Our impact framework contains a section listing what the document does not establish. It was the hardest page to write and it is the one we would keep if we had to lose the rest.
That reach figures are exposure, not outcome. That satisfaction scores are perceived value from attendees only, with no third party validation. That speaker progression into external advisory roles is only partially measured and depends on partner confirmation. That we evidence contribution to the early links in our causal chain and do not claim sole causation of the later ones.
None of that is required of us. Every one of those statements gives a sceptical reader something to hold against us.
Because the alternative is a document that cannot be checked, and a document that cannot be checked is marketing.
There is a version of impact reporting in this sector that survives only because nobody reads it closely. It uses layered numbers, mixes exposures with outcomes, and describes activity in language that implies result. It works, in the narrow sense that it satisfies reporting requirements. It also makes the whole category of impact claims less believable, including the true ones.
Stating limitations changes what you can then assert. Once we had written down that our satisfaction data is attendee self-report, we could no longer describe it as evidence that the conference changes behaviour. That forced a different question: what would evidence of behaviour change actually look like, and are we willing to collect it.
We are not there yet. Follow-up measurement at six and twelve months is the gap in our evaluation and we have said so publicly rather than filling it with proxies.
If you publish impact material, the exercise is simple and uncomfortable. Take each number and write beneath it the sentence a hostile but fair reader would write. If you cannot live with that sentence in public, the number is doing more work than it can bear.
Then publish the sentence.