AGM Webcasting Platform Selection: A 2026 Buyer's Guide for Listed Companies

Most company secretaries pick an AGM webcasting platform on a feature-checkbox basis — registrar integration, live captioning, hybrid attendance, recording, ballot tooling — and a vendor that ticks every box wins the bake-off. Six months after the meeting, when external counsel responds to a regulator's information request or to a shareholder challenge filed under the local equivalent of UK Companies Act 2006 §311(2), the secretariat discovers the platform produced authentication logs the registrar can read but not the evidence file external counsel needs to sign off on. The platform streamed the meeting fine. The platform did not survive the review.

The 2026 buyer's question for AGM webcasting is not "does the platform stream video?" — every viable vendor clears that bar. The question is whether the platform survives an external counsel review six months after the meeting, against the strictest applicable disclosure and shareholder-rights regime, with a single archival artefact the secretariat can hand over without redaction or reconstruction. This guide is a procurement-defensibility exercise written for the company secretary, general counsel, or CFO running the RFP. It is the platform-side companion to the compliance-side companion for the company secretary and the production-side companion, and it mirrors the format we use in the parallel RFP framework for the webinar format.

What does an AGM webcasting platform need to do that a webinar platform cannot?

The default assumption in a procurement shortlist is that an AGM webcast is a webinar with a vote attached. That is the assumption that breaks the evidence file. A webinar platform is engineered for one-to-many broadcast with an audience the host curated. An AGM is a regulated shareholder proceeding where the platform must prove, after the fact, that every person who voted was entitled to vote, that every question put to the chair was visible to the chair before the resolution closed, and that the recording the secretariat archived is the same recording the audience saw — bit for bit, frame for frame.

Four capabilities separate an AGM platform from a webinar platform with extra features. First, registrar authentication — the platform must accept a credential issued by the registrar against the share register on the record date, not a public link or a self-service signup. Second, server-side ballot clock — the open and close of every resolution must be timestamped on the platform's server, not on the shareholder's browser, because a client-side timer can be manipulated and is not defensible. Third, immutable hash-chained event log — every authentication event, ballot open, vote cast, question received, and resolution close must be written to an append-only log where each entry's hash incorporates the previous entry's hash, so any tampering becomes mathematically detectable. Fourth, evidence-file generation as a first-class output — the platform must produce a single archival package the secretariat can hand to external counsel without manual reconstruction, not a bag of CSV exports and an mp4.

A platform that meets three of those four is not three-quarters of an AGM platform. It is a webinar platform with shareholder-meeting marketing. The evidence-file gap is the one that bites, and it bites only after the meeting is over.

Which jurisdictions does the platform need to satisfy out of the box?

The platform decision is jurisdictional before it is technical. A platform that satisfies German notice and identification requirements may not satisfy the Italian quorum and ballot record requirements, and a platform optimised for US Delaware practice may not even acknowledge the EU shareholder identification regime. The procurement question is which regimes the company is bound by — and the practical rule is to pick the strictest applicable regime as the binding spec, because every other regime is then satisfied as a side effect.

The five regimes a serious 2026 shortlist needs to test against:

United Kingdom — Companies Act 2006 §307A and §311. §307A sets the notice period and content requirements for a traded-company AGM. §311 governs the contents of the notice including statement of rights to appoint a proxy. The platform must support electronic notice dispatch with proof of receipt and a documented shareholder-question intake that reconciles to §319A's right of members to ask questions at a meeting of a traded company.

Italy — Codice Civile Article 2370bis governing telematic attendance and voting at the assembly. The platform must produce a meeting minute that the notary can certify, with attendance and vote records reconciled to the share register as of the record date.

Germany — Aktiengesetz §118a introduced under the Virtual Hauptversammlung legislation governing the virtual general meeting. The platform must support shareholder questions in advance of the meeting in line with §131(1a) and live questions if the bylaws permit, with the chair's response logged against the question.

European Union — Shareholder Rights Directive II (Directive (EU) 2017/828) as transposed into each member state, particularly the Article 3a shareholder identification regime that gives issuers the right to identify shareholders against intermediary chains. The platform must accept the SRD II identification feed and authenticate beneficial holders, not just nominee accounts.

United States — Delaware General Corporation Law §211 permitting virtual-only and hybrid meetings under §211(a)(2). The DGCL framing is permissive but the platform must still satisfy the corporation's bylaws and any SEC proxy rules engaged by the meeting type.

A platform that ticks "international" without naming the specific statutes and the specific platform features that satisfy them is a platform that has not been tested in the regimes it claims to cover. The RFP question is statute-by-statute: which feature satisfies which section number, and where in the platform documentation is that feature documented.

How should the secretariat evaluate the evidence file?

The evidence file is the single deliverable that decides whether the platform survives external review. The seven artefacts the company secretary signs off on, listed in the order an external counsel would expect to see them:

1. Notice of meeting plus proof of dispatch. The notice itself with the dispatch log timestamped against §307A notice periods.
2. Register reconciliation as of the record date. The list of authorised attenders cross-referenced to the share register and the SRD II identification feed.
3. Unedited broadcast recording. A single continuous master with no edits, in a format external counsel can open without bespoke tooling, with a cryptographic hash committed before archive.
4. Vote log with ballot hashes. Every resolution opened and closed, every vote cast, every change of vote permitted by the platform, each line entry signed against the chain.
5. Question log. Every question submitted, the timestamp of receipt, whether the chair took the question on the floor, and the executive response with link back to the broadcast timestamp.
6. Attendance roster. Continuous attendance log with join and leave events, reconciled to the registrar's authentication chain.
7. Certified minute with hashes. The signed minute, with the hashes of the underlying artefacts written into the minute so any later tampering is detectable by recomputing the chain.

If the platform cannot output all seven in a single archival package — same file structure every meeting, same hash format, same retrieval URL — it is not fit for purpose. A bag of CSV exports plus an mp4 plus a screenshot of the registrar dashboard is not an evidence file. It is the absence of an evidence file dressed as a deliverable. This mirrors the documentation rigour we apply to the regulatory artefacts for earnings call broadcasts, where Reg FD and MAR Article 17 demand the same single-archival-package discipline.

Which red flags appear in the procurement RFP response?

A platform's RFP response reveals more about its real architecture than its sales deck. Six specific failure modes that should disqualify or down-rank a vendor:

No documented authentication chain back to intermediary feeds. If the response says "we integrate with major registrars" without naming which registrar APIs, which credentials the platform issues, and how the credential is bound to a beneficial holder on the register, the platform is authenticating into a webinar room — it is not authenticating into a shareholder proceeding.

Ballot clock running client-side. If the resolution open and close timestamps are recorded against the shareholder's browser clock, the timestamps are unreliable and externally challengeable. The platform must run the clock on its own server, with NTP-synced UTC, and write the timestamp before acknowledging the vote.

Recording stored on a shared cloud bucket without WORM (Write Once Read Many) policy. If the recording can be overwritten, deleted, or re-encoded by anyone with platform admin access, the recording is not archival. WORM enforcement at the storage layer with a documented retention period satisfies the EU Market Abuse Regulation Article 17 five-year retention standard and analogous requirements elsewhere.

Captioning bolted onto the player not the encoder. If captions are rendered by a JavaScript layer on the playback page, the archived recording does not include the captions and the accessibility evidence does not survive replay. Captions must enter at the encoder so they are baked into the broadcast and the archive.

No jurisdictional fluency in references. If the vendor's reference list is six companies in one jurisdiction and the response cannot name the specific statutes those references' meetings were tested against, the vendor has not done meetings under the regimes binding the issuer. References must be jurisdiction-matched, not just industry-matched.

No documented response time for evidence-file extraction. If the vendor cannot commit in the contract to delivering the full evidence file within a specified business-day window from the meeting close, the vendor has not industrialised the extraction. An evidence file produced ad-hoc on request is an evidence file that will be late when external counsel needs it.

How should you stress-test a shortlisted vendor before contract?

Three tests, executed in order, before any contract is signed. They cost real money. They are worth real money.

Paid technical pre-AGM rehearsal with a synthetic registrar feed. The vendor runs a full rehearsal against a synthetic feed of fake shareholder records that mimics the issuer's actual registrar structure — share classes, joint holders, nominee accounts, beneficial owners identified under SRD II. The rehearsal exercises authentication, ballot, and question intake under load. The output: a sample evidence file the secretariat can review before the live meeting.

Evidence-file dry-run reviewed by external counsel. The rehearsal evidence file goes to external counsel for sign-off before the live event. Counsel's review identifies whether the seven artefacts are present, in the expected format, with the expected hashes — and identifies the gap before the gap matters. This is the test most procurement processes skip and most post-meeting disputes regret.

Failover drill on the live network the day of dress rehearsal. The vendor demonstrates encoder failover, network-path failover, and registrar-feed failover on the actual network configuration the issuer will use on the day. A failover that works in the vendor's lab and fails on the issuer's network is the worst-case scenario for an AGM. Test it before the day, on the network of the day.

Pricing transparency: what should the procurement quote include?

A quote that prices "AGM platform — €X per meeting" is a quote that will surface €Y of unbudgeted line items in the final invoice. The procurement specification should require itemisation against the following lines:

• Platform licence per meeting, with a defined attendance cap and overage rate.
• Registrar integration setup, per registrar, with renewal or re-test fee if the registrar changes API version.
• Captioning provider, named, with the per-hour rate and the SLA for accuracy and reference-dictionary loading.
• Archive retention, with the per-year storage rate for the full retention window — typically five years to satisfy MAR Article 17 — and a defined extraction fee for retrievals within the window.
• Evidence-file extraction SLA, with the contractual business-day window and the penalty for missed delivery.
• Post-meeting forensic-support hour rate, for the inevitable counsel request months after the meeting. Forensic support priced at "case-by-case" is forensic support that will be expensive at the worst possible time.

A vendor that responds to the itemisation request with a single bundled number is a vendor whose internal cost structure does not match the deliverables. That is a procurement risk before it is a budget risk.

Vendor scorecard: 14 criteria the company secretary should rank before contract

The shortlist evaluation ranks each vendor against fourteen criteria, weighted by the secretariat's risk appetite and the issuer's jurisdictional exposure. The criteria — drawn from the architecture questions above — group into four families.

Authentication and identity (criteria 1–3): registrar API integration depth; SRD II beneficial-holder identification; credential-binding to share register on record date.

Proceedings integrity (criteria 4–7): server-side ballot clock; hash-chained event log; question intake with chair-visibility audit; live caption chain at encoder.

Archive and evidence (criteria 8–10): WORM enforcement; seven-artefact evidence-file generation as default output; documented extraction SLA in business days.

Procurement and risk (criteria 11–14): jurisdictional reference fluency; itemised pricing transparency; post-meeting forensic-support rate; contractual liability cap proportionate to the meeting's regulatory exposure.

A vendor scoring above threshold on all fourteen is rare. A vendor scoring above threshold on the proceedings-integrity and archive families, even with weakness in pricing transparency, is workable with a sharp contract. A vendor failing on the authentication family is a vendor for a different meeting type, not an AGM. The same scorecard logic applies to the cross-IR-format pattern we use for investor day broadcasts, where the artefact-defensibility test runs in parallel even when the regulatory exposure differs.

When does the buy-vs-bundle decision favor unified vendor over best-of-breed stack?

The best-of-breed argument is intuitive: pick the strongest webcasting vendor, the strongest registrar-integration partner, the strongest captioning provider, and the strongest archive provider, and assemble a stack that beats any single-vendor offering on every individual capability. The argument is sound on every layer except the layer that matters most for an AGM — the evidence file.

The evidence file is the artefact where the multi-vendor stack breaks. Every vendor in the stack produces its own logs, its own timestamps, its own hash chains if it produces any at all, and the secretariat is left to reconcile them into a single artefact external counsel can sign off on. The reconciliation is manual, expensive, and brittle. A vote logged in the registrar's timezone, a question logged in the captioning provider's UTC offset, a recording timestamped against the encoder's clock — three timestamps for one event, none of them authoritative against the others, all of them externally challengeable.

A unified vendor — one platform owning authentication, proceedings, captioning, archive, and evidence-file extraction end-to-end — produces a single chain with a single clock and a single signature. That is the artefact external counsel signs off on without question. The best-of-breed stack can match this only if the issuer is willing to commission and pay for a dedicated integration layer that reconciles the vendors' outputs into a unified evidence file, with the integration layer itself subject to the same WORM, hash-chain, and SLA discipline. Few issuers commission this. Most discover at the post-mortem that they should have.

The unified-vendor decision is not always correct. For a single-jurisdiction issuer running a low-contestation AGM with a long-standing registrar relationship, the best-of-breed stack is defensible and often cheaper. For a multi-jurisdiction issuer running a contested resolution under SRD II identification with activist shareholders on the register, the unified vendor is the procurement default and the burden of proof should fall on a best-of-breed alternative.

Ready to scope your AGM webcasting platform RFP?

The platform selection is one of the few procurement decisions where the wrong choice does not surface until six months later, in an artefact-review by people who were not in the room when the contract was signed. The companies that get this right run the RFP as a procurement-defensibility exercise — strictest applicable regime as the binding spec, seven-artefact evidence file as the non-negotiable output, jurisdictional reference fluency as the credibility test. If you are scoping an AGM platform RFP for the 2026 or 2027 meeting cycle, our virtual event production services cover the platform-and-production scope end-to-end, and our team can walk you through the scorecard against your shortlist.