Leadership1 distinct publisher3 min readUpdated
DragonGC's third annual count found 14 Fortune 1000 companies below 80% support, down from 25. What they disclosed next reads like a project report, not an apology.
The Board Room · Leadership desk

Compiled by The Board RoomSomething wrong?How this is made
Follow any of these and your For You feed starts watching them — no settings page required.
DragonGC has published its third annual accounting of how companies with adverse say-on-pay votes responded the following season, in a memorandum by Nicholas Sasso, Erin Conlan, Jennifer Dorney, Neil McCarthy, Sophia Ojjeh and Leo Tadikonda [1]. The interesting material is not the rebuke but the response, which now arrives in proxies as a work log: investors invited, investors met, percentage of shares covered, and which directors were in the room [13][15].
The population is small and shrinking. Reviewing Fortune 1000 say-on-pay results for the 2025-2026 annual meeting season, DragonGC identified 14 companies with support below 80%, against 25 in 2024-2025 and 21 in 2023-2024 [6], which is 44% fewer than the prior season [1]. Nearly all 14 had an announced engagement program, some more detailed than others [7]. All 14 recorded an improved vote at their 2026 annual meeting, with increases ranging from 3.9% to 66.1%, on results DragonGC calculated from for and against votes only [8][9]. Near-universal improvement makes the engagement program table stakes rather than a differentiator.
The benchmark, then, is what the visible programs contained. PENN Entertainment invited 17 shareholders representing about 48% of outstanding shares and met nine representing about 36% [10], covering roughly three quarters of the share base it approached [4]. Independent board members attended 100% of the meetings held with investors among PENN's top 30 holders [10]. UnitedHealth Group contacted 46 shareholders representing about 60% of shares and held discussions with 21 representing about 51%, including all 12 shareholders it identified as having voted against the proposal, with independent directors in every discussion [11]. By share weight that is about 85% of what it set out to reach, even though it met fewer than half the investors it contacted [2]. GE Aerospace reached out to holders of about 54% of shares and engaged holders of about 48%, with independent directors engaging holders of about 29% directly and the then Compensation Committee Chair leading or joining many meetings according to shareholder preference [12] - director-level contact covering roughly 60% of the engaged base [3].
The other pattern is timing. Intel, Goldman Sachs and Live Nation Entertainment ran ongoing or multi-stage processes instead of a single post-vote push, with Intel cycling through results review, off-season outreach, feedback incorporation and in-season discussion, and Goldman starting compensation and governance engagement before its 2025 annual meeting [14]. That is a standing calendar item, not a fire drill.
Why any of this happens is worth restating plainly. The vote is a non-binding advisory vote required by SEC rules over the Item 402 disclosures, including the CD&A and compensation tables [2], and most years most companies clear 80% support [3]. Adverse results are typically driven by an against recommendation from ISS, Glass Lewis or large institutional investors applying their own compensation policies, which is the enforcement mechanism the statute does not supply [4]. Companies on the receiving end nearly always respond with an engagement program the next season [5].
Two things to watch. First, whether the count keeps falling; a benchmark drawn from 14 companies is thin, and a further drop would say more about issuers pre-clearing pay design with proxy advisers than about engagement quality [6]. Second, whether director attendance rates become a standard disclosure line, since PENN's 100% figure and UnitedHealth's every-discussion claim set a bar that a comp committee chair cannot meet by delegating to investor relations [10][11]. The material here documents outreach and describes it as informing board consideration of responsive actions [15]; the pay changes those meetings produce are the half that decides the following vote.
Ranked by verification strength, evidence, and original report placement.
Adverse say-on-pay outcomes are typically driven by an adverse voting recommendation from one or more of ISS, Glass Lewis and large institutional investors for violating their executive compensation voting policies; while SEC rules require only a non-binding advisory vote, in practice these entities provide an enforcement mechanism.
All 14 companies had an improved say-on-pay vote at their 2026 AGM, with increases ranging from 3.9% to 66.1%.
DragonGC published its third annual report on how companies with adverse say-on-pay votes responded with shareholder engagement programs the following season; the post is based on a DragonGC memorandum by Nicholas Sasso (Product Specialist), Erin Conlan (Legal Analyst), Jennifer Dorney (Head of Marketing), Neil McCarthy, Sophia Ojjeh and Leo Tadikonda.
SEC rules require public companies to hold a separate shareholder advisory vote to approve executive compensation, covering compensation disclosed per S-K Item 402 including the CD&A, the compensation tables and other narrative executive compensation disclosures.
Most years for most companies the say-on-pay vote passes with greater than 80% support from shareholders who vote on the matter; sometimes the approval rate is less than 80%, and sometimes the resolution receives less than a majority and fails.
Companies that have received an adverse say-on-pay vote nearly always respond with a shareholder engagement program during the following season.
Evidence-backed comparisons of source perspectives and observed adoption signals. Read the methodology
Which Builder, Operator, and Investor concerns the observed source mix emphasized—not a truth score.
Evidence, demonstrated adoption, hype gap, incentives, and confidence are assessed independently, each on its own current evidence. How these are measured.
Single-source vendor analysis of public filings
The factual spine - season counts, vote improvements, and company-level outreach statistics - is drawn from SEC filings and 2026 proxy statements, which is verifiable primary material, and the methodology (Fortune 1000 universe, for/against basis, named sources) is stated. But the cluster has exactly one item, the underlying results table is not reproduced in the supplied text, company figures are issuer self-disclosure, and no second publisher or dataset corroborates the counts.
Engagement response is near-universal among affected filers
Adoption of the practice itself is well documented at the level of the affected population: nearly all 14 sub-80% companies disclosed an engagement program in their 2026 proxies, and named issuers quantified outreach reaching 36-51% of outstanding shares with independent directors present. It is bounded, however, to 14 Fortune 1000 companies in one season as counted by one vendor, so it evidences a concentrated practice rather than broad measured uptake.
Mild overstatement via implied causation
The writing is restrained and numeric rather than promotional, but the juxtaposition of near-universal engagement programs with universal vote improvement invites a causal read the data cannot support: no counterfactual, no companies-that-did-not-engage comparison, and no analysis separating outreach from actual pay-plan changes or shifting proxy adviser recommendations. The falling sub-80% count is likewise presented without testing alternative explanations.
Vendor research with marketing authorship
The analysis is produced by DragonGC, a governance-services firm, and the author list includes its Head of Marketing alongside a Product Specialist and a Legal Analyst - a structure typical of thought-leadership content that supports a commercial governance offering. It is now an annual franchise, which further ties publication to brand visibility. The supplied source discloses roles but no commercial relationship to the companies analyzed, so the incentive is presentational rather than evidenced as conflicted.
Moderate-low
The discrete facts are likely accurate because they are traceable to proxy statements and vote filings, but the cluster rests on a single vendor item with no cross-publisher corroboration, no reproduced results table, and an unresolved causation question at the center of the story. Confidence in the numbers is higher than confidence in the implication.
invest
The SEC stopped grading exclusions, so issuers now mark their own proxy ballots1 distinct publisher
leadership
Boards Prune ESG Pay Labels But Keep the Specifics, Reshaping the Next Incentive Cycle1 distinct publisher
build
The chokepoint moved: ABF film, not lithography, now caps China's accelerator output1 distinct publisher
invest
Half a book in two memory names: what Situational Awareness's 67% month actually shows1 distinct publisher
Distinct publishers with included, body-backed reporting in this cluster.
1 article · August 17, 2026