What Delivery Can Prove - and What It Cannot

RESEARCH QUESTION

What can evidence about delivery legitimately establish, and where does it stop supporting claims about reception or later outcomes?

ABSTRACT

Communication often leaves a strong record on the sending side. A script can be compared with its revision. A recording can be replayed. A presentation can be timed. A released file can be retrieved. Those records can support precise claims about what was made, changed and delivered.

The problem begins when evidence about delivery is allowed to carry a larger inference. A changed pause does not by itself establish a changed interpretation. A more fluent performance does not by itself establish greater understanding. A different outcome after a different delivery does not, without stronger evidence, identify delivery as the cause. This note separates what a sender-side record can prove from what it can only make plausible.

KEY FINDINGS

1. Delivery evidence can be very strong about the delivery itself. Recordings, scripts, files and logs can establish observable properties and changes on the sending side.

2. Inspectability is not the same as causal importance. The fact that one part of the communication journey is easier to examine does not mean it explains what happened later.

3. Larger claims require additional evidence. Moving from "this changed" to "this caused a receiver-side outcome" adds an inference that the sender-side record alone cannot settle.

EVIDENCE AND ANALYSIS

Delivery leaves a record

Some parts of communication are unusually inspectable. If two recordings exist, timing, wording, pauses, emphasis and other observable properties can be compared. If two drafts exist, changes in content and sequence can be identified. If a message was distributed through a system, the system may establish that a file was released or made available.

That evidence matters. It allows claims that would otherwise depend on memory or impression to become testable against an object. The bounded claim is straightforward: a sender-side record can establish what was present in that record and how one version differed from another.

The boundary appears at the next step. The same recording does not contain the receiver's interpretation. The same file does not contain what someone retained a week later. A delivery record can be complete for one question and silent about another.

Inspectability is not causal proof

This creates an evidentiary asymmetry. Communication work often preserves detailed evidence about production and delivery while leaving much thinner evidence about what happened after exposure.

That asymmetry can pull explanation toward what is easiest to inspect. If a speaker changed the pacing and the meeting later went better, the pacing change is visible. If a revised email produced fewer questions, the revised wording is visible. If a manager used a new script and the team later complied, the script is visible.

But visibility does not identify cause. Methodological work on validity and causal inference makes the same general point in more formal settings: claims depend on the inferences used to move from observations to conclusions, and more ambitious claims require stronger support.[1,2] The communication application is bounded. Evidence that a delivery changed can support a claim that the delivery changed. A claim that this change caused a later receiver outcome needs additional evidence about that causal link.

The comparison problem

A common source of overreach is the before-and-after comparison.

Imagine two presentations. In the second, the speaker leaves more space, changes the opening and shortens the explanation. The second meeting also ends with a better commercial outcome. That is useful information. It is not a controlled comparison.

The audience may differ. The prior relationship may differ. The decision process may differ. Timing, budget, expectations and competing events may differ. Even when the communication itself is the most visible difference to the sender, it may not be the only relevant difference in the system.

This is why a successful second outcome does not retrospectively prove the explanation for the first. It can increase confidence in a hypothesis. It does not, by itself, isolate the mechanism.

The evidence asymmetry

The sending side often has an archive. The receiving side often has traces.

A recording may preserve the complete audio event. A receiver-side record may consist of one comment, one survey item, one action, one silence or one later recollection. The richer record can therefore dominate the explanation simply because it supports more detailed description.

That is an analytical risk. Detail and explanatory reach are not the same property.

COMMODYN's earlier Research Note What Is Landing? makes the complementary receiver-side point: comprehension, interpretation, retention, expression, action and onward transmission should not be collapsed into one outcome.[3] This note adds the sender-side counterpart. Evidence can be precise and still be precise about the wrong part of the claim.

What a stronger claim would need

There is no single required method for every communication question. The evidence needed depends on the claim.

If the claim is "the wording changed", version comparison may be enough.

If the claim is "the audience heard version B rather than version A", distribution or exposure evidence is needed.

If the claim is "version B was understood differently", receiver-side evidence is needed.

If the claim is "the delivery change caused the difference", the claim becomes causal and the evidentiary burden rises again.

The practical discipline is not to distrust sender-side evidence. It is to stop extending it after it has answered the question it can answer.

WHAT THIS NOTE ESTABLISHES

Evidence level: methodological synthesis and analytic application to communication; no original empirical data.

The note supports three bounded propositions. Sender-side records can provide strong evidence about what was produced and delivered. Their inspectability does not establish that the observed sender-side feature was the most important cause of a later outcome. Claims about receiver-side effects or causation require evidence that reaches beyond the delivery record itself.

CLAIM BOUNDARY

This Research Note does not claim that delivery is unimportant, that sender-side evidence is weak, or that delivery effects cannot be studied.

It does not claim that a later outcome is unrelated to delivery merely because other explanations are possible.

Its boundary is narrower: evidence about delivery establishes delivery-level facts. Moving from those facts to claims about reception or causation requires additional inference and, where the claim matters, additional evidence.

SOURCES & NOTES

1. Kane, M. T. (2013). Validating the Interpretations and Uses of Test Scores. Journal of Educational Measurement, 50(1), 1-73. DOI: 10.1111/jedm.12000. Kane's argument-based approach is used here as methodological prior art for the principle that the validity of a conclusion depends on how well the evidence supports the inferences required to reach it. This is not treated as direct empirical evidence about organisational communication.

2. Shadish, W. R., Cook, T. D., & Campbell, D. T. (2002). Experimental and Quasi-Experimental Designs for Generalized Causal Inference. Boston: Houghton Mifflin. Used for the bounded methodological distinction between observing a difference and establishing a causal explanation for that difference.

3. van Poelgeest, P. (2026). What Is Landing? COMMODYN Research Note RN-01, version 1.0. Used for the established COMMODYN distinction between different receiver-side outcomes and the rule that evidence for one outcome should not silently become evidence for another.

4. van Poelgeest, P. (2026, 24 August). What Delivery Can Prove - and What It Cannot. LinkedIn article. This is the original public precursor of RN-03 and is included for provenance, not as independent evidence for the claims in this note.

PUBLICATION INFORMATION

AUTHOR Peter van Poelgeest

PUBLISHER COMMODYN Research

EVIDENCE TYPE Methodological synthesis and analytic application; no original empirical data

VERSION 1.0

ORIGINAL PUBLIC PROVENANCE 24 August 2026

COMPETING INTERESTS No competing interests declared.

AI ASSISTANCE AI tools were used for research support, editorial development and document production. Source selection, interpretation, claim boundaries and publication decisions remain the author's responsibility.

RECOMMENDED CITATION van Poelgeest, P. (2026). What Delivery Can Prove - and What It Cannot. COMMODYN Research Note RN-03, version 1.0.