.jpg)
If you've been following the proxy-voting market this year, you’ll have watched the ground shift. J.P. Morgan has dropped external proxy advisors for its US voting and moved the decisions in-house, using its own AI tool. A senior SEC official has publicly backed AI's role in proxy voting. Rules-based voting is becoming widely available, and the key question for stewardship teams is how well it holds up under scrutiny.
Speak to any stewardship team that has been doing rules-based voting, and they’ll tell you the most demanding work is justifying those rules:
- Defending a vote against management to a compliance or risk team that wants more than a line-item output.
- Deep-diving on a contentious holding before its AGM, when the rule-based output is "refer" or the rationale has to stand up to scrutiny.
- Building a new rule from scratch when the policy needs to evolve and you don't yet know where the line should sit.
- Answering a client question, "why did we vote that way?", with something more substantial than "the policy said so".
Most rules engines can’t help with this; yet it's also where client and regulatory scrutiny homes in.
Custom research reports, built for the work that follows the rule
This is what ProxyBeacon's custom research reports are for. Every report is cited to source and built around the specific question you're trying to answer rather than a generic template.
One request we hear more and more is peer comparison. Often the only way to judge a company's behaviour is to see it next to a peer. Is Chevron's CO2-linked variable pay genuinely calibrated, or cosmetic next to Exxon's? Are their low-carbon capex disclosures comparable, or do they describe very different commitments?
ProxyBeacon produces structured side-by-side information on whatever you need to compare, whether that's climate KPIs, executive pay calibration, low-carbon capex or board composition, with every data point cited and explicit flags where a company hasn't disclosed.
Every claim links back to the underlying filing, and every gap is acknowledged rather than glossed over, so the report stands up when someone checks it.
See it in practice
If you'd like to see a worked example, please get in touch. We don't publish the demos publicly, but we're happy to talk you through one privately.
