A client reads the video report, sees completion rate at 38% on the YouTube line and 94% on the CTV line, and asks the obvious question: what went wrong on YouTube? Nothing went wrong. One line counts people who chose to keep watching an ad they were allowed to skip after five seconds, and the other counts a non-skippable spot on a connected TV that nobody could skip and probably nobody was holding. Averaging those two into a single “video completion rate: 61%” tile, which is what most templates do, produces a number that describes neither placement and invites exactly the question you just got.
Video campaign report metrics break more often than display or search metrics because the same word means different things across formats. Completion, view, and even “video play” are defined by the platform, not by the industry, and a six-second bumper and a thirty-second CTV spot are not competing on the same scale. The fix is not a better chart. It is refusing to blend placements that answer different questions, and saying in the report which definition you used.
A 40% completion rate means nothing until you know whether the ad could be skipped
Start with the mechanic of the placement, because that sets the ceiling before targeting or creative gets any credit or blame.
| Format | Can the viewer skip? | What a high completion rate tells you |
|---|---|---|
| Skippable in-stream (YouTube, programmatic video) | Yes, usually after 5 seconds | Genuine signal: the creative earned the remaining seconds |
| Non-skippable in-stream (15 to 20 seconds) | No | Mostly a delivery and viewability signal |
| Bumper (6 seconds) | No | Near-structural, close to an impression count |
| CTV and OTT | No, and often no click either | Delivery quality, ad pod placement, fraud check |
| Social feed video (Meta, TikTok) | Yes, by scrolling, instantly | Harsh signal: no forced exposure at all |
Read down that table and the YouTube versus CTV comparison stops being a performance story. A 38% completion rate on skippable in-stream means roughly four in ten people who could have left after five seconds stayed to the end, which is a creative result worth discussing. A 94% completion rate on CTV mostly means the spots served in valid pods and the stream did not break. Both belong in the report. Neither belongs in the same tile as the other.
That is also why a single dashboard layout across channels falls apart on video faster than anywhere else, a problem worth reading about in its own right in why one dashboard fails every channel.
View-through rate and video completion rate are not the same number
This is the one that causes real credibility damage, because both get abbreviated and both get called “views” in conversation.
Completion rate is the 100% quartile: impressions that reached the end of the creative, divided by impressions served. It is a clean ratio with a fixed definition in the export.
View-through rate depends on what the platform calls a view. On skippable YouTube in-stream, a view is counted at thirty seconds, or at completion for shorter creative, or on an interaction. So a thirty-second asset and a fifteen-second asset generate views under different conditions, and VTR is not comparable between them. On Meta, the counted events are different again: a three-second video play is one metric, ThruPlay (fifteen seconds or completion, whichever comes first) is another. A template with one tile labelled “VTR” fed from two platforms is adding apples to a different fruit entirely.
Two practical habits fix most of this. Put the definition in the report, in a short line of text under the tile, naming the platform and the threshold. And when a client asks which number matters, answer with the question they are actually asking: VTR for whether the audience chose to watch, completion rate for whether the full message landed, quartiles (25/50/75/100) for where they stopped. The quartile curve is the most useful chart in a video report and the most frequently left out, because it shows the drop-off point rather than the aggregate. If 50% of viewers make it past the halfway mark and only 12% finish, your creative has a problem in its second half, and no single rate will tell you that.
CPCV punishes long creative, not weak targeting
CPCV reporting goes wrong in a specific way. Take two placements in the same campaign, both hypothetical but ordinary. A six-second bumper buys at a €12 CPM and completes 85% of impressions, so 850 completed views per thousand, giving a CPCV of about €0.014. A thirty-second skippable unit buys at a €14 CPM and completes 28%, so 280 completed views per thousand, giving a CPCV of €0.05. Three and a half times more expensive per completed view.
Shown as a league table sorted by CPCV, the bumper wins by a distance. But the bumper delivered six seconds of message and the skippable unit delivered thirty seconds to a self-selected audience, and a client who bought the thirty-second asset to explain a product feature did not get the same thing for their money. CPCV compares formats of equal length fairly and formats of unequal length not at all.
So use CPCV inside a format, not across formats. Comparing CPCV between two publishers running the same fifteen-second asset is a real efficiency question. Comparing CPCV between bumper and in-stream is a creative-duration question wearing a cost metric’s clothes. If the client is going to see one cost line per placement anyway, CPM plus completion rate side by side is harder to misread than CPCV on its own.
CTR belongs in a video report, just not at the top
Video CTR is typically a fraction of a percent, and on CTV it often does not exist at all because there is nothing to click. Leading a video report with CTR trains the client to judge the campaign on the metric that the format is worst at, the same trap covered in display reporting metrics that beat CTR as the headline.
Where CTR earns its place is as a secondary signal next to completion: a creative with high completion and near-zero clicks did its storytelling job without producing intent, which is a useful finding for the next flight. Keep it in the table, out of the KPI row.
CTV and OTT add a completion metric paid social does not have
CTV OTT ad metrics come with structurally high completion and no scroll, which changes what you report. Reach and frequency carry more weight here than any rate, because the buy is closer to a television buy than a performance buy. Frequency capping failures also show up in CTV reporting before anywhere else, and a client who saw the same spot five times in one evening will ask about it regardless of what the dashboard says.
Two caveats worth writing into the report rather than explaining over email. Co-viewing means an impression can represent more than one person, so reported reach on CTV and household reach are different concepts. And completion near 100% should be treated as a hygiene check: if it drops below the usual range for that inventory, something is wrong with delivery, not with the audience’s interest.
Paid social video sits at the opposite end, where the counted “view” is short, the drop-off is immediate and three-second metrics dominate. That asymmetry is why the social campaign metrics template needs its own rows rather than borrowing the video ones.
View-through conversions, with the caveat stated up front
View-through conversions matter in video reporting because clicks are rare and the format works on exposure. They are also the easiest number in the report to overstate. If you include them, state the attribution window and the fact that they are a modelled or exposure-based credit, in the same line as the figure, before the client’s internal analyst finds the discrepancy against the brand’s own platform. Reporting view-through conversions in a separate column from click-through conversions, never summed into one “conversions” total, keeps the conversation honest. The wider principle, which metrics deserve the headline and which deserve a footnote, is worked through in the twelve campaign metrics every client report should show.
Split video campaign report metrics by placement in the template
The reason most video reports blend placements is that the template has one row for video, and the account manager fills it with whatever the export aggregated. Adding a second and third row per format takes ten minutes once and then stops being a decision.
In Screebl, that looks like a reusable template where one row is grouped by placement with completed views and spend, another holds the quartile chart, and a creatives slider shows the actual assets in phone and laptop frames next to their drop-off numbers, since “the second version held attention longer” lands better when the client can see both. Rates like VTR and CPV are computed from the totals rather than averaged from row-level percentages, which removes one of the quieter sources of wrong numbers in video reporting. If you want to see which plan carries client branding alongside your own, the pricing page lists it.
The answer to the question that started this, written into the report rather than the reply email, is two sentences: YouTube completion is at 38% because viewers can skip after five seconds and this is the creative’s real retention; CTV sits at 94% because the spot cannot be skipped, so we read it as a delivery check and judge that line on reach and frequency instead.
