BPN Exit Poll vs Results: We Called Bengal’s Political Shift with 66.9% Seat Accuracy
The results of the 2026 West Bengal Assembly election have now confirmed what the BPN and Statscope India Exit Poll had projected before counting day: West Bengal was witnessing a decisive political shift in favour of the Bharatiya Janata Party. While the final scale of the BJP surge exceeded even our upper projection range, the broader direction of the election, the collapse of the ruling establishment across large regions, and the emergence of a clear BJP majority were all identified by our model before the first counting trend emerged.
The BPN and Statscope India Exit Poll projected the BJP in the 165-185 seat range, clearly above the majority mark of 148. The final result saw the BJP cross the 200-seat mark, establishing a dominant mandate. More importantly, our seat-by-seat projection model achieved 66.9% constituency-level accuracy across declared seats, correctly projecting 196 constituencies out of 293 declared seats.
In modern Indian elections, especially in a politically polarised and geographically complex state like West Bengal, achieving nearly 70% exact constituency accuracy across 294 seats is statistically significant. The final deviation largely came from the scale of late-stage consolidation in favour of the BJP in several close constituencies that were initially projected as narrow AITC holds.
What makes this performance even more significant is that the BPN and Statscope India model was among the very few independent election exercises that openly projected a BJP majority in West Bengal at a time when much of the mainstream political discourse still portrayed the contest as highly uncertain.
The core foundation of our projection was the identification of a structural anti-incumbency wave developing across large parts of North Bengal, Jangalmahal, industrial belts, border districts, and urban clusters. Our analysis suggested that the 2024 Lok Sabha consolidation in favour of the BJP was not merely a parliamentary phenomenon, but a sign of deeper organisational and voter realignment that could translate into Assembly seats if sustained.
We also identified early that opposition fragmentation, which historically benefited AITC in Assembly elections, was weakening in several regions. In many constituencies, the anti-AITC vote consolidated far more sharply than in previous elections, leading to a much higher seat conversion rate for the BJP than conventional swing models anticipated.
One of the biggest strengths of our projection was the district-level trend mapping. The model correctly anticipated that North Bengal and western districts would become the backbone of the BJP surge, while urban and semi-urban belts around Kolkata would witness far tighter contests than previous Assembly elections. Our projections also captured the vulnerability of multiple high-profile AITC strongholds that had previously been considered difficult for the BJP to breach.
At the same time, this exercise has also provided valuable lessons for future election modelling. The primary gap between our projection and the final result was not the direction of the verdict, but the intensity of the swing in a cluster of close constituencies. Several seats that were internally classified as “lean BJP” or “toss-up” ultimately broke more decisively towards the BJP during the final stretch of the campaign and voting phases.
That is why we believe the remaining 20-30% variance in seat prediction was largely marginal rather than structural. In other words, the broad political reading was correct, but the magnitude of the consolidation exceeded our median assumptions in specific districts.
Our methodology intentionally avoided overdependence on any one source of data. While extensive ground surveys and local feedback formed an important pillar of the model, they were not given exclusive weightage. Equal importance was also assigned to historical voting patterns, constituency-level trend shifts, demographic behaviour, organisational strength, candidate impact, turnout analysis, parliamentary-to-assembly vote conversion patterns, and region-specific swing modelling.
This balanced methodology was designed to avoid one of the most common weaknesses in Indian election forecasting: excessive dependence on raw field sentiment without structural electoral analysis.
Most importantly, BPN and Statscope India believe that polling agencies and election forecasters must be willing to publicly evaluate and audit their own projections after the results. Across India, very few pollsters release detailed accountability reports comparing their predictions against actual constituency-level outcomes. We believe this transparency is essential for improving electoral modelling and maintaining public trust.
No other pollster is currently doing what we are attempting to do: publicly taking accountability for our numbers, openly discussing where the model succeeded, where it underperformed, and how future methodologies can be improved.
As India moves into the next cycle of major state elections, the West Bengal exercise will now become a benchmark for refining our future models. The objective is not merely to predict winners, but to continuously improve constituency-level precision through transparent post-result analysis.
Projection vs Result
BPN & Statscope India Projection: BJP: 165-185 AITC: 99-118 INC: 8-11 CPI(M): 0-2
Final Result: BJP: 207 AITC: 80 INC: 2 CPI(M): 1 Others: 3
Seat-by-Seat Accuracy: 196 correct out of 293 declared seats
Seat-by-Seat Accuracy Percentage: 66.9%
Major Call Accuracy: Correctly projected BJP majority in West Bengal
