Driving scores recorded by Zuno SmartDrive among Mumbai motorists declined during the first eight days of Ganeshotsav 2026, with the share of trips classified as clean runs falling from 73.7% to 56.2%, according to an analysis of 219 scored car trips from drivers across two comparable periods. The analysis was published on 22 September 2026 and covers trips made during 14 to 21 September — the period beginning on Ganesh Chaturthi — set against those recorded in the days immediately before the festival.
The data comes from Zuno SmartDrive, a driving-score feature of Zuno General Insurance. The comparison is internal: the analysis measures Mumbai against itself, not against another city or a national average. Figures described as findings cleared a two-proportion z-test at the 95% confidence level; other figures are noted in the published methodology as descriptive.
The analysis draws on 114 scored car trips from 42 Mumbai drivers in the pre-festival period and 105 trips from 40 drivers during the eight festival days. Percentages are calculated only among trips carrying a driving score, as the proportion of recorded trips that receive a score was still changing as SmartDrive adoption grew in the period.
Key Findings
How the scoring system works
The SmartDrive score is measured on a scale of 0 to 100 and reflects five aspects of how a car was driven: speed, braking, acceleration, cornering and phone use. A score of 95 or above is classified as a clean run, defined as a drive with no sharp or sudden events across any of those five dimensions. A single hard brake or a tight corner can move a trip that would otherwise score a 96 down to a 93, dropping it from clean-run to the next band.
The system does not assess road conditions, traffic density or whether a route was diverted. It records the car's movements and scores them against consistent thresholds regardless of context. This matters for interpreting Ganeshotsav findings: what the data records is how the car moved, not why it moved that way.
Where the missing clean runs went
When a city-level average falls, the common assumption is that more drivers were performing badly. The Ganeshotsav data does not support that reading for Mumbai. Trips scoring below 85 rose from 7.0% to 8.6%, a shift of under two trips in a hundred that did not reach statistical significance.
The band that grew was the 90-to-95 range, which nearly doubled from 12.3% to 24.8% of all trips. Drives that had previously cleared the clean-run threshold by a small margin appear to have slipped below it, while trips at the lower end of the scale did not increase materially. The analysis describes the conclusion in direct terms: the drives that left the top band went one step down, not to the bottom.
This distinction carries practical weight. A city where scores dropped because more drivers were braking sharply and handling phones at speed tells a different story from one where the distribution shifted because congested roads and tighter corners introduced single events into otherwise smooth drives. The Mumbai data is more consistent with the latter.
What individual score components showed
For a smaller subset of trips — those that Mumbai customers opened in the app — the five components of the driving score could be examined separately. This group comprised 281 trips from 58 drivers before the festival and 152 trips from 41 drivers during it. Because this is a more self-selected and engaged group than the headline sample, the analysis treats component-level figures as showing direction only, not magnitude.
Braking scores improved from 89.25 to 91.43 across the two periods. Phone-use scores improved from 94.02 to 95.81. Speed scores showed almost no movement, going from 95.07 to 95.03. Acceleration scores fell slightly, from 91.71 to 90.87. Cornering scores fell from 93.32 to 90.9.
The braking improvement is notable given that the overall clean-run rate declined. It suggests drivers were braking earlier and more gently — a pattern consistent with anticipating pedestrians stepping into traffic near pandals or procession routes. Cornering scores fell, which the analysis attributes to turns being physically tighter during the festival, rather than drivers taking corners more aggressively. Speed, which is the component most directly controlled by the driver independently of road conditions, did not change.
Late-night driving and the shift in timing
The single largest proportional shift in the dataset was in trips beginning between midnight and 4am. One trip in 114 fell in that window in the pre-festival period. During the festival, 10 trips in 105 did. The analysis notes this is the sharpest single change in the data and is consistent with festival observances — particularly the aarti, which can run late into the night — creating journeys that would not otherwise occur.
The distribution of driving across the day changed in other ways. Trips taking place between 11am and 5pm fell from 34.2% of all recorded drives to 23.8%, though this figure did not reach the significance threshold used in the analysis and is noted as descriptive. Evening and night driving rose correspondingly. The shift indicates that during Ganeshotsav, Mumbai's driving activity moved later in the day across the sample, with afternoon activity giving way to drives concentrated in the early evening and beyond.
Longer trips and what they may reflect
Trips of 15 kilometres or more rose from 19.3% of Mumbai's recorded drives to 33.3% during the festival. Short trips under 2 kilometres declined. The median trip distance moved from 4.6 km to 6.6 km, though that specific change only reached a p-value of 0.10 under the Mann-Whitney test used for distance data — above the 0.05 threshold — and is not presented as a standalone finding. The share of longer trips did clear the test.
The analysis identifies two plausible explanations for the increase in longer drives but does not claim to distinguish between them. Roads closed around procession routes or pandals would add distance to journeys that would ordinarily be shorter, routing drivers around sections of the city they would normally pass through. Separately, people driving further across the city to visit family or attend specific pandals would also produce longer trip distances. Both effects may be present simultaneously, and the data — which carries no route, pin code or locality information — cannot separate them.
The analysis carries no locality or sub-city information. There is no Lalbaug figure, no Andheri figure, and no ward or pin-code breakdown. The data records a city and nothing finer, so no area-level conclusion can be drawn and none is offered.
The data also does not constitute evidence of worse driving in any general sense. The proportion of trips scoring below 85 — the range that would indicate consistently poor driving events — was statistically unchanged. The analysis is explicit about this: the sample measured how cars moved, not how the city was arranged. Conclusions are confined to the four findings in the table and do not extend to broader assessments of driver behaviour or road safety.
Weekend-specific figures are published in the methodology appendix rather than the main analysis. Weekend trips scoring 95 or above fell from 55.2% to 43.6%, and weekend driving after dark rose from 37.9% to 59.0%, but both figures are drawn from counts in the teens and thirties and are presented for completeness rather than as reportable findings.
The analysis uses a two-proportion z-test at the 95% confidence level for all percentage-based comparisons, Welch's t-test for average scores, and the Mann-Whitney test for trip distance, given that distances are skewed and a mean would be misleading. A test result below 1.96 — or a p-value above 0.05 — means the sample cannot separate the observed change from noise. The analysis does not claim this means the change is absent, only that the sample is insufficient to establish it.
Percentages are calculated over trips that carry a driving score. Roughly half of recorded trips receive a score, and that proportion varied between the two periods. Using a denominator that mixed scored and unscored trips would introduce distortions that are not present in the underlying driving patterns.