TV and CTV ad creative

Linear broadcast is scheduled television: you buy a daypart, a program, or a rotation, and the spot airs to whoever is watching. CTV/OTT is streaming inventory bought digitally: the spot is served to a household or device, often with frequency controls and audience targeting closer to what digital buyers expect.

The creative implications matter more than the buying ones. Linear gives you no interactivity at all: the viewer cannot click, and the only actions available are remembering you, searching for you, or scanning a QR code before it leaves the screen. CTV inventory can support interactive overlays and QR-to-action formats, and it reports delivery metrics linear never could. But CTV also enforces broadcast-standard technical rules: 16:9 aspect ratio is the norm, and a vertical or social-media crop will be rejected on 16:9-only inventory. Do not recut a feed ad to length and call it a TV spot; the aspect ratio alone can kill it before a human ever judges the creative.

The honest concession: this channel gives you reach and memory, not clicks. There is no CTR, no same-day creative signal, no comment section telling you the hook missed. You trade feedback speed for attention quality, so the creative review has to happen before the spot ships, not after.

A :15, a :30, and a :60 are three different writing problems that happen to share one rule: the opening 3 to 6 seconds must earn attention. That window is shared across every duration. Spend it on atmosphere, a slow logo resolve, or brand wallpaper and the rest of the spot plays to a room that has already picked up a phone.

• —:15 has almost no runway. Establish the problem or the product immediately, then land one memorable brand-plus-CTA beat. There is no room for a second idea. If the script has a subplot, it is a :30 pretending to be a :15.
• —:30 earns a full miniature arc: a clear setup, a proof point or an emotional turn, and a payoff. This is the classic broadcast unit, and the most common failure is a middle that wanders away from the promise the opening made.
• —:60 can carry a real narrative: escalation, demonstration, objection handling, even character development. The risk inverts: with this much room, the selling idea can stall while the film gets admired. Every act still has to move the idea forward.

If narrative structure is new territory, the three-act logic in storytelling structure applies here more literally than anywhere in short-form: TV is where those frameworks came from.

A logo at the open and an end card at the close is not brand integration; it is bookends on someone else's film. Strong spots carry a brand, name, or value-proposition beat in the middle third, where the story is doing its heaviest work. The working heuristic: a viewer who sees only the middle ten seconds should still know who the ad is for and roughly what is being sold.

This is a craft judgment, not a platform rule, but it exists because TV viewing is partial by nature. People join spots late, leave early, and half-watch through conversation. A spot that only resolves its brand in the last two seconds is betting the entire media buy on complete, attentive views, which is not how living rooms work.

TV inverts the short-form default: sound is normally on, and the mix is a first-class craft surface. Dialogue intelligibility, music that supports rather than smothers, and a deliberate relationship between voiceover and picture are all reviewable before a single dollar of media runs.

The highest-value audio asset in the channel is a distinctive sonic logo or audio mnemonic: a short, ownable sound that accumulates recognition across exposures. A spot that introduces or reinforces one is building an asset the next campaign inherits. A spot that ends on generic library music is spending media on audio nobody can attribute. The broader craft toolkit in sound design still applies; the difference is that here you can rely on the audio actually being heard.

Legal copy on TV is a visible craft signal, not a checkbox. A disclaimer is not present just because it flashes on screen. The working standard: allow roughly 3 seconds minimum on screen for the first legal line, plus roughly 1 additional second for each additional line, at a size and contrast a viewer can read. Rushed, tiny, low-contrast, or obscured legal copy is a flaggable defect, and on regulated categories it is a rejection risk.

The 10-foot rule extends the same logic to everything on screen: the viewer is across the room, so text and QR codes must be legible at typical living-room distance (as a working equivalent, think roughly 36px text on a production frame). A QR code also needs dwell time; roughly 3 or more seconds on screen is the working floor for a scannable code. A QR that appears for one second is decoration.

Beyond legibility, broadcast and CTV inventory enforce regulatory and technical standards that feed platforms do not. Loudness rules under the CALM Act govern how loud a US spot may play relative to programming; this is measured in post, not eyeballed from frames. Photosensitivity standards flag strobing and flashing sequences. Prescription-drug spots require Important Safety Information and fair-balance voiceover, and their absence is a likely rejection. Alcohol creative carries talent-age restrictions on who can be shown consuming or promoting. None of these are optional craft preferences; they are gates the spot must clear before it airs.

On linear, the CTA menu is short: an end card, a verbal instruction, a memorable URL, or a QR code. Everything must be actionable after exposure, from memory. That pushes the craft toward simplicity: one URL, spoken and shown, short enough to survive the walk to a laptop. “Search [brand]” often beats a URL nobody will retype. The layered CTA thinking in CTA architecture still applies, but the primary action is always deferred.

On CTV, interactive and QR-to-action CTAs are legitimate, but do not award the creative points merely for being clickable. The on-screen instruction, the destination promise, and the timing all have to make the action usable: a viewer holding a remote is not a viewer holding a thumb over a link. If the interactive layer fails, the spot should still work as a pure brand film, because for many viewers that is what it will be.

The channel does not give you a click stream, so build a decision loop from directional signals and say out loud that they are directional. Working practice, not guarantees:

• —Recall and brand-lift surveys measure whether the spot registered and who it registered for. Slow, but the honest core metric of a memory channel.
• —Matched-market tests: run the spot in some markets and not others, then compare branded search, direct traffic, and sales movement between them. The cleanest causal read available without a click.
• —Vanity URLs, promo codes, and QR scans attribute a slice of response directly. They undercount, so treat them as a floor, never the total.
• —CTV delivery metrics: completion and viewability are real numbers on CTV, framed as directional. In The Ad Bench's working calibration, viewability around 93% and view-through around 70–90% for :15 to :30 spots are the bands worth sanity-checking against, not targets a spot is promised to hit.

What you must not do is import feed-video jargon. There is no hook rate on linear TV, no save rate, no scroll to stop. Reach, frequency, recall, and response codes are the vocabulary; anything else is borrowed language describing data the channel does not produce. For the cost side of the buy, the CPM logic in what is CPM is the one metric that translates cleanly.

Upload the spot as a video file, paste the script, or drop a public URL. The analyzer classifies the medium automatically and scores long-form and connected-TV creative against a dedicated long-form rubric rather than feed-video criteria; you can also override the format explicitly. The rubric's primary driver is the broadcast story arc, brand integration, and clarity of the selling idea across the full spot, not a social-feed scroll-stop.

Concretely, the review checks the duration-calibrated act structure above, the shared 3-6 second opening window, the middle-third brand beat, dialogue and mix quality, sonic-logo reinforcement, legal-super timing and legibility, the 10-foot rule on text and QR codes, and placement-appropriate CTAs (end card, verbal, URL, QR for linear; interactive allowed on CTV). It flags regulatory territory it can see, such as visible strobing, missing prescription-drug safety information, or alcohol talent-rule risk, and it marks what cannot be judged from frames or script, such as CALM Act loudness, as a verify-in-post check rather than pretending to measure it. If the placement is not stated, the review says so and evaluates the craft without inventing a delivery environment. Inputs up to 30 minutes are supported on this rubric; longer inputs are flagged as outside its calibrated scope.

What the score can tell you: whether the spot's structure, brand integration, legibility, and CTA are sound before you commit media. What it cannot tell you: whether the campaign will perform. See the scoring rubric for how the score is built.

The Ad Bench methodology: The checks, timing floors, and calibration bands in this article reflect The Ad Bench's current long-form scoring rubric and creative-review approach. They are designed to identify creative risk before media spend, not predict or guarantee campaign performance.

1. The Ad Bench methodology, long-form / connected-TV format rubric. Reviewed August 24, 2026.