Hold rate calculator

Views held is the completion-style figure your report offers: a completion, a fifteen-second view, a thruplay, a watch to seventy-five or a hundred per cent. Any of them works and none of them is comparable to the others, so pick one, write down which, and hold it.

Three-second views is the short-view count, from the same report and the same rows.

Impressions is the number of times the video was served. All three have to describe the same dates and the same placements, and the third output will say so when they do not.

Against impressions, the hold rate blends two effects: how many people started and how many of those stayed. A creative that nobody starts and a creative that everybody abandons can produce identical figures.

Against the hook, it is retention alone — of the people the opening stopped, what share reached the end. That separation is the whole reason this page exists, because the two failures need opposite work.

At 12 per cent hooked and 25 per cent retention, three impressions in a hundred produce a held view. Lifting retention to 40 per cent takes that to nearly five without a single extra impression; lifting the hook rate to 20 per cent takes it to five as well, and costs a different kind of work. Having both numbers makes that a choice rather than a guess.

The count between the two is where the money went. Seventy-two thousand views that started and did not stay is a specific, legible number, and it is usually the largest single loss in a video campaign.

It is worth looking at what that count implies about the edit. A body that repeats the opening, a demonstration that starts thirty seconds in, a slow build toward a reveal, a testimonial with no stated subject: each of these reliably produces a steep drop immediately after the hook, because the promise of the first second is not being paid off.

Where the platform offers a retention curve rather than two counts, read the curve. It puts the drop at a timestamp, and a timestamp is an edit instruction.

The count is also the honest way to size the opportunity before anybody starts editing. Recovering a fifth of seventy-two thousand abandoned views is fourteen thousand more people reaching the end, which can be compared directly against what those extra held views are worth and against the cost of re-cutting the video. Stated as a movement from twenty-five per cent to thirty, the same change sounds like a rounding error and gets deprioritised.

Whether people were persuaded. Attention is not intent. Creatives that hold beautifully and convert poorly are common, usually because the edit is interesting and the offer never arrives with enough clarity to act on.

Anything comparable between definitions. A hundred-per-cent completion rate and a fifteen-second thruplay rate are different measurements. Reports that quietly change which one they show produce performance stories that never happened.

Anything comparable between lengths. A completion on a fifteen-second cut and a completion on a ninety-second one are not the same commitment, and threshold-based measures can favour whichever length the platform's definition suits.

Whether the audience is right. A poor retention figure across every creative usually means the campaign is being shown to people with no interest in the subject, which no edit fixes.

Pay off the hook immediately. If the opening promises a result, show the result early rather than saving it. The structure that works on a feed is almost the reverse of the one that works in a cinema.

Cut the middle. Most underperforming edits are a strong opening, a long explanation and a good ending that almost nobody sees. Removing the explanation usually raises retention and the absolute count of people reaching the end at the same time.

Keep something changing. Static frames, long single shots and a talking head with no cut are where feed audiences leave. Motion, cuts and on-screen text buy attention cheaply.

Test length deliberately, and judge it on both rates. A shorter cut nearly always raises retention; whether it raises the number of people who reach the end is the question worth answering, and it is answered by the counts rather than the percentages.

And where the hold rate is already high, stop editing. A retention figure above about three-fifths means the constraint on the creative is how many people start it, and the work belongs in the opening, the placement and the audience instead.

The four combinations are worth naming, because each one points somewhere different.

Hook low, retention high. A good video almost nobody starts. The work is entirely in the opening, and the rest of the edit is already proven on the people who reach it.

Hook high, retention low. An opening that recruits people the body cannot keep. Sometimes the opening is overpromising, sometimes the body is slow, and the retention curve says which.

Both low. Usually the audience rather than the creative, especially if it holds across several very different executions. Test one known creative against a second audience before re-cutting anything.

Both high, outcome poor. The most frustrating case and the most common in practice: an ad people watch and nobody acts on. The offer, the call to action and the landing page are where that is decided, and no amount of editing reaches them.