What I Learned Implementing VOICE on a Real Engagement Book

Most of what has been written about VOICE so far explains what it says. This note is different. Over the past weeks I implemented the method in software, end to end, and ran it against a realistic book of engagement objectives — this is what breaks, what holds, and what it means for any team planning to adopt it.

On June 8, 2026, the FIR published VOICE, the first structured method for evaluating the effectiveness and the influence of shareholder and bondholder engagement. It came out of more than a year of work coordinated by the FIR's Commission Dialogue et Engagement, with a pilot group initiated by FRR and ERAFP and supported by the PRI, bringing together some fifteen investors and academics.

The method in one paragraph

VOICE makes one distinction and builds everything on it. Effectiveness asks whether your engagement activities took place and reached their milestones. Influence asks whether the company's behavior actually shifted in a way attributable to you. The method grades influence on a five-level scale, and each level is harder to earn than the last. Assessment happens when an objective closes, and the standard is causal contribution, not correlation.

Lesson one — the hard part is not the scale, it's the definition

VOICE defines engagement as iterative, intentional interaction cycles based on specific objectives, aimed at influencing a company's transparency, operations or strategy. That definition deliberately excludes occasional dialogue whose purpose is information gathering.

Apply that definition honestly to a typical engagement log and a large share of the lines fail it. The meeting where you asked questions about the climate plan and noted the answers is dialogue, not engagement — no formalized request, no objective, no timeframe. Teams that adopt VOICE will discover that their first task is not scoring influence; it is deciding, line by line, what actually counts as an engagement. Expect the count in your next report to go down before its quality goes up — and expect to need a sentence explaining why.

Lesson two — it's scored at closure, but the evidence is captured or lost months earlier

The method assesses influence when an objective closes. Every team I have worked with over fifteen years closes objectives the same way: in December, from memory, from mail archives, in a hurry.

That works, barely, for activity reporting. It cannot work for VOICE, because the upper levels of the scale depend on evidence that exists only at the moment things happen. The analyst who hears a company representative say "your letter came up in our board discussion" holds a level 5 signal in her hands. If it is not recorded that day, with a date and a context, it is gone. Retrospective scoring collapses everything to level 2 or 3 — not because there was no influence, but because nobody can prove there was.

The single highest-impact change a team can make is to move evidence capture to the moment of the interaction. Everything else in the method depends on it.

Lesson three — level 4 is an evidence-management problem, and partly not your own evidence

Level 4 requires convergent bundles of evidence — what the French text calls faisceaux d'indices. In practice that means assembling several independent signals that point the same way: the timing of the company's change against your asks, the specificity of what changed against what you requested, statements made in meetings, and — critically — the context of what everyone else was doing.

That last element deserves attention. Situating your own contribution requires knowing which other investors were engaging the same company on the same theme at the same time. That information exists — scattered across published engagement reports, coalition disclosures and stewardship statements — but no team maintains it internally, because it is a market-data problem rather than a record-keeping one. Any serious VOICE implementation has to answer the question: against what background of convergent engagement am I claiming my own influence? Teams should decide early how they will source that context.

Lesson four — level 5 changes analyst behavior, and that is the point

Company acknowledgement sounds rare, and publicly it is. But acknowledgement in the method can be private, and private acknowledgement happens more often than teams realize: in the debrief after an AGM, in a side remark on a call, in an email thanking you for the framework you suggested. Once analysts know that a recorded acknowledgement is the difference between level 4 and level 5, they start hearing them. The method does not just measure the work — it changes how attentively the work is documented, which is precisely what a good measurement framework should do.

Lesson five — closure discipline is the hidden prerequisite

Scoring at closure assumes objectives close. Most engagement books I have seen contain objectives that are years old — neither achieved nor abandoned, kept open because closing them feels like admitting defeat. VOICE quietly forces the issue: an objective that never closes is never scored, and an honest level 1 or 2 at closure is more useful than a permanent maybe. Teams adopting the method should set a closure convention — a maximum age, or an annual review at which every objective is closed, escalated or explicitly renewed — before they score anything.

Where this leaves a stewardship team in 2026

The regulatory context makes this more than methodology. In July 2025 the AMF published the summary of its SPOT inspections on voting and engagement policies, and its central criticism was that engagement reports describe activity without analyzing outcomes. VOICE is the profession's most serious answer to that criticism to date. But the method only works on top of a record that captures evidence continuously, situates it against the market, and closes objectives with discipline.

And if you'd rather start on your own, I've published a free scoring template that lets a team grade its first objectives on the VOICE scale in a spreadsheet. It will carry you to your first honest scores. It will also show you — around objective thirty — exactly why this eventually needs a system.

Written by Amine Gharby, founder of engage insights (amine@engageinsights.io). Provided for information only, not as legal, voting, or investment advice.