PowerPoint
PowerPoint is the honest baseline: it is paid for, it is installed, and nobody needs training. Any argument for something else has to be worth the cost of not using the thing everyone already has.
What it cannot do is bind. Slide masters, layouts and placeholders are real, and a placeholder is something assistive technology reads correctly. What is missing is anything that enforces them: you can always draw a shape beside the placeholder, and to a screen reader that shape is not a heading. The Accessibility Checker tells you afterwards, which is a different thing from a format in which it could not have gone wrong. The same gap shows up downstream, where a deck built from your own data is a file your script assembles, with nothing to check it against.
Google Slides
Google Slides is the one a lot of people use without ever having chosen it: it comes with Workspace, it opens in a browser on any machine, and real-time collaboration with a version history has worked there for longer than anywhere else. It also has a serious API. A batch update that replaces text across a copied deck is a working template merge, and Apps Script puts that within reach of somebody with no infrastructure at all.
The two things it does not offer are the two this table is about. There is no version you run yourself, so where your material is processed is Google’s answer rather than yours. And a slide is still shapes at coordinates: the merge does not know what your fields mean, so it inherits the formatting of the first character it replaces and copes badly with text longer than the placeholder it lands in. It is PowerPoint’s model with much better collaboration, not a different model.
Canva
Canva solved the problem PowerPoint left open: it made a good-looking deck reachable for somebody who is not a designer, and it made brand assets something a team can share rather than email around.
The cost is where it all sits. The decks, the images and the brand kit are on somebody else’s platform, under their terms and their roadmap, billed per seat as the team grows. For a lot of organisations that is a perfectly reasonable trade. For the ones that have to answer where their material is processed, or that would rather not have their visual identity live in an account they do not control, it is the whole question.
Gamma
Gamma is very good at the thing it does. If you need a deck this afternoon and nobody is going to audit it, generating one is the sensible answer, and pretending otherwise would be silly.
What it is not is a place where your organisation’s rules live. A Gamma theme carries your fonts, colours and logo, so the output can genuinely look like you; what a theme cannot do is decide what a slide is. The generator still picks the structure, and nothing checks the result against a rule you wrote down. Add that there is no self-hosted version and "our decks" stays a thing on their platform. Deckyard is aiming at the step after that: not the fastest first draft, but the deck you can put in front of a board without checking every slide first.
Deckyard
The trade Deckyard makes is to put the rules in the format. A slide is a record with a declared type, the theme carries every colour and font, and anything filling a field is validated the same way whether it is a person, a script or an agent. Other tools can fill a template from data too; the difference is that here the type of the slide is declared and the system checks the filling against it, rather than dropping text into a box on a drawing. That is where the unbreakable brand, the accessible output and the integrations all come from; they are one decision, not four features.
And that is also the cost. There is no marketplace of 3,000 templates, the integration list is short, and the software is younger than everything else in this table. Self-hosting is real work: a server, upgrades, backups, somebody who owns it. If nobody in your organisation is that somebody, a managed instance exists, but then you are paying for hosting rather than getting it free with a licence you already own.