"Post on Tuesday at 11 a.m." exists in a thousand variants. Almost all of them rest on social platform engagement data — that is, on when people scroll a feed. On when people actually click a link, there is almost no public data at all.
We looked at ours: 723,783,029 redirects through short links over 12 months.
The result runs against expectation. Links are created on a sharp office-hours curve, and clicked almost evenly around the clock. The hourly spread for creation is 2.43x; for clicks it is 1.38x. Day of week barely registers: best to worst differ by 10%.
Method
Data: 723,783,029 redirects and 6,768,413 links over the last 12 months.
Times are UTC. This matters for reading everything below, and we return to it under limitations.
Link creation time is approximated by the first recorded redirect for that link. Redirect statistics carry no separate creation timestamp.
Filtering: none. Test links, service links and automated traffic all remain in the sample.
No country dominates: the largest accounts for 28.5% of redirects, the second for 19.3%. The audience is spread across the world, and that is the key to everything below.
Creation versus consumption
| Hour UTC | Clicks | Links created | Gap |
|---|---|---|---|
| 00 | 4.44% | 2.77% | −1.66 |
| 01 | 4.40% | 2.79% | −1.61 |
| 02 | 4.47% | 2.89% | −1.58 |
| 03 | 4.50% | 2.67% | −1.82 |
| 04 | 4.26% | 2.54% | −1.72 |
| 05 | 3.92% | 2.94% | −0.97 |
| 06 | 3.57% | 3.06% | −0.51 |
| 07 | 3.39% | 4.12% | +0.73 |
| 08 | 3.35% | 4.47% | +1.12 |
| 09 | 3.46% | 5.07% | +1.61 |
| 10 | 3.57% | 4.99% | +1.41 |
| 11 | 3.87% | 5.21% | +1.34 |
| 12 | 4.14% | 5.25% | +1.11 |
| 13 | 4.27% | 5.82% | +1.55 |
| 14 | 4.43% | 5.87% | +1.44 |
| 15 | 4.61% | 6.18% | +1.57 |
| 16 | 4.42% | 5.25% | +0.83 |
| 17 | 4.29% | 4.66% | +0.38 |
| 18 | 4.38% | 4.54% | +0.16 |
| 19 | 4.39% | 4.27% | −0.12 |
| 20 | 4.32% | 4.04% | −0.28 |
| 21 | 4.62% | 3.85% | −0.77 |
| 22 | 4.52% | 3.54% | −0.98 |
| 23 | 4.43% | 3.19% | −1.24 |
Both curves come from the same global sample, measured the same way. They behave completely differently.
Link creation is a textbook working day. Trough at 04:00 (2.54%), rising from 07:00, peak at 15:00 (6.18%), declining into the night. Trough to peak: 2.43x.
Clicks are close to flat. Trough at 08:00 (3.35%), peak at 21:00 (4.62%). The spread is only 1.38x, and half the hours sit inside a 4.2–4.6% band.
This is not simply a timezone artefact. Both samples are spread across the same countries — if adding up timezones were the whole story, it would flatten both curves equally. The difference lies elsewhere: link creators are few and clustered in a handful of countries, while the audience is numerous and lives everywhere. You publish on your clock. Your readers do not.
The clearest moment is 08:00 and 09:00 UTC. Those are the hours when link creation most exceeds clicks (+1.12 and +1.61 percentage points). They are also the hours with the fewest clicks of the day. Peak production coincides with the trough of consumption.
Day of week barely matters
| Day | Redirects | Share | vs average |
|---|---|---|---|
| Monday | 103,607,360 | 14.31% | +0.2% |
| Tuesday | 101,607,204 | 14.04% | −1.7% |
| Wednesday | 102,834,699 | 14.21% | −0.5% |
| Thursday | 104,182,942 | 14.39% | +0.8% |
| Friday | 109,494,541 | 15.13% | +5.9% |
| Saturday | 102,741,696 | 14.20% | −0.6% |
| Sunday | 99,315,936 | 13.72% | −3.9% |
Best day to worst is 1.10x. Friday genuinely is slightly better and Sunday slightly worse, but the scale deserves perspective: moving a post from Sunday to Friday buys roughly 10%, while the difference between a good and a bad headline is measured in multiples.
Combining day and hour, the busiest slot of the week is Monday 15:00 UTC (5,175,606 redirects) and the emptiest is Thursday 08:00 (3,305,798). A 1.57x difference. That is the entire prize available from perfect timing, for an internationally distributed audience.
The device mix shifts through the day
The split between phone and desktop is not constant:
| Hour UTC | Phone | Desktop |
|---|---|---|
| 03 | 56.84% | 35.86% |
| 08 | 43.96% | 44.28% |
| 13 | 54.43% | 35.77% |
| 20 | 56.98% | 34.35% |
Phone share moves by 13 percentage points — from 43.96% to 56.98%. And there is exactly one hour in the day when desktop overtakes phone: 08:00 UTC. That is the same hour holding the daily click trough and the surge in link creation.
Put differently, the only window when your audience is predominantly at a desktop is the very morning hour when the fewest people click.
What follows
Stop optimising publication time if your audience is international. The entire available gain is 1.57x between the best and worst slot of the week, and that assumes you hit it exactly. The realistic spread between a decent and a good hour is a few per cent. The same effort spent on the headline, or on the link itself, returns more.
Do not confuse your schedule with your audience's. The creation curve shows people publishing between 09:00 and 17:00 UTC. The click curve shows people reading around the clock, and most actively in the UTC evening and night. Those are different people in different timezones, and the first does not govern the second.
Day of week needs no planning at all. A 10% difference between the best and worst day does not justify rescheduling.
Check how your landing page looks on a phone. Mobile share never drops below 44% in any hour of the day and reaches 57%. Even in the "desktop" morning hour it is almost half.
If the link leads somewhere time-limited, remember the night audience. Clicks at 03:00 and 04:00 UTC represent 4.50% and 4.26% of daily volume — more than at 09:00. A promotion ending "at midnight your time" cuts off in the middle of the working day for part of your audience.
Limitations
Times are UTC and the audience is global. This is the central limitation and it cannot be argued away. The flat click curve is a property of summing across many timezones. If your audience sits inside a single country, your curve will be noticeably sharper than the one shown here, and the "timing barely matters" conclusion does not apply to you. How much sharper cannot be established from this data; that needs a separate per-country analysis.
Automated traffic was not filtered out. Crawlers, messenger preview fetchers and monitoring run around the clock and are evenly distributed by nature. Some of the observed flatness in clicks may be attributable to them. We cannot size that share from the available data, and that is a genuine weakness of this study.
Creation time is approximated by the first click. Links created in advance and published later shift the creation curve to the right of the truth.
Tablets are not broken out. The detector did not classify them separately; they are distributed between phones and desktops. No tablet share can be derived from this data.
This is one service's data. Audience composition and link types will differ elsewhere.
A short link is an ordinary HTTP redirect, and what you tag it with matters more than when you send it: understanding where redirects came from requires UTM parameters, not guesses about timing. The practical side is covered separately in short links in marketing and short links: pros and cons.
