Leave the contribution on the screen

TL;DR: I would spend the final slide on the contribution I want the audience to examine. Patrick Winston’s How to Speak is worth revisiting before an engineering review, research presentation or technical interview. Its closing-slide advice gives you a concrete edit to make immediately.

MIT identifies this recording as Winston’s January IAP 2018 talk. In the closing section, he argues for leaving the contribution visible while people ask questions and leave. His distinction is between generic conclusions and an account of what the speaker actually accomplished.

That is the part I would turn into a preparation rule: the final slide should help someone reconstruct what changed because the work was done.

A farewell slide can be polite. It gives the audience little to inspect. A contribution slide can keep the central claim, evidence and remaining boundary available during the most useful part of the conversation.

Build the closing slide before cutting the middle

Consider an illustrative presentation about improving a document search system. A weak ending might list the technologies used, say that the prototype was successful and invite questions. The audience still has to remember which problem was solved and what “successful” meant.

I would instead leave three compact elements visible: the search failure the work addressed, the evidence that the revised approach helped, and the important cases it still could not handle. Any numbers on that slide would have to come from the actual evaluation. The example does not need invented improvements to show the structure.

That ending also changes how I would edit the rest of the talk. If the contribution depends on handling ambiguous queries, the audience needs to see an ambiguous query and understand the comparison. Several slides about implementation details may matter less than that one example.

Writing the ending early creates a useful constraint. Every major section should help the audience assess the contribution or understand its limitation. Material that does neither has to earn its time another way.

Leave a question the audience can answer

For a review that requires a decision, I would place the requested next step beside the contribution: extend the experiment, fund a pilot or resolve a particular uncertainty. That is my application of the lesson, rather than a claim that every Winston talk requires a business approval.

Then I would rehearse the ending with a colleague who has not seen the deck. Ask them what they think the work established and what remains open. Their answer is more informative than whether they remember the exact title of the last slide.

The limitation is that a presentation heuristic cannot rescue weak work or missing evidence. Different settings also call for different endings; a teaching session, a research defence and an incident review have different responsibilities. A strong closing slide should serve those responsibilities.

What transfers is the respect for the audience’s attention. The final visible surface can do useful work even after the speaker stops talking.

Leave the audience looking at what you contributed, the evidence for it and the question that remains.