Cutroom
Guides
Most creative problems are arithmetic problems wearing a disguise. These are the numbers, worked out properly.
Fourteen a month for one winner. *Six is a coin flip.*
Most teams ship four to six creatives a month. At the published win rate, six is 0.42 expected winners. It is also a two in three chance the month ends blank. This page does the division. How many creatives buy one winner. What a thin month costs you. How to raise the count without adding a shoot day. You will finish holding your own number and a way to hit it.
A win rate is a fraction, and *the bottom half is negotiable*
Published benchmarks put winners at 5 to 8 percent of creatives. That is one in thirteen to one in twenty. Your own number is decided by the denominator. Most teams pick a flattering one without noticing. This page covers the three denominators, a definition of a winner that survives a quarter, and the two levers that move the fraction. By the end you will know which of the two is available to you this month.
Four causes, one chart, *and only one needs a new video*
An ad that dies produces the same shape whatever killed it. Four causes make that shape. Saturation, a spent pool, an auction shift and a false positive. This page gives you the fingerprint of each. It takes twenty minutes and three lines from your own account. By the end you will know which one you have, and whether a new video is the answer or the expensive mistake.
Five supplies of first-person video, *four already in the building*
A creator sells you three properties. A first-person voice, a real room, and specifics only somebody who has used the product can supply. Four sources inside your own business hand you all three for nothing. This page prices all five against each other. It also covers the one step that turns any of them into a finished vertical ad. You will finish knowing which source to use this week.
The invoice is about half of it. *The rest never gets billed.*
Editors price three ways. Hourly, per video, or on a monthly retainer. Each shape moves the risk somewhere else. One clause decides your bill more than the rate does. This page unbundles all three. Then it adds the three costs that never reach an invoice. You will finish holding a real number per finished ad, and knowing which half of the work to stop paying craft rates for.
Forty minutes of clock, *four minutes of decisions*
One thirty-second vertical ad, cut from a three-minute take, runs about forty minutes of hands-on work. Two revision rounds turn that into an evening. This page audits where the forty minutes go, step by step. Four of them contain every decision that changes the result. By the end you will know which four, and what it takes to stop paying for the other thirty-six.
Five rules, and *the fifth decides the other four*
Most creative testing frameworks are four rules long. A falsifiable sentence per creative. Verdict conditions written before launch. Kill on cheap metrics, promote on expensive ones. Log every row, including the ties. Each of those is here with its thresholds written out. Then comes the fifth rule, which decides whether the other four were worth writing down.
Three seconds is *eight words*, and none of them sell
Three seconds of normal speech is about eight words. That is the whole hook. It is not enough room to describe a product, a brand or a benefit. Here are the six shapes that do fit in eight words, four weak openings rewritten line by line, and the rule that decides whether yours survives a muted feed. You will finish with six openings written rather than one polished.
Three to five per set, and *the division shows why*
Three to five active creatives per ad set is the working answer for most budgets. The check is one division. Daily spend over ads live. This page runs that division at three budget levels, shows what ten ads on fifty a day actually buys, and sets the cadence that stops the delivery system relearning every week. You will leave with your own number.
Fatigue is a curve, and *its half-life is a quota*
Creative fatigue is a decline caused by repeated exposure, with budget and targeting held roughly constant. Every clause in that sentence rules out a decline the word gets blamed for. Here is how to plot your own curve against creative age, read the half-life off it, and convert that half-life into a monthly quota. You will finish with a number instead of an argument.
One rate, *six line items*, five of them negotiable
A creator quotes you one number. Inside it sit six separable line items. Concepting, the shoot, deliverables, revisions, usage rights and whitelisting. This page unbundles all six, shows how to cut five of them without changing anything on screen, then multiplies the result by the published win rate. You will finish holding a cost per winner.
One setup, one afternoon, *twelve openings written the night before*
Setup and nerve cost the same whether you shoot one video or twelve. That is the whole case for batching. Here is the plan in order. Twelve openings written the night before, one lighting position you never move, every take under three minutes, and the cutting session booked before you start. One afternoon covers a month of ads.
A cutaway is *an answer*, not a texture
A cutaway earns its place when a sentence raised a question and the shot answers it. Not near the sentence. On the exact words that raised it. Here are the three moments you never cover, the coverage ceiling by pace, and a twelve-minute sourcing method that replaces forty minutes of browsing. You will know where every clip goes before you start looking for one.
Two to four words a card, *and check it on a bus*
Most of a vertical feed plays with the sound off. That makes captions road signage rather than typography. Two to four words a card. Heavy weight, real contrast, clear of the platform chrome. Emphasis rationed to two or three words an ad. Here are the five rules, the transcription check that saves a week, and the test you run outdoors before anything goes live.
The output is not badly written. *It is generically written.*
Generated ads underperform, and video quality has nothing to do with it. Three named failures do. The script is the middle of everything ever written about your category. Nobody is standing behind the claim. Within months the whole category converges on one look. This page covers each failure in turn. Then the ten-second test that catches the first one. Then the line between work worth automating and work worth keeping.
Same body, six openings, *and the media is the bill*
A hook test is six ads that are identical after the first three seconds. You eliminate on the three-second view rate. You confirm on cost per acquisition, never the other way round. This page covers the method. Then the arithmetic showing ninety percent of the budget is media rather than production. Then the four ways a hook test quietly becomes six files and no lesson. You will finish able to run one this week.
Thirty seconds is *seventy-five words*. Count them first.
A natural read is two and a half words a second. So a thirty-second ad is about seventy-five words. Most first drafts run a hundred and twenty. This page gives you the block budget, a worked script that fits it, and the two structural fixes that buy back five seconds. You will finish able to count a draft before anybody points a camera at it.
Live slots over healthy span, *over the share that survives*
The cadence is a formula, not a calendar. Live slots, divided by how long your creatives stay healthy, divided again by the share that lasts. This page walks each of those numbers, then the five cheap refreshes to exhaust before commissioning anything new. Four slots, a three-week span and one in three surviving comes out at fifteen a month. You will finish holding your own quota.
Monthly cost over price per video. *That is the debate.*
Break-even is your fully loaded monthly in-house cost divided by an agency's price per finished video. Six thousand against five hundred is twelve videos a month. Most teams cannot brief and review twelve. This page runs that division, prices the four costs it hides, and shows the split teams actually settle on. By the end the debate is two numbers rather than an argument about culture.
*'I stopped setting the six-forty alarm'* beats any production value
A viewer decides person or advertisement inside a second. Three signals settle it. A detail too boring to invent. Imperfections of delivery left in while genuine faults are fixed. A speaker who says in four words why they would know. This page works through all three, with the lines to keep and the lines to cut. Production value is not on the list, and by the end you will know where that budget should go instead.
One take, *six ads*, if you paused between the blocks
A three-minute recording contains six blocks. An opening, a problem, a mechanism, a proof, an objection answered and an ask. Say them as standalone thoughts with clean pauses and they recombine. This page covers how to record them, which variant layers to work down, and which variants are real. By the end you will get about six ads out of one sitting, with no second camera day.
Four components, *and the invoice is the smallest*
One tested creative has four components. Production, internal time, delay, and the media spent reaching a verdict. For most small advertisers the last one is the biggest by a distance. Only the first arrives as a bill anybody negotiates. This page adds every component, multiplies the total by the published win rate, and ranks the three moves that shrink it. You will finish holding a cost per winner.
Creative volume: *winners needed, times thirteen to twenty*
Most small accounts make six creatives a month and then wait. Six is not a number that finds anything. This page runs the formula, prices the answer in credits, then checks it against the spend floor underneath. By the end you will have a monthly creative target you can defend in a meeting.
Cost per tested creative: *production is the smaller half*
Almost every version of this calculation counts one term and stops. It counts what the ad cost to make. The second term is the media spend it takes to read the result. For most accounts that term is the larger one. This page runs both, prices production at about 113 credits, and shows you how to compute your own spend floor.
Creative budget planning: *start from what you can read*
Most creative budgets get set by whatever the last invoice happened to be. That is backwards. The number of creatives your media budget can read is the input. The production line is the output. This page walks the four steps in order, works three real splits, and names the lines people leave out of both.
Video ad length: *the runtime is a result, not a decision*
There is no correct length. There is only a correct script. Cut everything that is not the claim, the proof or the instruction. Then read the runtime off the export. This page says what each duration carries, what an extra twenty seconds costs, and how to make the cut without picking a number first.
Credits calculator: *the whole price list, then five workflows*
Every charge here is a fixed number and there are five of them. Nothing is metered by seat, by render tier or by how busy the service is. This page lists all five, then prices the five workflows people actually run, from a 20 credit music track to a 374 credit rendered presenter ad. By the end you will know what a month costs before you subscribe to anything.
Hook rate benchmarks: *the only one that applies is your own*
Somebody in a meeting quoted a hook rate from an article. It described a different placement mix under a different denominator. Yours is the only benchmark that applies to your account. This page gives the ten minute procedure for computing it, the sample size you need before acting on a gap, and the openings worth testing.
Creative refresh: *replace on decay, not on a calendar*
A fortnightly refresh rule replaces creatives that are still working. It also keeps ones that stopped weeks ago. There is no correct number of weeks. This page names the three signals that say a creative is finished, the queue depth that lets you act on them, and what a replacement costs.
Commissioned against in-house: *compare the revision policy first*
Commissioned creators give you something in-house production cannot. A different face, a different home, a stranger who is not you. What they cannot give you is the fifteenth attempt at the price of the first. This page compares the two on structure rather than sticker price, and prices the in-house column at about 113 credits an ad.
Testing budget split: *size it from what you can read*
There is no correct percentage. Every article quoting one is guessing at your cost per thousand impressions. The testing share is the spend it takes to read one creative, multiplied by the number you intend to read. This page works it backwards in three steps and shows what the winner rate does to the timeline.
Time per ad: *count the handoffs before the hours*
The working hours on an ad are small. The elapsed days are enormous. Every handoff introduces somebody else's calendar, and calendars are where the weeks go. This page counts the handoffs instead of the hours, prices the revision loop in days, and shows how one filming session removes three of them.
Yes, end to end. *Two of the four jobs are still yours*
A machine takes a recorded take and returns a finished vertical ad today. The assembly is the dependable part of that. Deciding what to say is not automated. Judging whether it worked is not either. This page splits the question into four separate jobs and prices the two that are now mechanical. By the end you will know which part of your week you can hand over this month.
No single number exists. *Build yours out of five lines*
There is no single price for a video ad. Self-shot costs your hours and almost no cash. Commissioned costs a negotiated fee per deliverable. Produced costs people-days, which is where the frightening figures come from. This page gives you the five-line stack that produces your own number, and the one figure to divide it by. By the end you will have a cost per tested ad you can take into a budget meeting.
B-roll swaps the picture, *never the sentence underneath*
B-roll is the footage cut in over your main shot while the speaker keeps talking. In an ad the main shot is a person to camera. The b-roll is everything else the viewer sees while that person carries on. This page covers where the name came from, the four jobs a cutaway does, and the coverage ceiling nobody says out loud. By the end you will know which two seconds of your own script deserve a picture.
A hook is a purchase. *It buys the next three seconds*
A hook is the opening moment of an ad taken whole. The first line, the first frame, the first sound, and any text on screen. Its one job is turning a scrolling viewer into a watching one. On vertical video it has about three seconds. This page covers the four channels firing in those seconds, how the result gets measured, and the four beliefs about openings that cost money. By the end you will know what to change first on your next ad.
9:16 at 1080 by 1920. *The safe zone is the hard part*
Vertical 9:16, 1080 wide by 1920 tall, MP4 with H.264 video and AAC audio. Applying that takes ten seconds. What costs money is the interface the app draws on top of your frame. A logo in the corner ends up behind a button. This page covers the screen area each shape gives away, where the safe zone sits, and how to rescue horizontal footage. By the end you will know exactly which pixels are yours.
Twenty to thirty seconds, *and here is how to argue yours*
Twenty to thirty seconds for most direct response. Anything outside fifteen to sixty is a decision you should be able to defend out loud. The number sits where two curves cross. Viewers leave continuously, and a viewer has to understand the offer before they can act. This page covers where your own two curves meet, the four questions that set the number, and how to test a pair. By the end you will have a length for your account rather than for somebody else's.
The mechanism holds. *The percentage you read is folklore*
Almost certainly yes, and the mechanism is not arguable. A large share of feed video is watched with the sound off. An uncaptioned talking-head ad says nothing at all to those viewers. What does not exist is a credible universal lift figure. This page covers where the famous statistic came from, the four jobs a caption does, and which caption decision is worth testing. By the end you will know what to stop arguing about.
No user made it. *The format is the entire product*
UGC advertising means ads built to look like an ordinary person's own post. One person, a phone, a real room, speaking to camera. Almost none of it is user generated in any literal sense. The people are briefed, commissioned and paid. This page covers the three signals a viewer reads first, the four routes to getting the footage, and the four misreadings that cost money. By the end you will know which route gets you enough attempts.
One variable, enough volume, *and a rule written before launch*
Change one variable at a time. Give each version enough impressions to mean something. Write the verdict rule down before anything goes live. Almost every failed creative test breaks one of those three, usually the third. This page covers what to test first, how much volume a result needs, and why your cadence is capped by supply. By the end you will know how many creatives your month has to produce.
The ad did not get worse. *The audience ran out of new people*
Creative fatigue is the decline that happens when the same creative keeps reaching the same people. Nothing about the ad changed. The pool of viewers seeing it for the first time is shrinking, and repeat viewers respond less. This page covers the two numbers that diagnose it, the four causes that look identical to it, and the cheapest replacement available. By the end you will know whether to commission anything at all.
Fourteen tested creatives per winner. *Now count backwards*
Work back from the win rate rather than guessing at a number. Published benchmarks put winners at roughly 5 to 8 percent of creatives. Expect about fourteen tested creatives per winner. This page covers the three numbers that size your month, what actually counts as a separate creative, and what one attempt has to cost. By the end you will have a monthly target you can defend in a meeting.
The offer does most of it. *Production value does least*
The offer decides most of it. Then how specific the claim is. Then whether the right viewer reaches the ask. Production quality comes fourth at best, and in a feed a polished look often costs you performance. This page ranks the four levers, shows what a checkable claim looks like, and gives the coverage ceiling that keeps a face in the ad. By the end you will know where the next hour belongs.
Automate the transformation. *Never automate the invention*
Use it for assembly. Transcription, cutting, captions, footage search, export. Do not use it to decide what the ad says or whether it worked. One rule divides the two. Does the step transform material you supplied, or invent material you did not? This page applies that rule to the features on the market, prices the safe half, and names the two purchases to refuse. By the end you will know which demo to walk away from.
Six ads from one take, *and none of them is a slice*
Vary the opening, the length and the pace against one recorded take. Do not slice the same file into pieces. One three minute take carries several genuinely different ads, because what changes between them is the entrance rather than the argument. This page covers why slicing fails, how to record six entrances in one sitting, and what each further version costs. By the end you will know how to get six ads out of twenty minutes.
One person, chest up, *and no scene to build*
A talking-head video is one person speaking to camera, framed from the chest up, with no constructed scene around them. In advertising it is the dominant vertical format. A face saying something specific is the cheapest credible thing anybody can film. This page covers where the name came from, the four production decisions that carry it, and the coverage ceiling that protects it. By the end you will know how to get several ads out of one recording.
The transcript is the timeline. *The words carry the timecode*
You edit the transcript instead of the track. Every word carries the timing of the audio it came from. Delete a line and that audio goes, and the picture rebuilds around the gap. Highlight a phrase and footage lands over exactly those words. This page covers the timing data that makes it work, the trade it asks for, and the four jobs it hands back to a timeline. By the end you will know whether to point at words or at clips.
An EDL is a list of decisions, *not a video file*
An edit decision list is a plain, machine-readable record of which piece of source plays when, in order. It contains no picture and no sound. It was invented so an edit made on cheap equipment could be recreated exactly on expensive equipment. This page covers where it came from, the four facts inside one edit event, and why every automated editor still produces one. By the end you will know which files in your archive are worth keeping.
Nothing broke. *The people who had not seen it ran out*
Usually because the audience already saw it, not because the ad got worse. Four causes account for nearly all of it. Frequency. A change you made. Auction and seasonal shifts. Measurement drift. This page covers how to tell them apart, the order to check in, and what a replacement costs on the one occasion it is fatigue. By the end you will know whether to commission anything at all.
Not on a calendar. Swap it *when frequency and cost climb together*
Change it when frequency climbs and cost per result climbs with it. Not on a schedule somebody wrote in an article. For most small accounts that lands between two and six weeks per creative. This page covers the two numbers that trigger a swap, why your cadence is set by supply, and what a replacement costs. By the end you will know how often your own account can actually refresh.
Early views over impressions. *That is the whole formula*
Hook rate is the share of impressions that reach a short early threshold, commonly three seconds. It measures the opening alone rather than the ad. That is what makes it useful, and also what makes it trivially easy to inflate. This page covers the formula, why cross-platform comparisons are meaningless, and how to build a baseline from your own ads. By the end you will know what number your next opening has to beat.
Not for the testing. *For four shots a phone cannot get*
Not for volume testing. A phone, a window and a willing person outperform a booked shoot there. Yes for four specific cases. A second location. Controlled lighting. A demonstration nobody has filmed. A hero film you will run for a year. This page covers what each route is good at, the arithmetic that decides it, and how to brief a shoot when you do book one. By the end you will know which half of your month needs a camera crew.
Native is grammar, *not a filter you add at the end*
Match the grammar of the feed rather than the look of a commercial. Vertical, one person speaking, captions on, the first frame already mid-thought, no title card, no logo sting. Native is a set of decisions made before you record, not a treatment applied afterwards. This page covers the seven signals a viewer reads first, the three endings that undo all of them, and the coverage ceiling. By the end you will know what to change on your next recording.
Arcads bills per render, *so your keep rate is the price*
The billable unit in generated UGC is one rendered video. It bills whether you publish it or delete it. So the bill is not the ads you run. It is that number divided by your keep rate. This page covers the meter, the three costs it hides, and the other meter that bills a filmed take instead. By the end you will be able to size a tier before you pay for one.
Avatar platforms bill minutes, *and a retake bills again*
The meter counts minutes of video generated, not minutes you keep. A ninety second ad regenerated twice bills four and a half minutes to deliver ninety seconds. This page covers how that meter behaves, the habit that halves it, and what the same thirty videos cost on a meter that bills a filmed take instead. By the end you will take a multiplied number to their live tiers rather than a hopeful one.
A credit is not a price *until you find the conversion*
The meter here is credits. A credit means nothing until the vendor states what it buys: a generated video, a second, a minute, an avatar pass or a voice pass. Two tools can both sell a thousand a month and differ tenfold in what arrives. This page covers how to find that conversion, why a credit meter counts attempts, and what thirty ads cost on a table published in full. By the end you will be able to line two plans up in the same column.
Caption tools bill per clip, *so count last month first*
The meter is how many clips you process in a month, with certain features gated by tier rather than by volume. The number you need is not your plan for next quarter. It is what you actually posted in the last thirty days, and that is almost always smaller. This page covers the meter, the agency case that breaks it, and what the same stage costs when the captions arrive inside the cut. By the end you will know which tier to buy and which number forced it.
Two meters at once: *a seat each, then hours of speech*
Descript charges per person and caps transcription hours per month. Two different constraints can bind first, and most people arrive with only one number. This page covers both meters, the recordings that quietly double one of them, and what thirty short ads cost on a meter with no seat charge at all. By the end you will know which count decides your tier.
Clipping tools bill the upload, *so yield sets your cost*
The meter counts what you feed in. A three hour stream consumes three hours of allowance whether it hands back twenty clips you would post or two. So yield decides your real cost per clip, and only your own footage knows yours. This page covers the meter, the three costs it does not price, and what a purpose-shot take costs instead. By the end you will be able to turn any tier into a cost per published clip.
Public tiers, then a quote: *localisation decides which*
The public plans meter video minutes and editor seats. The top tier is a quote rather than a number. Nine languages means nine times the minutes for roughly the same effort, so localisation is what moves a team off the published plans and onto a phone call. This page covers both meters, how to get two quotes you can actually compare, and what a short vertical ad costs on a table with no quote in it. By the end you will know which meter binds you and what to negotiate on.
Worth it when the ad informs. *Not when it must be believed*
Arcads earns its money when your ad explains, demonstrates or lists, and nobody in it has to be trusted personally. It stops earning when the ad is testimony, because a generated performer saying a product changed their life is the one register this category still loses. This page covers where that line falls, what the render meter really costs behind thirty published ads, and what those thirty cost when a real person films. By the end you will know which pile your winners sit in.
Worth it when a stranger *could deliver the same script*
Read your script out loud. If a stranger reading it sounds completely normal, HeyGen is a sensible way to make that video and the subscription pays. If it would be odd coming from somebody who never used the product, an avatar cannot fix it and the money is wasted. This page covers the test, the three cases where it clearly pays, and what the same videos cost when a real person films. By the end you will know whether to buy minutes or to buy the edit.
Worth it if captions are slow. *Usually they are not*
Submagic is worth the money when hand-captioning is the minute that costs you most, which is true for accounts already posting tightly written clips. If your clips take eight seconds to reach the point, captions are the cheap part and you are about to make the easy part easier. This page covers the audit that tells you which, the meter you would buy on, and what the job costs when captions arrive inside the cut instead. By the end you will know which stage of your week to spend on.
Worth it if you already record. *Upload one hour and count*
Opus Clip is worth it when long recordings pile up as a by-product of something you already do. You do not have to guess: upload one real hour, count the clips you would genuinely publish, and divide. This page covers the prerequisite, what yield does to the real price, and what a purpose-shot take costs when the archive is not there. By the end you will know whether to mine or to shoot.
Free below ten ads a month. *Then it bills you evenings*
CapCut is worth using for almost anybody making a handful of ads a month, and it costs nothing in money. It costs about forty minutes per thirty-second ad, counting the fiddling and the export. That is under seven hours at ten ads a month, and twenty hours at thirty. This page counts those hours at your own volume, names the cases where the free tool still wins, and prices what removing the timeline buys instead. By the end you will know which of the two your month can afford.
Worth it above four hours of speech. *Below two, rarely opened*
Descript earns its price when you record more than about four hours of speech a month, and it is hard to argue against past ten. Below two hours you are buying a capability you rarely open. This page covers the count that decides it, the two meters behind the tier, and what thirty short ads cost on a meter that bills nobody per person. By the end you will know whether your material is long-form speech or short ads.
Worth it when a video *has to exist more than once*
Synthesia earns its price when the same content must exist in several languages, several regions or several role-specific versions, and get reissued when the product changes. One video, made once, in one language is the case where the platform cost is hardest to justify. This page covers the multiplier that is the whole business case, where the output stops working, and what a short vertical ad costs on a table with no quote in it. By the end you will know whether your video has a multiplier in it.
CapCut makes the first ad. *Cutroom makes the ninth*
You sell a product and you cut the ads yourself. CapCut is where you do it. The first few go fine. Then the catalogue grows, the winner fatigues, and every new ad starts from an empty timeline. This page covers what breaks at twenty ads a month, what twenty consistent ads cost in credits, and how to shoot a catalogue in one morning. By the end you will know what your ninth ad has to cost for the month to work.
Descript finishes the episode. *Cutroom finishes the ad*
You are the brand. Your face is what closes. Descript already runs the podcast, the demo and the long recordings, and it is good at those. The thirty second vertical ad at the top of the funnel is a different build with different rules. This page covers what that build needs, where forty minutes an ad goes, and what six founder-led ads cost in credits. By the end you will know which part of your Sunday you can hand over.
HeyGen teaches the lesson. *Cutroom makes the ad that sells it*
You built the course and a generated presenter delivers the modules. One face, one voice, forty lessons, no reshoot when a price changes. That is the right architecture for a curriculum. The ad that persuades somebody to buy the course is a different format. This page covers what that ad needs, why a lesson clip run as an ad underperforms, and what eight launch ads cost in credits. By the end you will know what to record before your next launch.
A UGC ad is an ad *shaped like somebody's own post*
A UGC ad is a paid advert made to look like an ordinary person's own post. One person, a phone, a real room, talking to camera about something they use. The letters stand for user generated content, and the name is misleading. The person is almost always briefed and paid. This page covers the plain definition, one real ad broken down second by second, and what making one costs today. By the end you will know which choices to make before you press record.
B-roll in an ad is *the picture that proves the sentence*
B-roll is footage cut in over the main shot while the speaker carries on talking underneath. In a vertical ad that means the product, the hands, the result, anything other than the face. It lands over the exact words that need it. The audio never breaks, which is what separates it from a scene. This page covers the definition in plain words, one thirty second ad broken down clip by clip, and the coverage ceiling. By the end you will know which two seconds of your own script deserve a picture.
Creative testing is *running ads to find the one that works*
Creative testing means running several versions of an ad against each other so the account tells you which one works. The winner gets the budget. The rest get switched off. This page covers the definition in plain words, one test written out with real numbers, and the arithmetic that decides whether your testing is worth doing at all. By the end you will know how many ads your month has to produce.
Transcript-based editing means *you edit the words, not the clips*
Transcript-based editing means the software turns your audio into text and remembers which second every word came from. You change the video by changing the text. Delete a line and that audio goes with it. Highlight a phrase and footage lands over exactly those words. This page covers the timing data underneath, one paragraph edited three ways, and the point where a text editor becomes a finished cut. By the end you will know whether pointing at words suits how you think.
One take in. A finished ad out. See it on your own footage before you decide anything.