Methodology

    How AdLens analyzes and scores video ads

    AdLens uses multimodal AI to examine a video’s visual, verbal, audio, and time-based creative signals. It turns those observations into six explainable scores, timed evidence, and practical recommendations—before media results are available.

    The short answer

    AdLens applies classes of multimodal models to the creative itself. The system observes what appears and happens, what is said or written, how the story develops, and how brand and product signals work together.

    Scores summarize the strength of observable creative signals. They are decision support, not a promise of sales, lift, recall, or platform delivery. The report keeps the model rationale and video evidence visible so teams can review the reasoning rather than relying on a number alone.

    Transparent where it matters

    Every score is connected to a metric definition, model rationale, and observable evidence from the creative, so teams can review how the result was reached.
    Inputs

    What the models examine

    The analysis combines signals across the whole creative. No single frame, word, logo, or metadata field determines a score by itself.

    Visual and temporal signals

    Frames, scenes, pacing, composition, product use, logos, on-screen actions, and how these signals change over time.

    Language and on-screen text

    Spoken claims, captions, supers, calls to action, narrative structure, and consistency between words and visuals.

    Audio and emotional direction

    Speech, music, energy, tonal shifts, and the role sound plays in attention, meaning, and brand encoding.

    Creative context

    The supplied category, language, audience, and campaign context can help interpretation when available. Not every field is required or equally complete.

    How to read a score

    All metric and overall scores use a 0–100 scale. The interpretation band provides a quick signal; the rationale and creative evidence explain the result.

    0–65

    Needs work

    Important signals are missing, inconsistent, delayed, or difficult to follow.

    66–85

    Effective

    The creative communicates well, with identifiable opportunities to make it stronger.

    86–100

    Strong

    The relevant signals are clear, consistent, and well integrated across the video.

    These bands support interpretation; they are not universal pass/fail rules. Review the evidence, objective, audience, and benchmark context together.

    Metric library

    Six views of creative effectiveness

    Explore full report examples

    Score versus benchmark

    A score describes the creative. A benchmark describes its position within an eligible comparison set. They answer different questions and are calculated separately.

    Read the benchmark methodology

    Review the evidence

    Each report connects scores to rationale, timestamps, attention changes, scene observations, detected brand cues, and recommended edits where the data is available.

    Open the interactive demo