“The values of Sangh flow like blood in my veins… the day I die, I want to be draped in the BJP flag” whispered an audibly emotional and teary eyed Kalyan Singh to a huge gathering of party workers in the Jhulelal Park of Lucknow on January 22nd 2013. It was the homecoming of the prodigal son who had redefined Hindutva of the 90’s by expanding the social base of the Bharatiya Janata Party by taking it beyond the Brahmins and Baniyas of heartland. If ever any political leader deserves the title of “Hindu Hriday Samrat” it is this 3 term Chief Minister of Uttar Pradesh who didn’t blink an eyelid to give up all power and pelf for Ram Janam Bhoomi in 1992. Yet, just about a year later, the Kalyan Singh led BJP lost the 1993 elections despite being the single largest party.

“Despite having a clear lead of over 4% against our nearest rivals (the unlikely alliance of Mulayam Singh and Kanshi Ram), we did not get close to a majority because of small internal sabotage by some of our own Brahmin leaders in 30 seats” a former Kalyan Singh aide had explained to me in Mathura many years ago. Kalyan Singh himself had averred on many occasions that “the day even 50% Hindus forget their internal differences and have joint socio-political ambitions, no power on Earth can stop BJP from winning any election”. Throughout his active political life until early 2000s, Kalyan Singh was willing to take any amount of risk to achieve this Hindu unity. He was the original beacon of United Spectrum of Hindu Votes – USHV – in the heartland.

“Prime Minister’s Varanasi road show was scheduled for Sunday, while Saturday was just supposed to be a visit to the temples, but the local units felt that his images on TV and Social Media could electrify voters in the Gorakhpur division that was voting on Saturday… PM readily accepted the local unit’s suggestion for an impromptu road show” confessed a BJP strategist on March the 4th. Narendra Bhai, not unlike Kalyan Singh, is willing to take any amount of risk to sustain USHV today, he willingly risked over exposure of a Prime Minister in a local state election like Uttar Pradesh despite the amount of criticism and bad press he received after the Bihar debacle. History will probably judge Modi as the leader who rescued an ancient civilization from decaying under the overt onslaught of political secularism.

Uttar Pradesh 2017 was not an easy election for BJP. It was a hard fought battle at every level. Yes BJP did have the cushion of 2014, but what must not be forgotten is that this was the first time the party was in real contention in the state assembly election after more than 2 decades. The Sanghathan grassroots infrastructure has always existed across the state, even when the party was in the wilderness for 20 years which is why BJP could make that quick transition from 10 MPs to 73 in 2014, but it needed reactivation and rejuvenation at multiple levels. For an entire generation of voters in small town and rural Uttar Pradesh, BJP was essentially a ‘national party’ and the local assembly battle was preconceived to be between SP and BSP. This is where the Sanghathan had to play its role to change the perceptions.

For many months before the elections, Amit Bhai Shah organized “Kamal Melas” in the UP hinterland which attracted huge footfalls. Each Kamal Mela was a full-fledged exhibition with an RJ, a puppet show, a magic show, a techno virtual reality booth and a ‘selfie with Modi’ stall and what have you. Designed to attract the young rural audience it was a huge hit among all sections of the society, including hundreds of young Muslim couples in each town. For instance, today, many Delhi journalists and editors are suddenly discussing about the Ujjwala Yojana of providing free LPG to the rural poor, but very few people know that it attained cult status among rural women because of such Kamal Melas where Modi government’s schemes were extensively propagated to an audience that was in awe.

The other aspect of the Shah strategy was to inorganically bring in lateral talent from other parties who had assembly and block level personal equations with large number of voters. As per our estimates, there were at least 100 such new entrants who had been given tickets by BJP. Final part of the Shah strategy to make BJP viable in the local assembly election was the tie up with small OBC based parties in eastern Uttar Pradesh which ensured that BJP would remain in contention in each and every seat of Poorvanchal no matter what the party’s core strength was.

Despite all these efforts, BJP began the election season with a near disaster in first phase on the 11th of February. This was probably the one minor chunk in the Shah armor. The party’s reluctance to involve external professional talent for political and electoral management (which had paid rich dividends in 2014) meant that the saffron campaign ship lacked the tightness of strategic output. While BJP was heavily concentrating on the east, the western flank was left vulnerable. The last ditch effort to bring Hindutva element to the core by Para dropping Yogi was more cosmetic than a deeprooted strategy. Four factors affected BJP’s chances adversely in western UP;

  1. The mismanagement of Jats who had been grumbling against the party for quite some time and BJP refused to deploy more resources to assuage them
  2. Lack of a seat by seat strategy like that of central and eastern UP which meant that the party relied heavily on polarization as the sole weapon
  3. The unwillingness to strike a deal with small local caste based parties here as experimented in Poorvanchal
  4. The SP-INC alliance had much fewer legislators in this region which also meant lesser localized anti-incumbency that helped them perform better

As we went on to report later in our Midpoint Analysis, BJP did not cave in like Bihar and instead fought back valiantly from the 3rd phase to take back the initiative. This was a remarkable comeback from the party, but the western UP underperformance did leave a tall ladder to climb. Our pre-poll survey had shown a solid lead to BJP in the west which had meant winning more than 45 seats in phase 1 itself – had that been the case, UP election would have come to an end effectively on the very first day itself. Now, BJP had underperformed by about 4-5% which meant that the party had to cover a lot of ground in the next few phases. Narendra Modi who till then had remained mostly at the top level by only planning to do some half a dozen rallies had to be pressed back into action. PM was everywhere, crisscrossing the state and doing dozens of rallies to resuscitate the USHV and support the sagging morale of the party cadre. Amit Bhai Shah too maintained a strict vigil over the Karyakarta brigade and redoubled his efforts from phase 3. In central UP, where our pre-poll showed the Grand Alliance to be ahead of the BJP, the saffron party surprised by its grand showing, especially in the Lucknow-Kanpur urban belt. Since then BJP has fought inch by inch all the way up to Varanasi to regain the lost momentum. The fight has been bloody and rough, make no mistake of that.

The 5 X factors of UP social mapping of 2017;

  • Muslim vote split was a reality just as in 2014. This is possibly the starkest difference as compared to Bihar where 91% of the Muslim vote had strongly consolidated in favour of the Mahagathbandhan. Here in UP, even the expectation of a 70-80% Muslim vote consolidation for Akhilesh Yadav doesn’t seem to have materialized as the split is more like 60-40 in favor of the Grand Alliance. Yes, Muslims have voted tactically to defeat BJP, especially in the 1st phase in West UP, but after that we discovered that even Muslim voters were impacted by localized factors like anti-incumbency.
  • The gender gap of BJP votes is a significant finding of our exit polls which we had never observed in our pre-poll surveys! Historic average of female to male voters for BJP in UP always stood at 1 or 2 points negative because the saffron party was always considered as a male dominated system (it was equal in 2014), but this time there seems to be a huge divergence as 3% more women have tended to vote for BJP as compared to men cutting across caste lines! Importantly, there isn’t much difference between Muslim men and women voters (only 0.5% difference) instead this is more of a Hindu gender cleavage, so triple talaq may not be the main issue here. One of the reasons for this variance could be the Ujjwala Yojana and its massive impact on even those women who haven’t received any LPG cylinders. In fact, this gender gap is a nightmare scenario for pollsters like us as a 3 point variance is too significant, especially bearing the fact that many women voters (accompanied by male family members to polling booths) may not have been too forthcoming about their vote to BJP. For instance, even if the gender variance is 2 or more points higher than our current projections, there could be a landslide for the saffron party that we may have missed. We must also remember that our Math models give higher weightage to male votes than female votes because of a patriarchal societal structure, so that could only increase our error in judging the extent of a landslide in case of a deeper variance.
  • Relative turnout of voters is an important subset of Indian psephology that most pollsters miss out on – the 5Forty3 Bihar math models were accurate because of this very factor. We have a very complex but also very accurate system  to measure relative turnouts which involves tracking specific polling booths dominated by particular caste/religion groups and then comparing the same with historic averages. In 2017 UP we found that Brahmin votes were significantly less enthusiastic about going out to polling booths as compared to 2014. We also found that Brahmins were especially more ‘lazy’ in seats where NDA candidates belonged to OBCs and MBCs. Yet, the overall Brahmin turnout was slightly higher as compared to their recent history. This is definitely one negative aspect that affects the USHV adversely.
  • There was a big divergence among Jat voters between our pre-poll survey and exit polls which indicates that the ‘Biradari’ had taken a decision to go against the BJP on the night before the election, especially in phase 1. This 12% change in Jat vote preference as compared to our pre-poll and almost 36% change as compared to 2014 is a factor in at least 40+ seats of West UP.
  • The Non-Yadav OBC votes did consolidate in favor of BJP just as our pre-polls suggested which should give Modi a big booster dose in his efforts to win Uttar Pradesh. Beyond 2017, the import of this is going to be much more significant on the long term trajectory of USHV and the continuing dominance of BJP as the main pole of Indian politics.

Historically, Uttar Pradesh assembly elections have always been closely contested affairs with many sub-regional variations and hugely localized assembly level contests. Being such a vast state, UP has always been politically unstable and volatile. Even in the last 2 elections when SP and BSP won simple majorities the vote-share gap was not significant at all, instead it was just the receding of BJP into sub-20% levels that made the contest structurally bipolar and created a majority. We were expecting something similar to play out this time with Maya collapsing to around 20% levels and BJP going beyond 35% to proclaim a big saffron majority. That has clearly not happened. In fact, this election had all the hallmarks of the 1990’s phenomenon wherein BJP was in pole position while the 2 caste based opposition parties fought against it by consolidating their own bases. Of course, the one momentous difference from then and now is the larger than life presence of Prime Minister Modi and his undeniable popularity – the only extra question that we asked voters in our exit poll was what would be their likely voting pattern in 2019 and the results were staggeringly in favour of Modi.

BSP has been the big X factor of this election because Maya has refused to fade away without a fight. Of course, one important obstacle she faces is the very adverse seat conversion ratio which limits her overall impact, but she has definitely had a massive impact on both the frontrunners. We must also remember that pollsters tend to underestimate BSP in almost every poll simply because her voters are very much below the radar and also hard to track by fieldworkers. Thus, even if Maya over-performs by about 2-3 points as compared to our current projections she could potentially upturn the election in many ways by upstaging Akhilesh Yadav because her incremental vote can now mostly come from Muslims. The UP math is such that BSP in the 20 to 25% range would eat out mostly BJP voters whereas in the 25 to 30% range Maya would completely damage Akhilesh Yadav and his grand alliance.

The SP-Congress alliance was born with the sole ambition of preventing a Muslim vote-split, but that doesn’t seem to have materialized on the ground to the extent that they had hoped. Since it was a hastily announced marriage, the synergy was missing at many crucial phases, especially as both the parties contested against each other in more than 2 dozen seats – 1 glaring example of this was Rahul Gandhi’s home seat of Amethi where SP and Congress fighting each other seems to have added a big chunk to Garima Singh’s (BJP candidate) margin. Yet, Prashant Kishore (PK) did manage the election pretty well after taking over the Grand Alliance’s war room. PK’s work was visible in his capacity to manage seat-by-seat contest much better than BJP. One of the biggest worry for the Grand Alliance is that in UP, past data suggests that voters tend to vote out at least 2/3rd of the sitting legislators in each subsequent election which is why we observed high degree of anti-incumbency against many sitting Samajwadi legislators, especially in those districts where the party had swept in 2012.

BJP’s biggest strengths in this election were Narendra Modi and USHV. The significant non-Yadav OBC attraction of the saffron party was a reality that was clearly visible in the data. Poorvanchal did give BJP a happy hunting ground because of great caste and social coalitions that it had built. Yet BJP did underperform by about 3 percentage points from our pre-poll numbers which needs to be clearly explained and articulated through the following factorials;

  • Setback BJP suffered in the Jat dominated western Uttar Pradesh meant a lot of catching up to do for the saffron party
  • BSP never fully collapsed and in fact gained 4 percentage points from our pre-poll survey which affected BJP’s USHV more than the alliance’s Muslim votes
  • Contest became increasingly localized which meant that some of the core voters tended to behave along local factors (Brahmin laziness in OBC seats is one example) rather than along nationalistic perspective like in 2014
  • Smaller sub-regional caste ethnic parties like RLD, the Nishad party, Peace Party etc. were propped up with enough money by opposition strategists at the last moment which may have hurt USHV at the local level
  • Although Amit Shah’s strategic depth was refreshingly strong unlike Bihar, BJP did lack the professional touch of outside talent like PK that could have gone beyond the internal intrigues of the party to create more sustainably winning strategies
  • Some voters were also a tad disappointed that BJP had not “transferred” at least some benefits of the demonetization drive – this is probably because of the campaign by many local BJP leaders who had promised benefits along the Ujjwala scheme lines after Notebandi
  • The Sangh was not as active as in 2014, especially at polling booth mobilization

These relatively small factors brought BJP within striking distance of the opposition. Yet, 4% is a good enough lead for a clear majority going by past 2 elections. In fact, the upside risks of BJP adding a point or 2 above our current projections are higher than downside risks. If BJP adds even 1 or 2 points from here, or if the grand alliance falters by another percentage point, then there could even be a saffron landslide. On the other hand, the counter argument is that BJP, by not touching 35% in our projections has rendered itself vulnerable to the 3% error of margin rule which could leave it below the majority mark in case of greater attrition than out current projections.


Any conversion of vote-shares into seats is fraught with danger, but we are testing a new seat conversion algorithm based on our new tech tool MAPi – Micro Analytical Projections (INTELLIGENCE) – that gives sufficient weightage to district level performances of different political parties. Both MAPi and our new algorithm are still in Beta versions, so there could be errors too, but we still believe this is some progress we have made in our endeavor to bring in rigorous math into Indian elections. Based on all our exit poll findings, here are 4 key factors that one needs to consider about the Uttar Pradesh assembly;

  1. After all the rounds, we believe there are some 90 close contest seats and whoever wins maximum of those should end up getting a disproportionate size of majority. Thus a 250+ like performance is still possible for BJP if it manages to win 75% of those close contests.
  2. BJP not only had the largest pool of seats to pick from as it was in top contention in 370+ seats, but also since it had very few sitting legislators it suffered from very little localized anti-incumbency. In essence, the saffron party had to win 1 out of every 2 seats that it was in strong contest
  3. SP-Congress alliance has a double obstacle – it is in strong contest in under 300 seats which means it has to win 2 out of every 3 seats it is in contention and also a large number of its sitting legislators face huge localized anti-incumbency
  4. Unless BSP has silently scored 50% of Muslim votes consistently in all phases and all districts, the party is not in a position to cross the 120 seat barrier as of now because of adverse seat conversion ratios

If BJP manages to win Uttar Pradesh, as we are projecting based on our rigorous exit polling exercise, then all opposition to NaMo would have been vanquished for the foreseeable future. What would be left as forces of opposition would be some left fringe groups of Communists and the likes of Mamata Banerjee. Of course, Arvind Kejriwal could emerge as a potential challenger provided he matures as a politician and controls his obsessive Modi hatred. As one long time BJP watcher puts it succinctly, “If BJP win UP it is also winning possibly 2 other states as an added gift; Bihar, because Nitish is already fed up with his alliance partners and Tamil Nadu which is desperately looking for leadership and political stability in the post-Amma era”. In more ways than one, Uttar Pradesh 2017 could also be deciding the political trajectory of India for at least the coming decade.

A note on our Exit Poll:

This is by far the largest sample size for a single state election that we have ever done. Since price per response rate came down to roughly 55  rupees due to fieldworkers only having to gather near polling booths for talking to respondents and due to very small questionnaire size, we decided to take the risk of conducting such a big poll with a sample size close to 20k. Our exit poll survey covered a whopping 59 districts and 132 assembly segments spread across 411 geographic locations of all the subregions of the state! We also ensured absolute adherence to our base algorithm of caste and class groups of Uttar Pradesh thereby giving adequate representation to all social groups.

SI No:

Exit Poll Activity



No: of districts targeted



No: of assembly segments covered



Target Sample Size



No: of respondents achieved