with bundles that can carry the cost per sale.
2. Pick the model: sell in the ad, or warm up in the ad and sell on a page.
3. Cut three openings onto the same body, each with a different first five seconds.
4. Launch in Demand Gen with conversion bidding toward purchases, with enough budget per ad for a few dozen sales.
5. Judge on cost per sale against your break-even, after enough sales to trust the number. Use view rate and quartiles only to decide what to edit.
6. Recut the winner: tighten wherever the curve drops, then test a shorter version against it.