How Vote Record Check is built

What it measures

For each firm we show the share of its public proxy votes cast with management, by category and season, next to peers of a similar size. Support is the plain tally of how a firm voted. It is neutral. A high or low figure is not good or bad, and engage insights takes no view on how anyone should vote.

United States: Form N-PX

US figures are computed from SEC Form N-PX, the annual filing in which funds disclose every proxy vote. Since the 2024 season these are filed as structured data, so we read each individual vote rather than a firm's own summary. We map the filing's categories to five plain buckets: director elections, say on pay, shareholder proposals, mergers and capital, and other.

• Seasons. A season is the twelve months to 30 June, the N-PX reporting period. We hold the three most recent seasons.
• Which filing counts. When a fund amends a filing, the later one supersedes the earlier, so each period is counted once.
• Series to complex. A fund family files under one or more identifiers. We group all of a family's filers into one complex, so the figure reflects the house, not a single fund.

Europe: SRD II Article 3g

There is no central European equivalent of N-PX. Under SRD II Article 3g each manager or owner publishes its own voting record on its own website, in its own format. We read two kinds of disclosure. Where a firm publishes a full downloadable record of its individual votes, we read every vote the same way as the US filings. Otherwise we use the firm's own published figure for the share of resolutions it voted with management. We include a firm only where that figure is at the level of resolutions, so it sits on the same axis as the others. Firms that publish only meeting level statistics, such as the share of meetings with at least one vote against, or only a search tool, are not placed on the chart, because that is a different measure. Each European firm shows the source and the year behind its figure.

The coverage rule

A category is shown for a firm only when it cast at least 25 decisive votes in that category in the season. Below that, a percentage would swing on a handful of votes, so we show not enough data instead. This is why a firm that discloses only one kind of vote appears in that category alone.

Comparing like with like

Firms are grouped into size bands and only ever compared within their own band and jurisdiction. Where a firm's assets under management are not public, its band is estimated from the breadth of its voting, so the bands are indicative rather than exact. The chart is a plain distribution: each dot is one firm, the box spans the middle half of the band, and the line is the median. The searched firm is the only point in color. We never publish a ranking of firms against each other.

Split votes

Where two funds in the same complex voted differently on the same resolution, we list it as a split vote. It is a factual observation from the filings, often reflecting different mandates or sub advisers, not a judgement.

One data quirk worth naming

On shareholder proposals, the recommendation field in filings is populated inconsistently, so we do not rely on it. Instead we use the standing convention that management opposes shareholder proposals: a vote for a shareholder proposal is counted as against management, and a vote against it as with management. The same convention lets us read European records that give the vote but not the recommendation.

What the record does not show

Form N-PX, and its European equivalents, show how a firm voted. They do not show why. They carry no engagement history, no rationale, and no private dialogue with the company. A vote with management can follow months of pressure, and a vote against can be a first step. The public record is a starting point, not the whole story. That is the gap the engage insights platform is built to close.