Argument · Time-to-Winner

Signal speed beats scoring: why the fastest creative team wins.

By Orcool · July 16, 2026 · 6 min read

Why scoring feels like the answer

Creative scoring had a good year. Score every ad before it runs, rank them, put budget behind the top of the list. It feels like control, and next to pure guesswork it is a real upgrade. But a score answers one question and one only: of the ads in front of me, which looks strongest right now. It is a snapshot of a single creative, and a snapshot cannot tell you the one thing that decides whether spend pays off, which is timing.

What a score cannot see: the closing window

Every creative angle has a window. It opens when a hook is fresh and closes when the feed has seen it too many times. Scoring reads the pixels. It does not read the window. Two ads can carry the same score and have very different amounts of life left, because lifespan is a property of the angle and the market, not the polish on the frame.

We saw this directly when we pulled a panel of 1,061 live Meta ads and split them by format. Static image ran a median of 101 days still live. Video ran 66. Dynamic and DCO creative, which made up 57% of the panel, ran 33. Same production quality, very different clocks. A score taken on day one would have rated some of the shortest-lived work highly and told you nothing about how little runway it had left.

We wrote up the full durability breakdown in the 6% problem, and why ranking the library you already have is not the same as knowing which angle is missing in selection is not creation.

see the signal a shift in the feed  →  react days, not weeks  →  a winner sooner time-to-winner drops

The metric that survives: Time-to-Winner

The number that holds up under all of this is not a score on a creative. It is the count of days from starting a cycle to a concept that clears your KPI gate and keeps holding in rotation. We call it Time-to-Winner, and it is the number we actually sell against. Speed on its own can be gamed by shipping safe, forgettable ads that scrape past a soft bar, so it only counts paired with durability: the winner has to keep working, not clear the gate once and fade.

Under that metric, the fastest team is not the one with the sharpest scoring model. It is the one that turns a signal into a live test before the window closes. A faster ranking of stale angles still arrives late. Seeing the shift early, and reacting in days rather than weeks, is what actually moves the number.

Why a loop beats a score

Here is the part scoring cannot copy. A score is a fixed model applied to your ad. A loop is a model that learns your specific taste every time your team says yes or no. Each pick and each skip from the people who own the brand is a label. Dozens a week, no media budget required to start. Over a few cycles that feedback narrows what the system proposes toward the angles your decision-makers actually approve, so cycles-to-winner falls. Then the market outcomes, click-through and hold, calibrate the same model and catch its blind spots.

A static scorer trained on everyone's ads never learns the one thing that matters most in your account: what your own creative director says no to. That signal, the taste of the person who owns the brand, is the most valuable and the hardest to fake, and it is exactly what a fixed score throws away.

A score tells you which of today's ads is best. A loop tells you how to reach next week's winner faster. Only one of those compounds.

What this means if you run creative

Use scoring as a filter, not a strategy. Ranking a set of finished ads is fine for triage, and it beats sorting by gut. It will not tell you when to move or what to make next, so do not let it stand in for a plan.

Measure the clock, not the artifact. Track how many days and how many cycles it takes to reach a winner that holds. That number, not an average creative score, is what getting better looks like from one month to the next.

Shorten the loop between signal and test. The edge is not a nicer ranking of what you already have. It is spotting a shift early and reacting in days, so your budget lands inside the window instead of after it. Speed to the test, then durability of the winner, is the whole game.

Frequently asked questions

Methodology note: the format durability figures come from a single internal panel of 1,061 live Meta ads pulled from the public ad library and split by format, using median days still running as a lifespan proxy. One panel is a signal, not a universal law, and any client work referenced here is anonymized. The argument is about method; the numbers are included so you can weigh it for yourself.