Chapter 38: Minute Six Drop
At six minutes, four hundred listeners left me in the middle of a sentence I had refused to cut.
On Wednesday, PageTurn projected the prototype's minute-retention graph across the audio review room. The test used an anonymised random sample.
The line held through Carys entering the archive.
It held through the scratched safe.
Then it dropped while the narrator explained three insurance columns.
Four hundred anonymised sample listeners vanished inside sixty seconds. The fall was too concentrated to dismiss as ordinary drift.
My first note blamed the voice. I wrote it before replaying the sentence without the graph.
The actor had slowed on the policy dates and sounded cautious around the figures. I had supplied three clauses with nowhere natural to breathe.
“Can we hear the same performance without that paragraph?” the producer asked.
She loaded the marked comparison already prepared by PageTurn. The control used the actor's performance from the surrounding scene.
The actor's pace elsewhere retained listeners.
The fall began on my first explanatory clause and ended when dialogue returned. Voice, volume and playback quality stayed constant.
The evidence sat in my text, not her throat. Blaming casting would protect my pride and waste the remaining pilot money.
I crossed out ACTOR FIT on my pad.
Beside it, I wrote AUTHOR EXPLANATION.
The correction felt deservedly public inside the review record. The actor could see that I had withdrawn the criticism.
I opened the sixty-second script segment.
Carys currently recited policy number, cancellation date and insured vessel as internal summary.
The information was necessary. Without it, Mae Bell's vessel could not connect the call to the unexplained payment.
The lecture was not. It asked a listener to hold three numbers without a human reason to care.
I placed her colleague at the open safe.
“That policy ended on the fourteenth,” the colleague said.
“Mae called on the sixteenth.”
“Then the harbour owed no rescue.”
Carys turned the register towards her.
“It paid one on the seventeenth.”
The same three figures became an accusation and a contradiction. Each number now forced the other character to answer.
I removed the narrator's explanation.
The revised segment ran fifty-eight seconds. Silence around the register supplied the two seconds I had previously filled with explanation.
The actor read it once without changing her established voice.
Her pace sharpened because the characters now had something to resist. I did not ask her to sound younger, warmer or more dramatic.
The producer approved a targeted re-record rather than replacing the performer. The existing session remained intact outside that minute.
Its invoice entered my existing £6,000 pilot-production ceiling.
The cost reduced what remained for later tests. The budget ledger showed the charge against Saltmere alone.
No extra budget appeared because the problem was mine. Oliver and Pippa's ceiling did not change with my rewrite.
PageTurn randomised the old and revised audio across a fresh sample.
The minute-six drop flattened in the new version. PageTurn retained both A/B files and the random-assignment record.
More listeners reached the insurance payment contradiction. The lift continued beyond minute seven rather than producing a single curious click.
The overall prototype completion rate rose.
I signed the text-change reason and re-record charge.
The actor's file retained her original credit and clean performance note. My correction could not quietly survive as criticism of her work.
My editorial log retained responsibility for the removed paragraph.
The result did not make Saltmere the stronger audio proposition in every measure.
PageTurn opened the equal-summary comparison after the test closed.
Larkspur's audio-first construction still generated substantially more qualified shares.
Oliver and Pippa had built material people wanted to send to someone else. The equal summary supported that strength without exposing their script.
My revision had stopped readers leaving a necessary minute.
Those were different achievements.
Sunday's budget formula would have to value them without pretending they were the same.
The share column stayed in their favour even as my completion line recovered.