From Deadline to Talk: How CS Conference Timing Actually Works

Everyone learns the rhythm of their own subfield eventually: this deadline in the autumn, that conference the following summer. What is harder to see is the shape of the whole calendar, and whether the assumptions you have built up actually hold.

We track submission deadlines, notification dates and conference dates in the same place, so we measured the gaps.

What is in the data

220 conference editions where we hold both a submission deadline and a conference start date, of which 161 also have a notification date. Same caveat as always: this is a curated set weighted toward established, CORE-ranked venues, not a census of everything in computing.

Deadline to conference: about six months

The median gap between a submission deadline and the first day of the conference is 6.1 months. The mean is 6.4, the shortest in our data is 1.5 months and the longest is 16.9.

The spread is tighter than you might expect:

82% of conferences fall between four and eight months. If you are trying to work out roughly when a conference will be from a deadline you just spotted, six months is a good first guess and you will rarely be far wrong.

It varies by tier, and in the direction you would expect:

Flagship venues run longer pipelines. More submissions, more reviewers to coordinate, more rounds. The C figure rests on only seven editions so treat it lightly, but the A*-to-B trend is consistent across a decent sample.

Review takes about twelve weeks

Across the 161 editions where we have both dates, the median gap from deadline to notification is 12.0 weeks, ranging from 1.6 to 21.9.

The fast end is mostly conferences with lighter review loads or rolling processes. The slow end is exactly who you would guess: SIGMOD at 20.7 weeks, NeurIPS at 20.1, ICRA at 19.7. Roughly five months between hitting submit and finding out.

That is the part worth internalising. For a large venue, a third of the year passes between submission and knowing.

The number that should change your behaviour

After acceptance, the median time before the conference starts is 14.3 weeks. Three and a half months to arrange travel, which sounds comfortable.

The median hides the problem. The range runs from 59.6 weeks down to 0.9 weeks.

Eurocrypt 2026 notified authors on 4 May for a conference starting 10 May in Rome. Six days. Not six weeks: six days to arrange international travel and a hotel in Rome in May.

It is not an isolated case. SIGCOMM 2025 gave 4.7 weeks, IWQoS 2026 gave 5.7, ACL 2025 and ICDCS 2026 both gave 8. Several of those are in cities where a month’s notice means paying whatever is left.

The practical conclusion is uncomfortable but clear: for some conferences you cannot wait for the acceptance email before making travel decisions. Check the gap for your specific venue before you assume you have time. Where the gap is short and the destination is expensive, the choice is between booking refundable accommodation on spec and paying a premium later. We wrote separately about what a conference trip actually costs and whether the conference hotel block is worth it, both of which get sharper when the window is this tight.

September is a wall

Deadlines are not spread evenly through the year:

September carries almost three times the load of August or December. And within September the clustering gets worse: 31 weeks in our data carry three or more deadlines, and the week of 14 September 2026 carries twelve.

That single week includes ICRA 2027, NSDI 2027, Eurocrypt 2027, FAST 2027, HRI 2027, DATE 2027, ICALP 2027, ICASSP 2027, SOFSEM 2027, ICAART, VISAPP and FC 2027. If your work spans systems and robotics, or theory and crypto, that is not a week you want to discover in September.

Late January is the other pileup: the weeks of 19 and 26 January 2026 carried eight deadlines each, including ICML, IJCAI, USENIX Security, SIGGRAPH and Eurocrypt.

What to do with this

Start the slow things at submission, not at acceptance. With a six month median from deadline to conference and twelve weeks of that spent in review, waiting for the accept email leaves you roughly three months. For visa processing from many countries, that is not comfortable margin.

Look up your venue’s actual notification-to-conference gap rather than assuming the median. Three and a half months is typical; six days happens.

Check the September and late-January clusters before committing to a submission plan across subfields. Our deadlines page lists what is currently open, soonest first, which is the quickest way to see whether two of your targets land in the same week.

As of today there are 47 open deadlines in our data. The next one closes today: AAAI 2027. If you are reading this hoping for more time, that is roughly the point.

We update these dates continuously as conferences publish them, so the figures shift. The shape has been stable though: about six months out, about twelve weeks of review, and a tail of conferences that will give you almost no notice at all.