← all articles

screen recording storage on a mac: the real numbers

everyone asks this before they install anything, and nobody publishes an actual measurement. so here is one.

on a working install here, running on an m series mac at 1710 by 1107, the archive came to 26gb across 17 distinct days of capture. that is roughly 1.5gb per recorded day. more usefully, the split inside that 26gb is not what people expect, and it changes what your retention policy should do.

measured september 2026. your number will differ, and the reasons why are the interesting part.

the split that matters

the searchable index is 5% of the archive. the video is the other 95%.

whatsize on the measured installshare
video recordingsroughly 24.6gbabout 95%
searchable index (ocr text and metadata)1.4gbabout 5%
everything else (settings, extensions, derived notes)under 15mbrounding
total26gb100%

that ratio is the most useful fact on this page. the thing you actually search, the text pulled out of every frame, is almost free. the thing that costs you hundreds of gigabytes is the pixels, which you look at far less often than you think.

per day, on this install, that is about 82mb of index and about 1.45gb of video.

what a year actually costs

extrapolate and the decision makes itself.

at 250 working days a year, the same rates give roughly:

  • video: about 360gb a year. meaningful on a 512gb machine, painful on anything smaller.
  • searchable index: about 20gb a year. unremarkable on any modern mac.
  • both, for five years: nearly 2tb. not a plan.
  • index only, for five years: about 100gb. entirely reasonable.

so the honest answer to "can i keep my whole working life" is: the searchable part, yes, easily. the video, no, and you would not want to.

why it is smaller than you would guess

naive maths says continuous screen recording should be enormous, and four things stop it.

compression rewards boredom. modern video codecs encode differences between frames. a screen showing a mostly static document compresses to almost nothing, because almost nothing changed. the expensive content is video playback and fast scrolling, which is a small fraction of most working days.

focused window only. capturing one window rather than every pixel on every display is a large multiplier, especially on a two monitor setup. remynd captures the focused window for exactly this reason, which is also a privacy decision.

skipping unchanged frames. if the screen has not changed, there is nothing worth writing. a lot of a working day is you reading, thinking or away from the machine.

ocr output is text. a frame is measured in hundreds of kilobytes and the words on it in hundreds of bytes. that is the whole reason the ocr index size stays at 5% of the archive. apple's vision framework does the extraction on device on the neural engine, so it costs battery rather than bandwidth.

how much disk space should you actually plan for?

decide from your machine, not from the marketing.

  1. 256gb mac. index only, short video retention, 14 to 30 days. anything more will fight you.
  2. 512gb mac. the common case. 60 to 90 days of video plus an unlimited index sits comfortably.
  3. 1tb or more. 6 months of video is fine if you want it. you still probably do not need it.
  4. external drive. worth it only if you genuinely rewatch. most people search text and never scrub video.

whatever mac storage you have, the pattern behind all four is the same: keep text long, keep video short. almost every "i know i saw it" moment is weeks or months old, and the text answers it. the video is for the rare case where you need to see the thing rather than read it.

the retention policy question nobody asks

a retention limit is a privacy control before it is a storage control.

this is the part that gets missed when people treat disk space as purely an engineering question. an archive of everything you have seen is a concentrated target, and the cheapest mitigation available is not keeping it forever.

microsoft reached the same conclusion with recall, which ships off by default with biometric gating and a 15 minute inactivity timeout after the privacy reception of may 2024, assessed in is microsoft recall safe. apple came at it from the other direction, shipping siri ai with onscreen awareness on june 8, 2026 that reads your screen live and stores nothing, which costs 0gb and answers a much narrower question, as we covered in apple intelligence on mac.

so set the limit deliberately, and set exclusions at the same time. password managers, banking, health portals, anything belonging to a client. the storage saving is a bonus; the real point is that data you never captured cannot leak, and a shorter window is easier to defend under every regime in is it legal to record your screen at work.

what to check before you install anything

five questions, and storage is only the first.

  1. what is the per day rate on your hardware, measured rather than promised.
  2. is there a retention setting at all, and what is its default. remynd's is 30 days of recordings.
  3. can you exclude by app and by site, not just pause.
  4. where does the archive live, and can you export it in a format something else opens.
  5. what happens if the company disappears, which is not hypothetical in this category.

that last one has a track record. rewind defined screen memory on the mac, and in december 2025 meta acquired the company and shipped an update that disabled all capture on december 19, 2025, 14 days after the announcement. the storage on your disk is only yours if the format is. we set the general case out in what happens to your data when an ai app shuts down.

why the index is the part worth keeping

because searching is what people actually do, and searching only needs the text.

per the anthropic economic index for may 2026, the number one work task in sampled ai conversations is searching electronic sources for information at 4.95%, with reference searching second at 3.74%. that is 8.69% of conversations spent looking things up, while knowledge retrieval and enterprise search sits at 3.61% of request topics and personal ai assistant at 2.86%.

the demand is retrieval. retrieval runs on the 5%, not the 95%. which means the useful version of this product is much cheaper to keep than the scary version people imagine when they hear "records your screen."

remynd is built around that. it ships a retention period you set in settings, defaulting to 30 days of recordings, so the expensive 95% is bounded out of the box. capture, ocr and storage stay on your mac by default, it captures the focused window rather than every display, and you can exclude specific apps or sites entirely. sign in and the cloud agent do reach the network, and your history is read only everywhere in the app. the fuller architecture is in private ai on your mac.

download remynd for mac and set your retention on day one.

common questions

how much disk space does continuous screen recording use? +
on a measured install here, 26gb across 17 days of active capture on a 1710 by 1107 display, which is roughly 1.5gb per recorded day. that scales to something like 375gb a year at 250 working days. your own number will differ a lot depending on how much the screen changes, since compression rewards static content.
what actually takes up the space, the video or the text? +
the video, overwhelmingly. on the same install the searchable index was 1.4gb against roughly 24.6gb of recordings, so text is about 5% of the total and video is about 95%. that ratio is the single most useful fact for deciding a retention policy.
can i keep the search but delete the video? +
that is the right shape for most people, and it is what the 5% to 95% split implies. text extracted by ocr is tiny, so a year of searchable history costs roughly 20gb while a year of video costs hundreds. keeping text long and video short gives you most of the value for a fraction of the disk.
does screen recording slow down a mac? +
on apple silicon the capture and ocr run on hardware built for it, so the cost is mostly disk rather than cpu. the things that actually hurt are recording every display instead of the focused window, and recording frames when nothing on screen has changed.
what retention policy should i set? +
start with video at 30 to 90 days and text kept indefinitely, then adjust. most "i cannot find it" moments are weeks or months old, which is exactly the range where the text still exists and the video no longer needs to.