YouTube pre-roll ads got a lot more interesting between 2022 and 2026. Skippable in-stream is still the workhorse, but bumper ads, non-skippable in-stream, and the newer Shorts in-feed format each serve different jobs in the funnel.
Most pre-roll advice on the internet is still optimised for the 2018 era — "the first 5 seconds matter" type stuff. True, but not useful for actually structuring a YouTube programme. Here's the current playbook.
The four ad formats and what each is actually good at
Skippable in-stream (TrueView)
The classic. Plays before/during/after a YouTube video; viewers can skip after 5 seconds. You only pay if someone watches 30 seconds (or to the end of a shorter ad) OR clicks.
Best for: middle-funnel awareness + consideration. The cost-per-completed-view metric is honest — only engaged viewers cost you money.
Right length: 30-60 seconds for awareness, 90-120 for considered purchases (B2B SaaS, high-AOV ecommerce, considered services).
Common mistake: running a 15-second skippable. Skip rate ends up around 75-85% because the audience hasn't decided yet. The format works best when you give people a reason to keep watching past second 5.
Non-skippable in-stream
15-30 seconds, viewers can't skip. Sold on a CPM basis. Higher cost per impression but guaranteed completion.
Best for: brand awareness on a known message + creative that's been validated as not annoying. Risky for the wrong creative — you're forcing people to watch.
When it works: awareness campaigns where reach + frequency matter more than per-click economics. Particularly effective when paired with skippable in-stream (run skippable to identify resonant creative, then push the winners through non-skippable for guaranteed reach).
Bumper ads
6 seconds, non-skippable. Sold CPM. Designed for mobile.
Best for: top-of-funnel reach. The cost per impression is low and the format is short enough that people don't resent it.
Common use: as a frequency builder paired with longer skippable ads. Run skippable for the story; run bumpers for the brand reminder. Sequential targeting (skippable first, bumper to viewers who watched 25%+ of the skippable) is a powerful combo.
YouTube Shorts in-feed
Vertical, 60 seconds max, served between Shorts content. Bid on CPV or CPM.
Best for: reaching the audience that's moved away from long-form YouTube. Works well for DTC ecommerce — particularly for product demos and UGC-style ads.
Critical: use vertical (9:16) creative made FOR Shorts, not horizontal repurposed for vertical. The TikTok-native aesthetic outperforms produced TV-ad creative on this surface by 3-5x in our client testing.
The 5-second hook discipline
Whether someone skips a skippable in-stream ad is determined in the first 5 seconds. The rest of the ad runs (or doesn't) based on whether you earn the next 25 seconds.
What works in the first 5 seconds (from testing across our DTC + B2B client portfolio):
- A direct callout to who the ad is for ("If you run a 7-figure DTC brand...")
- A specific claim that contradicts conventional wisdom ("Most attribution data is lying to you")
- A surprising visual (something unexpected on screen)
- A direct question the target audience is currently asking
What doesn't:
- Logo first (looks like a TV ad → instant skip)
- Generic montage of "happy customers"
- Slow build to a punchline
- Music-led intros without dialogue
Targeting structure that works in 2026
YouTube's targeting moved heavily toward audience-based and away from keyword/topic targeting in 2024-25. The old playbook of "target keywords related to my product" is largely deprecated. The new structure:
For DTC ecommerce
- Custom segments built from competitor product searches + niche YouTube channels related to your category
- Customer Match using your existing customer email list — YouTube finds lookalike viewers
- In-market audiences for product categories you sell into
- Remarketing from your website (visitors who didn't convert + cart abandoners)
For B2B SaaS
- Custom segments built from job-title-relevant searches (e.g., "how to do incrementality testing" for marketing analytics audience)
- Customer Match from CRM data + free-trial users who didn't convert
- Affinity audiences for the psychographic match (e.g., "Avid Investors" for fintech, "Technophiles" for dev tools)
Bidding + budget
Three bidding modes for pre-roll, in order of how aggressive you should be with each:
- Maximize Conversions — let YouTube optimise toward your conversion event. Best for direct response when conversion volume is high enough (>30 conversions/week)
- Target CPA — set a CPA you're willing to pay; YouTube optimises within that. Use after Maximize Conversions has produced ~50 conversions
- Target CPV — for awareness-only campaigns where view-through is the goal
Budget recommendation for testing: minimum £100/day per ad set for 7 days to give YouTube enough data to optimise. Below this, you'll get noisy results that lead to bad creative-killed-by-poor-statistics decisions.
Measuring pre-roll properly
The metric that actually matters: incremental conversions, not the platform-reported "conversions" number.
YouTube's reported conversion data overstates impact by 30-60% in most accounts (view-through attribution + own-channel double-counting). The honest measurement is via geo holdout test or a properly-designed conversion lift study.
If your YouTube reporting says you're at £25 CAC and your blended CAC is £55, the truth is that YouTube is contributing somewhere in the middle — usually 60-75% of what the platform reports.
The takeaway
YouTube pre-roll, run well, is one of the more cost-effective awareness + consideration channels available in 2026. Run badly — wrong format for the funnel stage, weak hook, sloppy targeting — it's a money pit.
The brands getting it right treat each format as a different tool: skippable for story-driven persuasion, non-skippable for known-winning creative, bumpers for frequency, Shorts for the algorithmic-discovery audience. Most brands using "YouTube ads" as one undifferentiated thing are leaving the format leverage on the table.
