Our Methodology

1. How We Source Promises

Every promise tracked on ManifestoWatch is drawn directly from official election manifesto documents published by political parties. We do not rely on press conferences, campaign speeches, social media posts, or third-party summaries. Our sole primary sources are:

Manifesto PDFs — the official document released by the party before an election, usually available on the party’s website or submitted to the Election Commission of India.
Official party policy documents — supplementary policy booklets released alongside the manifesto that elaborate on specific commitments.

Each promise entry on our platform links back to the exact page number in the source manifesto so that anyone can independently verify the original commitment. Where a manifesto is available in multiple languages, we use the official English-language version as the canonical reference and note any discrepancies.

2. How We Verify Promise Status

Each promise is assigned one of six statuses based on verifiable evidence from government records, credible news sources, court documents, and independent research. We update statuses as new evidence emerges.

Fulfilled

The promise has been substantively delivered. There is clear evidence that the central commitment has been implemented or the stated outcome has been achieved. This may include an operational policy or scheme, enacted and implemented legislation, completed infrastructure, or achievement of a measurable target.

Example: a promise to establish a specified programme is marked fulfilled when the programme has been launched, is operational, and evidence shows that the promised service is being delivered.

On Track

Substantive implementation is underway and current evidence shows continuing progress consistent with the promise. This status is particularly useful for multi-year or directional commitments such as promises to expand, strengthen, increase, promote, improve, or further develop an existing policy or programme where the manifesto does not specify a clear numerical endpoint or completion deadline.

An announcement or approval alone is not sufficient. There must be evidence of implementation and progress, such as actual rollout, procurement, expenditure, beneficiary growth, geographic expansion, increased capacity, acreage, production, construction, or another relevant measurable indicator.

Example: “Further incentivise crop diversification with price support” — relevant programmes and procurement mechanisms are operational and measurable expansion is occurring, but the promise has no defined numerical endpoint at which it can yet be considered fully completed.

Partially Done

Meaningful action has been taken, but a substantial part of the promise remains undelivered. This may occur when implementation covers only part of the promised scope, major components are still missing, beneficiary or geographic coverage remains substantially below the commitment, progress has stalled, or a stated numerical target has only been partly achieved.

Example: “Build 2 crore houses” — 1.2 crore houses have been completed, leaving a substantial gap against the stated target.

Broken

The promise has been explicitly reversed, the relevant scheme or policy has been cancelled, or the government’s term has ended with no meaningful action taken towards fulfilment. A promise may also be marked broken when government action directly contradicts the commitment.

Example: a commitment is marked broken if the government formally abandons the promised policy or takes action directly contrary to it.

Pending

The government’s term is still active and there is no verifiable evidence of meaningful action — positive or negative — on the promise. It has neither been substantively started nor contradicted.

We do not normally mark a promise as broken while the term is ongoing merely because implementation has not yet started, unless there is an explicit reversal or contradictory action.

Disputed

Credible evidence materially disagrees on the degree of fulfilment. This may occur when government data conflicts with independent assessments, when the definition or measurement of the promised outcome is contested, or when authoritative sources reach different conclusions.

Example: government data may indicate that a target has been met, while credible independent analysis disputes the methodology or underlying measurement.

How we assess promises without numerical targets

Not every manifesto commitment contains a numerical target or a clear completion date. Promises using terms such as “expand”, “strengthen”, “increase”, “promote”, “improve”, “encourage”, or “continue” are assessed against the direction and scope of the commitment rather than an artificial percentage of completion.

For these promises, we look for three things:

Direct relevance — the government action must clearly address the specific commitment, not merely relate to the same policy area.
Implementation — the commitment must have moved beyond an announcement, proposal, or approval into actual implementation.
Observable progress — there should be evidence of expansion or improvement after the election, such as increased coverage, beneficiaries, capacity, acreage, procurement, infrastructure, expenditure, or another relevant measurable indicator.

Where these conditions are met and implementation is continuing, the promise may be classified as On Track. Where implementation remains materially limited, incomplete, or significantly short of the promised scope, it is classified as Partially Done. A promise is marked Fulfilled only when the substantive commitment itself can reasonably be considered delivered.

3. Confidence Levels

Every status assessment carries a confidence level that reflects the strength and consistency of the underlying evidence.

High

Multiple authoritative sources or strong primary evidence support the assessment. Government data, CAG reports, court records, parliamentary documents, official statistics, and credible independent reporting may be used together where available. The evidence leaves little uncertainty about the assigned status.

Medium

Evidence supports the assessment but is limited in scope or comes from fewer sources. There may be partial data, implementation may be recent, or comprehensive independent verification may not yet be available. The status is our best current assessment and may be revised as more evidence surfaces.

Low

Limited, incomplete, or conflicting evidence makes a definitive assessment difficult. The status reflects the best available information but carries significant uncertainty. We actively seek additional sources and encourage readers to submit evidence.

4. Editorial Independence

ManifestoWatch is editorially independent. We are not affiliated with, endorsed by, or funded by any political party, candidate, government body, or politically-aligned organisation. Our work is guided by the following principles:

No party affiliation — our analysts and editors do not hold membership in any political party, nor do they engage in campaign activities.
No political funding — we do not accept donations, sponsorships, or grants from political parties, their affiliates, or elected representatives.
Equal treatment — every party’s promises are assessed using the same criteria and evidence standards. No party receives preferential or harsher treatment.
Transparent sourcing — every status assessment links to the evidence on which it is based, so readers can independently evaluate our conclusions.

5. Submitting Evidence

We actively encourage citizens, journalists, researchers, and civil society organisations to submit evidence on any tracked promise. Crowd-sourced evidence strengthens our assessments and helps us catch updates we may have missed.

What we accept: news articles from credible publications, government gazettes and orders, RTI responses, court judgements, CAG and parliamentary committee reports, and data from recognised statistical agencies.

What we do not accept: social media posts, anonymous claims, opinion pieces without factual backing, or party press releases as sole evidence.

Review process: every submission is reviewed by at least one analyst. We verify the source, cross-reference it with existing evidence, and update the promise status and confidence level if warranted. Submitters are not identified publicly unless they request attribution.

6. Corrections & Contact

We take accuracy seriously. If you believe a promise status is incorrect, a source has been mischaracterised, or any information on this platform contains an error, please write to us with the specific promise, the correction you propose, and the supporting evidence.