Outcomes every instant request from an Indian student, by IST day · losses stacked as % of requests, fulfilled as the line
By hour of day over the whole range · where in the day supply runs out
Volume requests, and what they billed
Where each day's loss fell each quarter's share of that day's lost requests · counts on hover
By grade requests as bars, the loss rate as the line — a grade asks more because more students are in it
By subject the busiest 10 · loss rate on hover, because a long tail of small subjects has no rate worth reading
Was anybody offered it? of the requests that were lost, how many tutors each was pushed to
40.5% nobody was pushed 15,979 43.4% pushed to 1-2 17,125 11.8% pushed to 3-4 4,649 4.4% pushed to 5 or more 1,750
A lost request with no rows in the push log counts as “nobody was pushed”, because that is what it is. This separates a routing failure from a supply shortage: both leave the student without a tutor, and only one of them is fixed by recruiting.
…and at which hours lost requests as bars, the share nobody was pushed for as the line
How deep the offers went, by hour share of that hour’s lost requests · counts on hover
The line above is only the red band below. Splitting the rest apart separates a matcher that never started from one that gave up after two tutors — the same loss to the student, a different repair.
Could the idle tutors have taken it? under 100% is a shortage of tutors; well over is a routing problem · the tutors behind it →
fixed week 2026-09-30 → 2026-10-07 — not 2026-09-25 → 2026-10-08, unlike everything else on this page. Tutor presence is written once a night, so it is always a day behind and cannot be re-cut by date.
130% over the whole day 4079.6 sessions of idle capacity vs 3126.8 lost 44% in the hours carrying the loss 6, 18, 19, 20h
| Hour | Lost | Tutors online | Idle | Idle hrs | Rate | Capacity | Covers |
|---|---|---|---|---|---|---|---|
| 00 | 32.1 | 110.2 | 67.8 | 21.3 | 4.5 | 95.7 | 298% |
| 01 | 23.5 | 61.0 | 36.8 | 14.5 | 3.84 | 55.5 | 236% |
| 02 | 21.9 | 38.5 | 23.8 | 11.5 | 3.89 | 44.8 | 205% |
| 03 | 22.2 | 28.9 | 17.5 | 7.9 | 4.11 | 32.5 | 146% |
| 04 | 68.9 | 35.8 | 14.6 | 5.2 | 6.57 | 33.9 | 49% |
| 05 | 79.2 | 61.2 | 18.1 | 3.5 | 9.27 | 32.8 | 41% |
| 06 | 117.6 | 95.2 | 33.8 | 9.7 | 7.55 | 73.2 | 62% |
| 07 | 84.4 | 130.0 | 48.1 | 16.9 | 7.42 | 125.4 | 149% |
| 08 | 69.5 | 162.9 | 63.2 | 22.1 | 7.1 | 157.0 | 226% |
| 09 | 42.9 | 182.4 | 81.9 | 30.3 | 6.67 | 202.5 | 472% |
| 10 | 74.6 | 192.6 | 82.1 | 28.6 | 7.25 | 207.4 | 278% |
| 11 | 98.6 | 192.1 | 74.2 | 27.7 | 7.48 | 207.0 | 210% |
| 12 | 117.6 | 196.0 | 74.9 | 26.6 | 7.57 | 201.7 | 171% |
| 13 | 101.9 | 197.9 | 77.5 | 30.2 | 7.67 | 232.1 | 228% |
| 14 | 68.5 | 202.4 | 82.5 | 33.0 | 7.17 | 236.7 | 346% |
| 15 | 66.8 | 201.2 | 86.8 | 33.9 | 6.69 | 226.5 | 339% |
| 16 | 41.4 | 224.6 | 93.6 | 36.3 | 6.76 | 245.6 | 594% |
| 17 | 43.9 | 279.5 | 112.5 | 37.8 | 7.13 | 269.3 | 614% |
| 18 | 473.8 | 343.1 | 82.5 | 28.6 | 8.77 | 250.5 | 53% |
| 19 | 940.6 | 385.0 | 85.6 | 30.9 | 8.32 | 257.2 | 27% |
| 20 | 401.5 | 372.6 | 96.9 | 34.0 | 8.03 | 272.9 | 68% |
| 21 | 86.1 | 323.5 | 103.5 | 35.1 | 6.84 | 240.0 | 279% |
| 22 | 29.4 | 256.5 | 111.4 | 36.4 | 5.76 | 209.9 | 714% |
| 23 | 19.9 | 174.9 | 88.0 | 32.4 | 5.23 | 169.5 | 853% |
Capacity is idle tutor-hours × sessions per busy tutor-hour, not a count of idle tutors: the average idle tutor is online about 22 minutes of an hour, so counting heads overstates it roughly threefold. It is an upper bound on what better routing could recover — it assumes idle tutors would work at the same rate as busy ones, when some are idle precisely because they decline. An hour that still comes out under 100% on that generous assumption is genuinely short of tutors. Hours that lost nothing show “—” rather than 0%.
Day by hour lost requests · the whole range, whatever the charts above are showing
Shading is square-rooted, so a peak hour on a busy day does not flatten every other cell to white. Hover a cell for the day, hour and count.
By subject over the range, busiest first · “(unmapped)” is a request whose sub-topic has no v1 topic behind it
| Subject | Requests | Fulfilled | Lost |
|---|---|---|---|
| Mathematics - Class 9 and 10 | 75,374 | 69.2% 52,175 | 21.1% 15,885 |
| Science - Class 9 and 10 | 45,743 | 71.9% 32,896 | 16.3% 7,452 |
| Premium | 31,632 | 78.9% 24,943 | 10.7% 3,395 |
| Mathematics | 17,128 | 72.3% 12,382 | 18.9% 3,237 |
| Physics | 14,545 | 64.7% 9,416 | 24.4% 3,552 |
| English | 14,096 | 76.4% 10,767 | 12.2% 1,725 |
| Biology | 9,440 | 75.8% 7,159 | 13.8% 1,299 |
| Physical Chemistry | 4,125 | 62.4% 2,576 | 26.1% 1,075 |
| Organic Chemistry | 3,670 | 63.1% 2,315 | 25.8% 947 |
| Inorganic Chemistry | 3,206 | 66.7% 2,137 | 17.6% 564 |
| FILO internal | 224 | 0.0% 0 | 46.4% 104 |
| Social Studies - Class 6 to 8 | 204 | 7.8% 16 | 43.1% 88 |
| (unmapped) | 84 | 0.0% 0 | 45.2% 38 |
| Maths | 82 | 0.0% 0 | 40.2% 33 |
| Social Studies | 59 | 61.0% 36 | 10.2% 6 |
| Chemistry | 53 | 3.8% 2 | 35.8% 19 |
| Accountancy | 48 | 35.4% 17 | 35.4% 17 |
| Political Science | 30 | 26.7% 8 | 33.3% 10 |
| Economics | 28 | 0.0% 0 | 53.6% 15 |
| History | 21 | 38.1% 8 | 38.1% 8 |
By grade over the range · “—” is a student with no class on file, kept rather than dropped
| Class | Requests | Fulfilled | Lost |
|---|---|---|---|
| 1 | 79 | 69.6% 55 | 24.1% 19 |
| 2 | 17 | 76.5% 13 | 0.0% 0 |
| 6 | 600 | 70.3% 422 | 15.3% 92 |
| 7 | 648 | 67.1% 435 | 18.2% 118 |
| 8 | 1,210 | 67.6% 818 | 18.6% 225 |
| 9 | 85,570 | 70.0% 59,923 | 19.0% 16,281 |
| 10 | 56,694 | 73.3% 41,565 | 16.4% 9,307 |
| 11 | 28,776 | 72.6% 20,886 | 17.6% 5,070 |
| 12 | 43,311 | 71.0% 30,735 | 18.2% 7,875 |
| 13 | 2,720 | 69.8% 1,898 | 16.8% 457 |
| 14 | 118 | 55.9% 66 | 28.8% 34 |
| 15 | 100 | 42.0% 42 | 25.0% 25 |
| 20009 | 11 | 81.8% 9 | 0.0% 0 |
Days newest first · every request lands in exactly one column
| Day | Requests | Fulfilled | Expired | Cancel >20s | Cancel 10–20s | Cancel ≤10s | By tutor | Blocked | Billed ₹ |
|---|---|---|---|---|---|---|---|---|---|
| 2026-10-08 | 16,873 | 70.3% 11,859 | 14.2% 2,391 | 5.1% 860 | 2.6% 439 | 7.7% 1,296 | 0.1% 22 | 0.1% 11 | 132,931 |
| 2026-10-07 | 16,557 | 67.1% 11,111 | 15.1% 2,505 | 6.1% 1,017 | 3.3% 542 | 7.9% 1,301 | 0.4% 64 | 0.1% 17 | 123,359 |
| 2026-10-06 | 16,848 | 66.9% 11,273 | 16.1% 2,706 | 6.3% 1,055 | 3.3% 548 | 7.3% 1,237 | 0.1% 17 | 0.1% 12 | 127,178 |
| 2026-10-05 | 13,447 | 71.2% 9,569 | 12.5% 1,682 | 6.2% 831 | 3.0% 397 | 7.0% 948 | 0.1% 15 | 0.1% 7 | 110,716 |
| 2026-10-04 | 12,988 | 79.3% 10,305 | 7.0% 903 | 4.7% 604 | 2.3% 305 | 6.5% 838 | 0.1% 18 | 0.1% 16 | 125,959 |
| 2026-10-03 | 12,126 | 77.4% 9,386 | 8.1% 979 | 4.6% 559 | 2.6% 312 | 7.1% 866 | 0.1% 14 | 0.1% 10 | 112,833 |
| 2026-10-02 | 15,888 | 70.0% 11,127 | 13.7% 2,177 | 5.7% 906 | 3.0% 481 | 7.4% 1,179 | 0.1% 8 | 0.1% 11 | 127,936 |
| 2026-10-01 | 19,673 | 60.2% 11,839 | 20.5% 4,034 | 7.0% 1,386 | 4.0% 786 | 8.2% 1,604 | 0.0% 8 | 0.1% 18 | 124,876 |
| 2026-09-30 | 18,684 | 69.5% 12,979 | 14.8% 2,765 | 4.8% 905 | 2.7% 510 | 8.0% 1,492 | 0.1% 15 | 0.1% 18 | 134,904 |
| 2026-09-29 | 18,440 | 70.3% 12,967 | 12.8% 2,359 | 5.2% 968 | 2.7% 489 | 8.8% 1,620 | 0.1% 27 | 0.1% 11 | 128,867 |
| 2026-09-28 | 16,170 | 70.3% 11,368 | 13.5% 2,188 | 5.4% 878 | 3.0% 479 | 7.6% 1,224 | 0.1% 24 | 0.1% 9 | 122,607 |
| 2026-09-27 | 13,769 | 79.4% 10,936 | 7.2% 998 | 3.7% 509 | 2.1% 288 | 7.3% 1,001 | 0.2% 29 | 0.1% 8 | 131,574 |
| 2026-09-26 | 13,594 | 80.9% 10,995 | 6.2% 839 | 3.4% 464 | 2.0% 274 | 7.2% 983 | 0.2% 27 | 0.1% 12 | 120,777 |
| 2026-09-25 | 14,797 | 75.4% 11,153 | 9.5% 1,412 | 4.2% 623 | 2.7% 398 | 8.0% 1,187 | 0.1% 16 | 0.1% 8 | 120,217 |
Indian students only; internal accounts, package class 16 and throttled sessions excluded. “Lost” in the charts is expired plus cancelled after 20 seconds — a student who waited and got nobody. Cancellations under 10 seconds are counted but not read as supply.