Rules
Part of Designing video advertising with a checklist built around the audience job
Video advertising best practices as experienced teams run them, updated for 2027
Video advertising best practices connect truthful claims, suitable versions, accessible meaning, controlled rights, verified delivery, outcomes, and recovery.
What to take away
- Run every video ad through one loopbrief, build, instrument, launch small, read, refresh.
- Choose ratio, duration and hook by placement, not by house style.
- Fix the outcome definition and the platform view rule before you spend.
- Set refresh triggers as numbers in advance.
- Review claims, captions and destination in the same pass as the edit.
The workflow, brief to first live version
Experienced teams run video advertising as a short loop. Each pass leaves one decision and one number behind, so the next pass starts from evidence.
- Brief.Name the audience state, the placement, the outcome and the single claim the asset carries. One job per asset.
- Claim file.Gather substantiation before the script exists. The FTC's Rules of the Road for internet advertising requires online claims to be truthful and substantiated, so evidence sets the wording.
- Build.Cut to the placement spec: ratio, duration, hook timing, caption file, end card, landing page. Watch it sound-off on a phone before it leaves the edit.
- Instrument.Fire the conversion event in a live test. Set naming that survives a spreadsheet. Confirm the landing page matches the promise in frame one.
- Launch small.Change one variable at a time. Keep one placement per ad group where the platform allows it, and fund enough volume for the bidding model.
- Read and refresh.Book the first review date before launch, not after.
The first review checks the asset, the tracking and the destination against the brief, then names one change for the next version.
Platform practices by placement
Each placement is its own craft. Video and banner budgets often come from the same pot, yet the rules diverge, which is why the display advertising best practices comparison is worth reading beside this one.
Formats and typical durations
- YouTube
- 6s bumper, 15s and 30s skippable in-stream, in-feed up to 60s
- Meta (Facebook, Instagram)
- Reels and Stories 9:16 up to 90s, feed 4:5
- TikTok
- 9:16, roughly 9s to 60s
- CTV
- 15s and 30s, 16:9, non-skippable
What the placement demands
- YouTube
- Earn the first five seconds; skips start there
- Meta (Facebook, Instagram)
- Sound-off default, so captions are burned in
- TikTok
- Native framing and pacing; Spark Ads for creator cuts
- CTV
- Living-room viewing, so cap frequency hard
Short cuts carry one claim. A 30-second story trimmed to 15 seconds usually loses its proof, so rebuild rather than cut.
Budget, bidding and the numbers to watch
Set the outcome before the bid strategy. A lead, a purchase and a booked call are three objectives and three reports.
Maximize Conversions and Target CPA need conversion history. Google's Maximize Conversions guidance describes how the system sets bids to spend the budget at the best cost per conversion it finds. A common working figure is 15 conversions in 30 days per campaign before the model stabilizes. Below that, stay on manual or maximized clicks.
Write the attribution window into the report header and hold it steady across a test. A 7-day click window and a 1-day view window tell a different story from clicks alone. Your own 90-day baseline beats any published average, a case made in our video advertising benchmarks piece.
View rules decide what a report counts:
- YouTube counts a view at 30 seconds or completion, whichever comes first.
- Meta separates 3-second plays from ThruPlay, defined as 15 seconds or completion.
- TikTok counts a 6-second view.
- CTV inventory is usually non-skippable, so completion is high by design.
Relevance feeds cost. Google's Quality Score page lists expected click-through rate, ad relevance and landing page experience as inputs. All three can be fixed in the edit or on the page, and none of them require a bigger bid.
Set refresh triggers as numbers, and let any one of them start a rebuild:
- Frequency above three impressions per person per week.
- Click-through rate down 20% across two weeks.
- Cost per outcome up 25% against your baseline.
Run the video advertising checklist before the next flight goes live, so the new cut, the tracking and the destination are verified together.
Example: a claim caught in the first weekly read
An installer ran a 20-second in-feed video with the line "cut break-ins by half". The claim file held a monitored-alarm study that excluded camera-only systems, so the number did not match the product on screen.
On-screen text became "homes with monitored alarms reported fewer break-ins in one study". The end card linked to the study summary. A new version went up, the old asset stayed paused for the record, and missing captions were added. Cost per lead dipped during relearning, then recovered.
Common questions
How long should a video ad run before I judge it?
Two weeks, or enough spend to reach the conversion volume the bidding model needs, whichever comes second. Read frequency and cost per outcome together.
Do I need a separate cut for every platform?
Yes for ratio and duration. Reuse the footage and rebuild the edit. A 16:9 cut cropped to 9:16 clips captions and the end card.
Who owns the refresh decision?
One named person holds the trigger list and the calendar. Editors and buyers supply the numbers, and that person signs the rebuild.
Is a transcript enough for accessibility?
No. Captions, audio description, player controls and language all matter, and the right alternative depends on the content and the user.
