Well, there are some stats available in the spreadsheet specific to Ottawa office.
City | Month | App Count | AOR | AOR to IP | IP to Test Invite | Test Invite to test | Test to DM | DM to Oath Letter | Oath Letter to Oath | App. Complete in Days |
Ottawa | | | | | | | | | | |
| January | 9 | 87 | 148 | 274 | 9 | 26 | | | 558 |
| February | 6 | 205 | 73 | 169 | 4 | 51 | | | 528 |
| March | 4 | 215 | 40 | 127 | 23 | 97 | 9 | 6 | 505 |
| April | 2 | 186 | 62 | 118 | 7 | | | | 473 |
| May | 2 | 150 | 58 | | | | | | 429 |
| June | 2 | 156 | 80 | 153 | | | | | 406 |
| July | 4 | 119 | 72 | 148 | 4 | | | | 379 |
| August | 2 | 97 | 37 | | | | | | 340 |
| September | 5 | 116 | 43 | | | | | | 312 |
| October | 5 | 103 | 34 | | | | | | 285 |
| November | 5 | 85 | 38 | | | | | | 257 |
| December | 4 | 64 | 23 | 92 | 16 | 38 | | | 216 |
You can take a look at the numbers here. The reason why Ottawa appears "slow" is probably because the application count for Ottawa office is very, very low. There are only 50 applications from Ottawa in total for all of 2020.
For comparison, take a look at the stats for Mississauga/Toronto/Brampton offices below (Most of "Toronto" and all of "Brampton" applicants are processed in Mississauga IMO)
I'm adding up all the applicants from Mississauga, Brampton and (conservatively) 50% of Toronto applicants. The app count comes to 60 + 74 + 156 (311/2) = Approximately 290 applicants that are being processed in Mississauga office.
https://docs.google.com/spreadsheets/d/1U27V95kWlCVYWB0zye7DvqoXSkyqxgbA31eEJ_TKO6Y/edit#gid=2028470052
The App count is simply off the charts for GTA applicants when compared to Ottawa applicants. Since the app count for Ottawa is low, you don't see much updates from Ottawa applicants and it might appear that the office is much slower compared to the offices in the GTA (like Mississauga office).
Not sure if any of this makes sense but this is my take on why Ottawa office appears to be "very slow". The sample set for Ottawa is very small and hence the average processing times you see might not be properly reflected with such a small sample set.