Skip to content

The Change Desk by Broadcastwell

Your facts changed.
Did the AI answers?

After an incident statement, an acquisition or ownership change, or a price, plan, product or link change, we check what ChatGPT, Claude, Perplexity, Google AI Overviews and Google AI Mode tell your buyers. The $490 Record Check delivers findings within 48 hours of scope confirmation, with dated receipts for every answer and human review.

One company. One change event. Ten questions, five engines, three runs: 150 planned answers. Valid answers and exclusions are shown separately.

A clear public fact deserves a clear answer.

Buyer paths

A new price or next step

Your pricing page moves, a plan changes or a trial ends. We check the answer and the destination a buyer receives against the current official page.

Transaction state

An agreement or a completed deal

A company announces new ownership. We check whether answers preserve the announced status and what the public record says about products and support.

Published statements

A carefully qualified update

A statement describes one service, one organization and a dated finding. We check whether answers retain those boundaries and uncertainty.

From the changed fact to a usable record.

  1. Confirm the scopeWe agree one company, one change event and ten questions, with dated public sources.
  2. Keep the receiptsWe record three runs on each of the five engines, with complete answers, dates and cited sources.
  3. Review the distinctionsWe compare answers with the confirmed facts, separate errors from unknowns and review the findings.
  4. Deliver the next stepsWithin 48 hours of scope confirmation, you receive the evidence and a Correction List or clarification draft.

Evidence your team can work from.

  • Dated receipts for the answers we captured.
  • A source map showing the pages the answers cite.
  • Findings compared with the public record, including correct and unresolved cases.
  • A Correction List or clarification draft grounded in the evidence.
  • Visible planned and valid counts, with every exclusion explained.
  • A written report for an asynchronous decision.

Three studies, with review before findings.

Each study starts with a 30-answer preflight. Publication depends on its evidence gate and review. These cards will link to the study records when they are published.

In review

Where AI sends field service software buyers

Official pricing and demo destinations.

In review

Software transaction state

The distinction between announced, completed and continuing.

In review

ConnectWise Cloud Backup statements

What answers preserve, omit or confuse in dated public statements.

Study findings are not available yet. The statement study reports published statements and AI answers. Readers should use the official provider advisory for operational status.

Dated questions. Inspectable evidence.

We preserve the source bundle for each capture window and judge answers against the facts available then. A missing answer is separate from an error. We publish denominators, exclusions, correct cases and reviewed findings, with open answer tables under CC BY 4.0. Two reviewers confirm each adverse public finding.

Read the methodology

Corrections to our records are free, whether or not anyone buys. Write to hello@broadcastwell.com with the study, receipt and public source.

One company and one change event.

$490 once

Ten questions on five engines, three runs each: 150 planned answers. Findings within 48 hours of scope confirmation.

Refundable in full, no questions asked, if requested within 30 days of delivery.

The Record records what the engines stated on stated dates and promises no correction, no position, no leads and no revenue. Read the full terms.

For agencies

White-label Record Check: $490 per client report, for one company and one change event. Delivered to your agency under your brand. We never contact your client.

Before you order.

What counts as one change event?

One incident, an acquisition or ownership change, or a price, plan, product or link change for one company. We confirm the scope and dated public sources before the 48-hour delivery window starts.

Which engines do you check?

ChatGPT, Claude, Perplexity, Google AI Overviews and Google AI Mode. Ten questions, three runs on each engine, make 150 planned answers. We report valid answers and exclusions separately.

What do we receive?

Dated answer receipts, a map of cited sources, findings checked against your public facts, and a Correction List or clarification draft. Where the evidence cannot settle a statement, we say so.

When will we receive the findings?

Within 48 hours of scope confirmation. We agree one company, one change event and the public facts to check before that clock starts.

What is the refund policy?

Refundable in full, no questions asked, if requested within 30 days of delivery.

Can you guarantee an engine will change its answer?

We provide the evidence and proposed corrections. We do not control AI engines or promise a correction, position, inclusion, leads or revenue.

Can an agency buy it for a client?

Yes. A White-label Record Check is $490 per client report for one company and one change event, delivered to the agency under its brand. We never contact your client.

How can someone correct a study?

Write to hello@broadcastwell.com with the study, receipt and public source. Corrections to our records are free whether or not anyone buys a Record Check.

Put the evidence to work

See which buyer questions leave you out, the sources visible in those answers, and the three changes worth testing first. The Category Audit covers ten buyer questions on five engines, three runs each. AI answers vary, so repeated observations and disagreements are reported.

Start with the $490 Category Audit

Findings within 48 hours of your category confirmation. Refundable in full within 30 days of delivery.

One time price. No annual contract. All five engines included. The $490 credits once against the $2,900 Fix Sprint within 30 days of delivery, so the Sprint is $2,410.

Every figure comes from the published method, v1.1.