Ephoria

Async creative review: every question your note leaves open costs a day

Nobody is awake to explain what you meant. Here is the review that closes in a single handoff: one owner, one consolidated list, a timecode, a priority and an acceptance test on every line, filed before you close the laptop.

A housekeeper in a black vest and white gloves setting a pillow on a made bed
What is in here
  1. What does an extra hour of time-zone distance cost?
  2. Why does one clarifying question cost a whole day?
  3. What a note has to carry before it is worth sending
  4. One owner, one list, one delivery
  5. Which review tool should you use?
  6. The deadline is the end of your working day, in every US zone
The short answer

One owner, one consolidated list, one delivery. Where the two working days do not overlap, anything you leave ambiguous comes back as a question rather than a fix, so the note has to close every question before it is sent: a timecode on every line, a priority on every line, a reason on every line, and an acceptance test on every defect. Then file the whole list before you close your laptop.

What you get out of this
  1. The one study that varies time-zone distance rather than remote work, its sample, and the limit it cannot reach
  2. The mechanism behind the delay, from the coordination-cost literature rather than from a vendor blog
  3. What the four review tools do mechanically, quoted from their own documentation, including the one that hides comments from other reviewers
  4. A note template you can paste, and the checklist that decides whether a list is ready to send

It is 5.40pm and you have a cut open in one tab and a comment box in the other. The studio has been asleep for six hours. Whatever you write before you close the laptop is the whole conversation until tomorrow, and anything ambiguous comes back as a question rather than a fix. This page is about writing the note so those hours get spent once. Having taste about the cut is a separate skill.

The evidence, with its samples attached

Papers read as PDFs, September 3, 2026
11.0%drop in synchronous communication from one extra hour of time-zone separation, across 12,038 employees in 48 countriesChauvin, Choudhury and Fang, Organization Science, 2024
5.7hmean business-hours overlap in that sample. Nobody has measured a pairing at zeroSame study
0.743the p value on the same study's asynchronous coefficient, so async volume did not rise to replace what was lostSame study
12.7 daysa distributed work item, against about five days on one site, across 2,227 modification requestsHerbsleb and Mockus, IEEE TSE, 2003
24hthe response window Doist publishes for itself, and the only number in either of the remote handbooks I quote hereDoist
Four of these come from papers I opened as PDFs rather than from anybody's summary of them. The last is a company handbook. Every one is about software engineering or general knowledge work, because I found no study with a sample measuring creative or advertising production across time zones. The evidence here is borrowed and has to be handled as borrowed.

01What does an extra hour of time-zone distance cost?

About eleven percent of the synchronous communication. Chauvin, Choudhury and Fang measured it in Organization Science in 2024, across 12,038 employees of one multinational in 167 cities and 48 countries, using cities' daylight saving transitions as the shock. Their sentence: "an approximately one-hour increase in temporal distance reduces synchronous communication volumes by an average of 11.0%", with a 95 percent confidence interval of 6.2 to 15.5 percent.

The limit that study cannot reach

Two things in that paper matter more to a US brand than the headline does. The mean overlap in their sample was 5.7 hours and the average treatment removed 1.1 of them, so it measures shaving an hour off a relationship that had nearly six. A US brand working with an Indian studio has none. Asynchronous volume also did not rise to fill the gap: the coefficient was "negative but small and not statistically significant", at -0.007 with a p value of 0.743.

Four things this research gets quoted as saying

Flip them
Each card's back is what the paper reports about its own sample. I am putting the strongest published argument against this arrangement on the page rather than around it, because a buyer who finds it later will trust everything else here less.

02Why does one clarifying question cost a whole day?

Because the answer can only arrive on the next handoff, and the next handoff is twenty-four hours away. Espinosa and Carmel model coordination cost as four things rather than one: "communication, clarification, delay, and rework". Clarification is the one that hurts at distance. Their finding is that "when team members are separated by time, the need to clarify messages will introduce further delay, unless this happens during work-overlapping hours". Where the overlapping hours are zero, every question is a day.

What one ambiguous line costs

Follow it across the night
01You file the notelate afternoon yourtime, hours beforethe studio's daybegins02The studio opensthe listreads it cold, withnobody awake to askEVERY QUESTION STOPS HERE03One line readstwo waysthey either guess andrisk the round, orask and wait04The answerreaches themyour next morning, afull day after thenote went out05The fix lands theday afterand the round youthought was one hasquietly become two
01You file the notelate afternoon your time, hours before the studio's day begins
02The studio opens the listreads it cold, with nobody awake to askEvery question stops here
03One line reads two waysthey either guess and risk the round, or ask and wait
04The answer reaches themyour next morning, a full day after the note went out
05The fix lands the day afterand the round you thought was one has quietly become two
Nothing in this chain is anybody behaving badly. It is one sentence that could be read two ways, meeting a working day that has already ended. The only place to intervene is the first box.

Longer notes do not fix this. Espinosa and Carmel's catalog of what distributed teams do includes a tactic they call "break the e-mail chain", and their description of it is worth having in front of you: the chain "begins when one actor initiates a message, the receiver does not understand it fully and asks for clarification, the sender attempts to clarify, the receiver misinterprets again, and so on. Meanwhile, an entire week has gone by."

A note that needs one clarifying question costs a day. A note that needs two costs two.

Espinosa and Carmel's clarification cost, on a 24-hour handoff

03What a note has to carry before it is worth sending

Three fields per line, and the strongest published guidance on this comes from usability rather than film. Nielsen Norman Group's page on expert reviews says that "Including a severity rating for each issue discovered is key to making the findings actionable", asks for findings "mapped to where they occur in the design", and insists on "objective explanations, not subjective criticisms". Severity, location, reason. On a video, the timecode is the location field.

The note template, one line per issue

Copy it
ONE LIST. ONE OWNER. NOTHING ELSE IN THE THREAD.

[00:04.2]   P1  Product name unreadable at feed size.
            WHY     white type sits on the window highlight, no plate
            ACCEPT  legible on a phone, muted, at arm's length

[00:07.0 - 00:09.5]   P2  Beat runs long against the voiceover.
            WHY     the line ends at 8.1, the shot holds to 9.5
            ACCEPT  the cut lands on the last word

[00:12.8]   P3  I would try the wider frame here.
            TASTE   not a defect. Decline it if it costs a re-roll.

OWNER   one name. Nobody else replies in this thread.
ROUND   1 of 2.
FILED   17:40 local, before end of day. This is the whole list.
The ACCEPT line is the field most lists leave out and the one that saves the day. Priorities are yours to set. Whatever you call them, define them at the top of the list, because a studio left to guess your ranking will rank by whatever it opened first.

Most lists carry three fields and stop. The fourth is what would make the line right. A note saying a caption is unreadable can be answered six ways; a note saying it has to be legible on a phone at arm's length with the sound off can be answered once, without a conversation. Naming a defect when you are not a video person is the sister site's subject, in how to review AI creative when you are not a video person.

A tattooed hand sprinkling green herbs over a plated dish on a green ceramic plate, charred seafood and shaved white onion over an orange sauce, the kitchen behind it thrown out of focus
Hotel client - campaign reel. Half a second either side of this frame, the shot does not look like this. "The moment the herbs land" is a search somebody has to run; [00:03.4] is an instruction they can act on.

Is that a defect, or a taste note?

Answer three questions
Run it on the line you are least sure about. The point is to work out which priority the note belongs at, not to talk you out of it, because a taste note filed as a defect costs a re-roll and a defect filed as taste ships.

04One owner, one list, one delivery

A round is one consolidated list from one person, answered in one delivery. Anything looser is a conversation, and a conversation across a gap with no overlap is a week. Three people commenting on one cut over an afternoon produces contradictions somebody has to resolve without being able to ask which of you wins, so they guess. What a round contains contractually, and what a third should cost, is in what counts as a revision round.

Running one review, start to filed

One screen at a time
Step 1
Watch it cold, once, at phone size, before you open anybody's comments

Full screen on a laptop is not the viewing condition your ad will meet. Watch it once on a phone, muted, at the size it will run, and write down only what you noticed without trying to name it.

This pass is for defects that announce themselves. Everything you catch here is worth more than everything you catch on the fourth watch, because by then you know what was intended.

Step 2
Collect from everybody in one place, then edit the list yourself

Let the team comment wherever they like, then take the whole set and rewrite it as one list. Delete duplicates. Resolve contradictions before sending, because the studio cannot resolve them and will not know which of you outranks the other.

This is the step that makes you the owner. GitLab's handbook puts the standard for a useful reply plainly: "Answers like: 'will do', 'OK', 'it is on my todo list' are not helpful."

Step 3
Rank every line, and say what the ranks mean

Use any three-level scheme you like and define it at the top of the list. One that works: P1 is a defect that stops the ad shipping, P2 is a fix worth making if it is cheap, P3 is taste the studio is free to decline.

A list where everything is urgent tells the studio nothing and gets prioritized by whoever is holding the file at 9am. That person is not you and they cannot ask.

Step 4
Add an acceptance line to every P1

Write the condition that would let somebody else mark it done without checking with you. Legible on a phone. Cut lands on the last word. Product visible before the second second.

If you cannot write the acceptance line, the note is a feeling rather than a defect, and it belongs at P3 where declining it costs nothing.

Step 5
File it before you close the laptop, and stop adding

The deadline is the end of your working day, in every US time zone and both seasons, because the studio's next day starts after yours ends.

Anything you think of at 9pm belongs to the next round. Adding to a list after the handoff is how one round quietly becomes three.

1 / 5

About forty minutes for a fifteen-second cut, most of it in steps two and four. Step one is the cheapest and the one people skip, because watching cold is the only chance you get to see the ad the way a stranger will.

The same note, two ways, on a twenty-four hour handoff

Watch it run
An illustration of the mechanic rather than a transcript of a real client thread. The difference between the two runs is four extra words in the first message and a habit of writing them.

05Which review tool should you use?

Any of the three that let a reviewer without an account pin a comment to an exact frame. Frame.io, Vimeo Review and Dropbox Replay all do that, from their own documentation. Frame.io and Dropbox Replay also support range comments and drawing on the frame. Only Dropbox Replay's docs describe filtering comments by version. Loom is a different instrument and should not be where you collect notes.

Four tools, quoted from their own documentation

Read on one day
Four tools, quoted from their own documentation
DimensionWhat its own documentation says it doesUsable by a reviewer with no accountWhere it stops
Frame.ioComments post as a card and a bubble marking the timecode position; ranges are capped with the I and O keys; an annotation tool draws on the frameYes. YesIts own doc says the timecode is what the editor needs to know where to make edits. This whole page is downstream of that sentence
Vimeo ReviewReview links carry time-coded comments on the timeline, clickable to jump, plus a permission that lets a reviewer mark a video ApprovedYes. Yes, no account or team seat neededThe help article never says how a comment acquires its timestamp, and I am not going to guess on its behalf
Dropbox ReplayComments land on the timeline at the frame you picked; ranges by holding shift and dragging; drawing supported; comments filterable by versionYes. Yes, and quick reactions need no typing at allThe only one of the four whose documentation describes filtering comments by version
LoomReactions and comments appear on the video timeline as you leave themNo. No. Its doc says viewers not signed in cannot see comments or reactionsStrong for sending a spoken walkthrough, weak as the place a team's notes get collected
Read on September 3, 2026 from each vendor's own help pages rather than from a comparison post. The two features that matter most across a twenty-four hour handoff are ranges and version filtering, because between them they remove the two questions that would otherwise cost you a day: where exactly, and which cut.
What I could not find, and will not implyNo published study with a stated sample measures whether timestamped, prioritized, consolidated feedback reduces revision rounds. I looked. Everything above about how to write the note is craft consensus plus the mechanics each tool documents, and it should be read that way. The time-zone numbers are evidence. The template is a practice.

The vocabulary of a review that has to survive a night

Search it
7 terms
HandoffClock
The moment work passes from one side to the other. Across a US and India pairing there is one a day in each direction, and anything that misses one waits twenty-four hours for the next.
RoundProcess
One consolidated list from one owner, answered in one delivery. Two people commenting in parallel over an afternoon is a conversation rather than a round, and resolving it eats the round you thought you were already in.
Acceptance testNote
The condition that lets somebody who never read the brief mark a note done. Legible on a phone at arm's length. The cut lands on the last word. Without one, closing the note needs you awake.
PriorityNote
A three-level ranking you set and define in your own note. One that works: P1 blocks shipping, P2 is worth fixing if it is cheap, P3 is taste the studio is free to decline. A list where every line is urgent gets ranked by whoever opens it, and that person cannot ask you.
Range commentTool
A note pinned to a span rather than one frame. Frame.io caps a range with the I and O keys; Dropbox Replay sets one by holding shift and dragging. It removes the question of where exactly.
Time shiftingResearch
Working outside your own hours to keep a conversation alive. Chauvin, Choudhury and Fang found non-routine employees raised out-of-hours synchronous communication by 18 percent as distance grew, offsetting the in-hours drop.
Cold watchPractice
The first viewing, at phone size, muted, before you read anybody's comments or remember what was intended. The only pass where you see the ad roughly the way a stranger will.
Two of these are contract or research words and the rest are working habits. The habits are the half you control, and they are the half that decides whether a round closes in one delivery or three.

06The deadline is the end of your working day, in every US zone

Because the studio's day starts after yours has ended, a note filed at any point during your day is on their desk when they open. That makes the deadline generous and identical in every US time zone: send it before you close the laptop. The overlap arithmetic behind that, zone by zone and season by season, sits in your first ninety days with a creative studio, with the 72-hour schedule laid out half-day by half-day.

So the instruction is short. Block the last half hour of the day a cut lands, before anything else fills it, and file the list then. The whole risk in this arrangement sits on that half hour, and no amount of studio speed covers for it.

Before you send the list

Tick as you go - it remembers
0%
Seven checks, about five minutes once the list is written. The last one is the hardest and it saves the most: a list that grows after the handoff turns one round into three, and the studio has no way to tell you that is happening.

Questions people actually ask

Open what you need
What is an async creative review?

A review where nobody is expected to be online at the same time. One person collects every comment, resolves the contradictions, and files a single consolidated list with a timecode, a priority, a reason and an acceptance test on each line. The studio answers the whole list in one delivery. It works because the list is complete enough to be acted on without a follow-up question, which is the only thing that keeps the round to one.

How do you give feedback on a video ad across time zones?

In one list, filed before the end of your working day, with a timecode on every line. Where the two working days do not overlap, your note reaches the studio before they open, so the deadline is the end of your day rather than any specific hour. The expensive habit is drip-feeding comments as they occur to you, because each fragment can only be answered on the next handoff twenty-four hours later.

What should a timestamped video note include?

Four things. The timecode or range, so nobody has to search for the moment. A priority, so a partial fix is still the right partial fix. An objective reason, which Nielsen Norman Group's guidance on expert reviews frames as an explanation rather than a criticism. And an acceptance test: the condition that would let somebody mark it done without checking with you. Notes with all four can be worked through in one pass.

Does async communication replace the meetings a time-zone gap takes away?

The best evidence says no, with an important caveat. Chauvin, Choudhury and Fang found asynchronous volume did not rise to replace lost synchronous communication, at a coefficient of -0.007 and a p value of 0.743. But that was measured in a firm whose default was synchronous, on pairings averaging 5.7 hours of overlap. It is not a finding that async cannot carry the work when async is the design rather than the fallback.

Which tool is best for client feedback on video?

Frame.io, Vimeo Review and Dropbox Replay all let a reviewer without an account pin a note to an exact frame, which is the one feature that matters. Frame.io and Dropbox Replay add range comments and drawing on the frame. Dropbox Replay is the only one whose documentation describes filtering comments by version. Loom is for sending a spoken walkthrough: its own docs say viewers who are not signed in cannot see other people's comments.

How long should a client have to review a cut?

Read your contract, because in two of the three executed agreements I read for the companion piece, silence is approval. The windows and what they cost are in what counts as a revision round. Under a 72-hour turnaround, a missed window usually costs a day of calendar rather than a right.

The studio-side account of this distance, in the first person from India, is on the sister site in working with an Indian studio from New York. Read the two together: that one is about trust, this one is about the clock.

Closing every question inside the note takes about forty minutes on a fifteen-second cut. Forty minutes is your entire obligation in this arrangement, and the only stretch of the schedule entirely yours to control. Spend it and the round closes. Skip it and you have bought a slower studio without changing anything they do.

Why this studio

If the problem on this page is yours, this is the studio that fixes it.

  • Built, not brokered. The person who wrote this reads your brief and makes the ad: over a thousand finished creatives, film and stills, for DTC brands and hotels.
  • Fast enough to test. A first cut inside 72 hours, platform-native and built from your own product, then a monthly plan when you want a steady supply, and a number in the reply.
  • Honest about the odds. We measure and publish our own keep rate: 43% of what clears every automated check survives our eye, and you only ever see the survivors.
One cut, one round

Send a link. Get one finished ad, free, and test the note on it.

We build one finished cut from your product inside three days, before any money changes hands, and it is yours to run either way. File one consolidated list against it and see what comes back in a single handoff.

Replies within a day. Ad within three.
Badal Kariwal
Badal Kariwal

Runs Ephoria, an ad-creative studio, and builds the work himself: over a thousand finished creatives, film and stills, for DTC brands and hotels. Writes here about what buying and making ad creative involves, in dollars and in days, including where the studio loses. The person who reads your brief is the person who builds the ad. Send him something to make, or see the studio's work and plans.

Where the numbers came from

  1. Chauvin, Choudhury and Fang, Organization Science. Working Around the Clock: Temporal Distance, Intrafirm Communication, and Time Shifting of the Employee Workday - read 2026-09-03 as the open-access PDF; Organization Science 35(5):1660-1681, 2024. N = 12,038 employees of one multinational across 167 cities in 48 countries, identified off daylight saving transitions. One extra hour of temporal distance cut synchronous communication 11.0 percent, 95 percent CI 6.2 to 15.5. Asynchronous volume did not rise to replace it, beta -0.007, p = 0.743. Mean business-hours overlap in the sample was 5.7 hours. Non-routine employees raised out-of-hours synchronous communication by 18 percent
  2. Herbsleb and Mockus, IEEE Transactions on Software Engineering. An Empirical Study of Speed and Communication in Globally Distributed Software Development - read 2026-09-03 as the PDF; IEEE TSE 29(6):481-494, 2003. 2,227 modification requests from July 1997 to July 1999. Single-site items took about five days, multi-site items 12.7 days, more than 2.5 times as long, p less than 0.001. The paper attributes the delay to the number of people involved rather than to distance
  3. Espinosa and Carmel, Software Process Improvement and Practice. The Impact of Time Separation on Coordination in Global Software Teams: a Conceptual Foundation - read 2026-09-03 as the PDF; SPIP 8:249-266, 2003. Coordination cost modeled as communication, clarification, delay and rework; the break the e-mail chain tactic; and the Austin firm that moved its offshore partnership to Colombia
  4. Nielsen Norman Group. UX Expert Reviews - read 2026-09-03; "Including a severity rating for each issue discovered is key to making the findings actionable", findings "mapped to where they occur in the design", and "objective explanations, not subjective criticisms"
  5. Frame.io. Commenting on your media - read 2026-09-03; comments post as a card and bubble marking the timecode position, range comments are capped with the I and O keys, an annotation tool draws on the frame, and the doc states the timecode is what the editor needs to know where to make edits
  6. Dropbox. Give and receive feedback in Dropbox Replay - read 2026-09-03; comments land on the timeline at the selected frame, ranges are set by holding shift and dragging, drawing on the frame is supported, quick reactions need no typing or account, and comments can be filtered by current or previous version
  7. Vimeo. How to use and manage video review links - read 2026-09-03; review links work for people without a Vimeo account or team seat, time-coded comments sit on the timeline and can be clicked to jump, and a permission lets a reviewer mark a video Approved
  8. Atlassian. React to videos with emojis and in-video comments (Loom) - read 2026-09-03; comments and reactions appear on the video timeline, and "Viewers not signed into Loom will not be able to see video comments or emoji reactions"
  9. Doist. The Complete Guide to Asynchronous Communication - read 2026-09-03; "At Doist, we expect people to answer within 24 hours", plus a stated target mix of 70 percent asynchronous, 25 percent synchronous and 5 percent in person
  10. GitLab. Handbook: Communication - read 2026-09-03 as the handbook's own source markdown, because handbook.gitlab.com refuses automated requests; "You are not expected to be available all the time" and "Answers like: 'will do', 'OK', 'it is on my todo list' are not helpful"
  11. SEC EDGAR. Video Production Agreement, Coolfire Studios LLC and Global Personalized Academics - read 2026-09-03; dated July 21, 2015. The client has 24 hours to review and approve, after which a deliverable is deemed approved, with one round of revisions on scripts
  12. SEC EDGAR. Creative Services Agreement, Hasbro, Inc. and RAD Technologies Inc. - read 2026-09-03; dated August 28, 2023. Ten business days for the client to inspect and evaluate a deliverable, two business days for the agency to cure, and no deemed-acceptance clause

Every figure above links to the place it was published, with the date it was read. Numbers marked as ours are measured inside this studio and we say so where they appear. We do not print a statistic we cannot point at.