YouTube Ads for performance-minded businesses: which campaigns to use in 2026 and how to run a first test

When a client asks us whether they should "do YouTube", they usually have a TV ad in mind. In 2026 the question is a different one: YouTube is a surface on which Google runs several campaign types, some built to be seen and others built to drive conversions, with very different bidding and measurement rules. Picking one at random is the quickest route to deciding that "YouTube doesn’t work", when in fact the wrong tool was used or the wrong number was read.

What has changed: conversions moved to Demand Gen

For years the "performance" campaign on YouTube was the Video Action Campaign (and TrueView for Action before that). It no longer exists. Google folded it into Demand Gen in several steps:

• March 2025 An upgrade tool becomes available to carry settings and history from Video Action Campaigns over to Demand Gen.
• April 2025 New Video Action Campaigns can no longer be created.
• July 2025 Google starts automatically upgrading the remaining campaigns.
• April 2026 The last campaigns are automatically upgraded to Demand Gen.

In practice: if you want YouTube to deliver measured leads or sales, the campaign is Demand Gen, which serves on YouTube (in-stream, in-feed and Shorts) as well as Discover and Gmail. How to set it up — audiences, assets, controls — is covered in our Demand Gen guide; here we look at YouTube as a whole: which campaign for which goal, which formats, and how to measure.

The campaigns that run on YouTube, and what they are for

There are essentially three ways to buy YouTube inventory through Google Ads today:

Two details that are often missed. First: since April 2025 maximum CPV has not been available for new video views campaigns; it was replaced by target CPV, where you state the average cost per view you are willing to pay. Second: Performance Max also shows ads on YouTube, and if you do not upload a video it can generate one automatically from your images and text. In August 2026 Google also announced AI adaptation of uploaded videos into missing formats (vertical, square), which can be switched off in the campaign’s asset optimisation settings. If you are not controlling video in PMax, someone — the algorithm — is doing it for you; we cover this in our piece on PMax channel reporting and controls.

The formats: what the viewer actually sees

• Skippable in-stream. Plays before, during or after a video; after 5 seconds the viewer can skip. This is the format almost everything hinges on, and the reason the opening seconds matter so much.
• Non-skippable in-stream. Up to 15 seconds with no skip button. On TV screens longer durations exist too, up to 30 seconds, designed for living-room viewing.
• Bumper. Up to 6 seconds, non-skippable. It is for repeating a message, not explaining it.
• In-feed. A thumbnail with text in YouTube search results, on the home feed and alongside related videos: the viewer chooses to click and watch. Higher intent, lower volume.
• Shorts. Vertical videos between one Short and the next. For Google, a TrueView view on Shorts is counted at 10 seconds (or the end, if shorter) or on a click of the call to action.

What about connected TV?

A growing share of YouTube viewing happens on the television, and reach campaigns and Demand Gen can get there. For a business chasing direct conversions, though, there is practically no click on a TV: that is where you build memory, which you then harvest on Search or on your site. Worth knowing up front, so you do not judge those impressions on cost per click.

Creative: the ABCD framework

On YouTube, creative outweighs targeting. Google distils its research on effective ads into the ABCD framework, and according to Google ads that follow it show a 30% lift in short-term sales likelihood and a 17% lift in long-term brand contribution:

1. Attract — grab attention immediately. Tight framing, a person or the product on screen from the first frame, quick pacing. Not a logo spinning for three seconds on a white background.
2. Brand — make clear who you are, early and naturally. The brand (spoken as well as shown) in the first seconds, because most people will skip.
3. Connect — give people a reason to stay. A recognisable problem, a before and after, a demonstration. People remember stories, not feature lists.
4. Direct — ask for a specific action. "Request a quote", "try it free", not "find out more" mumbled at the end.

The way we apply it with smaller businesses is very concrete: a well-shot 15–30 second video filmed on a phone, with the customer’s problem in the first two seconds, the company name within the first five and a call to action both spoken and on screen, often outperforms an expensive ad made for television. Produce the same idea in landscape, vertical and square: otherwise on Shorts and some placements Google will adapt it for you, with results you may not like.

Bidding strategies, briefly

• Target CPM for reach and frequency campaigns: you pay for impressions and optimise the cost of reaching unique people.
• Target CPV for video views campaigns: you state what you want to pay on average per view.
• Maximise conversions / target CPA and maximise conversion value / target ROAS in Demand Gen and Performance Max, as elsewhere in the account.

The rule that matters on small budgets is the same as on Search: conversion-based strategies need conversions. In its Demand Gen guidance, Google suggests daily budgets of around 15–20 times the expected CPA, with an indicative floor of roughly 100 dollars a day per campaign. Since 1 April 2026 the API has also enforced a minimum of 5 dollars a day on Demand Gen campaigns. There is a lot of ground between the technical minimum and the recommended one, and that is exactly where you decide whether the test makes sense.

Measuring without fooling yourself

This is where most arguments about YouTube are won or lost. Google Ads can credit a video ad with three kinds of conversion, and they need to be kept apart:

Engaged-view conversions underpin much of the evaluation of Demand Gen and video campaigns, and rightly so: someone who watches ten seconds of an ad and buys the next day is not a coincidence. But they are not the same as a click. Our approach:

1. Always report click and engaged-view conversions separately, using the segments available in Google Ads. A "great" CPA that is 80% EVCs is a different piece of information from one made of clicks.
2. Do not lengthen the EVC window to make the numbers work. Three days is a reasonable compromise; stretching it to 30 grows the figures, not the sales. Note that since 11 August 2026 GA4 also lets you edit the EVC window, previously fixed at 3 days: make sure both platforms are telling the same story.
3. Look at what happens around it. Brand searches, direct traffic and total site conversions in the areas and periods where YouTube is live. If everything stays flat, the EVCs are probably describing people who would have bought anyway.
4. Measure incrementality where you can. Conversion Lift and Brand Lift studies now sit in the Experiments section of Google Ads. Brand Lift requires minimum budgets of several thousand dollars over a few days, varying by country and number of survey questions; for many smaller businesses the realistic route is a geo test with control areas, as explained in our incrementality testing guide.

When YouTube makes sense on a small budget, and when it does not

Where we see it work for businesses that are not large:

• Search is already well covered and growth can no longer come from expressed demand. YouTube creates the demand that Search then captures.
• The product needs showing: a machine, a service with a before and after, software that makes sense in thirty seconds of screen time.
• You have first-party lists and remarketing audiences (customers, site visitors, people who watched your videos) to start from.
• Tracking is reliable, consent included, and the account already has a steady flow of conversions.

Where it is rarely worth it:

• The overall budget is already thin. If you are struggling to buy all the qualified demand on Search, every pound on YouTube is taken from the most efficient part of the account. The reasoning is the same as in our budget guide.
• No video and no way of producing one. Letting Performance Max generate a video from your images is not a creative strategy.
• A long sales cycle with no CRM data connected. If a lead becomes a customer months later and Google Ads never hears about it, no video campaign can optimise for quality.
• Urgent-need businesses — emergency call-outs, "near me" services — where demand is triggered by a problem, not an ad.

A first test that tells you something

The test we most often propose to a business starting from scratch on YouTube with a clear focus on results:

1. Prerequisites (week 0). YouTube channel linked to Google Ads, primary conversions verified, consent passed correctly. Two or three 15–30 second videos built on ABCD, each in landscape and vertical.
2. One conversion-focused Demand Gen campaign, restricted to YouTube channels if you want to isolate the video effect, bidding on Maximise conversions with no target CPA for the first few weeks, audiences starting from customers and site visitors.
3. A budget fixed in advance, large enough to collect a few dozen conversions over 6–8 weeks. A purely illustrative example: with an expected CPA of £40, you need roughly £1,200–£2,400 a month to see 30–60 conversions; if that figure is not sustainable, the test is not ready.
4. A control area. If the business covers several regions, leave one or two comparable areas out of the targeting: comparing total conversions and brand searches between exposed and unexposed areas is the most honest read you have.
5. Decision criteria written down before launch: which click CPA, which share of EVCs and which effect on brand searches would make you carry on. You do not change the yardstick halfway through.
6. A creative review at the midpoint: look at view rates and where people drop off. If they leave in the first 5 seconds, the problem is Attract, not the audience.

If you want to know whether YouTube makes sense for your account before shooting a single frame, our free audit starts from the data you already have: uncovered Search demand, available audiences, tracking quality. You can see which services include video campaign management, and how we work in our process.