A conversation does more than read the document aloud.
Companies already create the information their people need. The challenge is helping employees find the time and attention to take it in.
Across the Table turns that material into focused conversations between a host and an expert. The format is informed by established research into how people process explanations, questions and information over time.
The research does not prove that every internal podcast will outperform every document. It does explain why the format is designed the way it is.
The second voice has a job.
Each episode uses two presenters: a host and an expert.
The host represents the listener. They ask the question an employee might genuinely be thinking, test what an answer means and draw out the useful detail.
The expert explains the source material clearly and accurately.
The important distinction is that this is not a document divided between two voices.
Research into instructional dialogue suggests that deeper learning can come from the questions within a conversation, particularly questions that prompt reasoning and explanation. Studies comparing dialogue with monologue indicate that the questioning mechanism matters more than simply hearing two different voices.
Evidence
Two voices are not the point.
Good questions are.
Plain language helps people into the material.
People often process explanations more effectively when language feels conversational rather than formal and impersonal.
That does not mean making serious material casual. It means explaining it as one person would explain it carefully to another.
Across the Table therefore writes episodes as real exchanges rather than turning documents into narrated summaries.
Document language
“Employees are required to familiarise themselves with the revised operating procedure.”
Conversation
“So what actually changes for someone doing the job?”
An illustrative design example, not research evidence or client material.
Evidence note
Research synthesised in Richard Mayer’s work on multimedia learning supports the personalisation principle.
One idea at a time.
There is no scientifically established “perfect” podcast length.
What the evidence supports more strongly is segmentation: breaking complex information into manageable units and giving people control over how they move through it.
So Across the Table focuses each conversation around a clear message. Longer source material can become a connected series rather than one overloaded episode.
Evidence note
What changed?
Why does it matter?
What happens next?
What does it mean for you?
An illustrative example. The number of episodes depends on the material.
Not shorter for the sake of being short.
Focused enough to follow.
Important messages rarely need saying only once.
Learning research consistently finds benefits when related information is revisited over time rather than delivered in one concentrated block.
That gives serialised communication a useful rationale. Instead of asking one piece of content to carry everything, a series can introduce an idea, return to it later and connect it with the next message.
Across the Table can therefore turn recurring or longer communications into a connected sequence rather than isolated audio files.
Introduce the message
Return to it in context
Connect it to what comes next
Evidence note
Audio does not have to replace the document.
Some people will want the full document. Others may be more likely to listen while travelling, walking or moving between meetings.
Audio gives employees another route into the same information. That is especially relevant where reading a long document is inconvenient or where people are away from a desk.
The original source remains available for detail. A transcript can sit alongside the episode for people who prefer or need text.
Listen
The episode, wherever they are.
Read
The original document and the transcript.
Revisit
Come back to it when it becomes relevant.
Evidence note
The format is AI-produced. The editorial decisions are not automatic.
Across the Table does not upload a company document and publish whatever an AI system produces.
The source material is read and structured first. The conversation is written around what the audience actually needs to understand. The finished episode is checked against the original material. The client approves the content before it goes out.
AI-generated voices are part of the production process, but they do not replace the need for editorial judgement, source checking and approval.
Original material
Human editorial work
Produced conversation
Client approval
What we can say with confidence
And what the research does not settle.
The evidence supports
The evidence does not establish
The research behind the format
Grouped by the part of the design each one informs. Open any group to see the references.
Dialogue and deep questions
Conversational language and segmentation
Spacing
Reading and listening
A useful counterpoint
Full evidence notes are available on request. Ask us for the reference list
The principle is simple.
Start with information the company has already created.
Find the message people actually need to understand.
Ask the questions that bring that message into the open.
Explain it clearly.
Then give people a convenient way to listen, revisit and follow the story.
Not a document read aloud. A conversation designed around understanding.