Operations
Part of Running connected TV advertising operations without delivery surprises
Before you sign off a connected TV advertising quality checklist
A working connected TV advertising quality checklist for UK teams: the four checks most often skipped before a campaign goes live, with a labelled pound example.
What to take away
- Most teams treat quality as a final sign-off, then discover the creative, the audience definition and the measurement plan disagree with each other.
- Build the checklist around four checkswhat actually plays, who actually sees it, what counts as a view, and who owns the fix.
- Every check needs a named owner and a written pass or fail, not a verbal nod in a launch call.
- The checks below are cheap to run before launch and expensive to run afterwards, when spend is already committed.
The four checks that stop most launches
Check what actually plays on the device
A file that passes your internal review can still fail on a specific television set. Test the finished asset on the oldest and cheapest device in your target mix, not the newest one in the office.
Four pre-launch checks
- Test asset on oldest target device
- Confirm which cut is live
- Verify audience segments exist
- Agree view definition before invoice
Look at safe areas, text legibility at viewing distance and audio levels. A call to action that sits inside an overscan zone is invisible on some sets. A quiet mix disappears under a drama soundtrack.
Version control matters as much as playback. If three cuts exist in the platform, confirm which one is live and which are paused. A mislabelled variant can run for days before anyone notices.
- Pass / [ ] FailPlayback: pass only if the finished asset plays correctly on representative devices in the target mix, including the oldest and cheapest device tested.
- Pass / [ ] FailSafe areas and legibility: pass only if essential text and the call to action are not cropped and can be read at normal viewing distance.
- Pass / [ ] FailAudio: pass only if dialogue and other key audio are clear and meet the destination platform's published audio specification.
- Pass / [ ] FailVariants: pass only if the approved cut is live, other cuts have the intended status, and the platform labels match the asset list.
For spot-based buying on broadcast video on demand, the same creative discipline applies as linear television. Thinkbox's introduction to spot advertising sets out the format conventions buyers are expected to follow.
Check that the audience definition is executable
A planning slide can describe an audience that no platform can actually target. Before launch, confirm the segments exist in the buying platform, in the same geography, with enough forecast volume.
Ask for a forecast that shows overlap between segments. Two audiences that look distinct in a deck can resolve to largely the same households once the platform applies deduplication.
Write down the fallback. If a segment underdelivers, do you widen it, swap it or pause? Decide before the first report, not during it.
- Pass / [ ] FailBuying route and inventory: record the actual route and service, such as DV360 or The Trade Desk, and the relevant inventory, such as Sky AdSmart, ITVX, Channel 4, Netflix, Amazon Ads, Roku, Samsung Ads or LG Ad Solutions. Pass only if the campaign order and available preview or test access identify the same service, format and targeting; confirm availability for the specific buy.
- Pass / [ ] FailSegment delivery: pass only if each planned segment exists in the named buying environment, uses the intended geography and has forecast volume that meets the campaign's documented delivery requirement.
- Pass / [ ] FailOverlap: pass only if the forecast shows segment overlap and the deduplication method is recorded.
- Pass / [ ] FailFallback: pass only if the plan says who can widen, swap or pause an underdelivering segment and who approves that action.
For the wider delivery process, see the connected TV advertising operations and delivery guide.
Views, currency and regulation
Agree what counts as a view before the invoice
A completed view, a viewable impression and a click are three different things. Agree which one the campaign is bought on, and make sure the platform reports it in the same way.
View definitions compared
Completed view
- What it counts
- Full ad watched
- Bought on
- Often
- Reconciles with Barb
- May not
Viewable impression
- What it counts
- Ad in view
- Bought on
- Sometimes
- Reconciles with Barb
- May not
Click
- What it counts
- Interaction
- Bought on
- Rarely
- Reconciles with Barb
- No
Agree the reporting naming convention at the same time. If the platform calls a metric one thing and your dashboard calls it another, the reconciliation problem starts before the first invoice.
- Pass / [ ] FailBought metric: record whether the campaign is bought on completed views, viewable impressions or clicks, and name the reporting source. Pass only if the order and platform report use the same agreed metric.
- Pass / [ ] FailDefinition: for a completed view, record the platform's completion-event rule and denominator; for a viewable impression, record the named measurement provider's definition and confirm it applies to the CTV inventory; for a click, record the platform's attribution rule. Do not treat these metrics as interchangeable.
- Pass / [ ] FailTarget and reconciliation: record the campaign's agreed target, metric labels and any mapping into the dashboard. Pass only if the reporting plan identifies how the platform figure will be reconciled with the dashboard and, where relevant, Barb currency.
Where you buy against UK audience currency, understand what is being measured. Barb publishes the joint industry currencies that underpin UK television audience measurement, and platform-reported figures may not reconcile with them.
If the numbers will not reconcile, say so in the plan rather than in a post-campaign argument.
Check the regulatory and consent position
Digital advertising in the UK sits under consumer and competition law as well as platform policy. The Digital Markets, Competition and Consumers Act 2024 covers areas including unfair commercial practices. For advertising claims, identify the relevant rules: the ASA's CAP Code for non-broadcast ads or the BCAP Code for broadcast advertising. Ofcom regulates broadcast television and certain on-demand programme services, so confirm the regime that applies to the service carrying the ad.
Check consent for any audience data used in targeting. Record the lawful basis under UK GDPR for processing personal data, check ICO guidance and establish whether PECR requires consent for any tracking or storage/access technologies used. Obtain and evidence consent where required. Also check that any price, saving or availability claim can be evidenced on the day it runs.
- Pass / [ ] FailClaims and rules: pass only if the relevant CAP Code or BCAP Code has been identified, claims have been reviewed against it, and supporting evidence is retained. Record any applicable Ofcom requirements for the service carrying the ad.
- Pass / [ ] FailAudience data and consent: pass only if the data, purpose and UK GDPR lawful basis are documented, ICO guidance has been checked, and any consent required under PECR has been obtained and recorded.
Keep the evidence, not just the approval. A signed-off claim with no supporting record is hard to defend if a complaint arrives months later.
Check ownership of the fix
Every check above needs one name against it. A checklist with no owner becomes a document nobody reads. For a worked example, a team paying £400 a month for a small streaming test can still lose a week of that spend if nobody owns creative approval.
Use a campaign launch sign-off sheet as the record. For every check, enter the pass or fail, named owner, evidence or test result, and any corrective action; record the overall launch decision and who approved it. Do not launch while a failed check has no owner or agreed resolution.
For role allocation, see the guide to connected TV advertising team roles.
Common questions
How long should the checklist take to run?
For a modest campaign, a few hours spread across the week before launch. The creative test is the slowest part because it needs a real device.
Can one person sign off every check?
Only in a very small team. Otherwise the person who built the audience definition should not be the only person confirming it works.
What if a check fails on launch day?
Pause the affected line rather than the whole campaign. A failed creative and a failed audience segment rarely need the same response.
Does this replace platform quality controls?
No. Platform checks catch policy and technical faults. This checklist catches the planning and ownership gaps platforms cannot see.
