If your training material is in Google Docs, you have the best possible starting point for conversion. The structure is real rather than inferred, which removes most of what goes wrong with PDFs.
A little preparation makes the result noticeably better.
Use real heading styles
This is the one thing that matters most.
A converter splits on structure, and in a Google Doc that means Heading 1, Heading 2, and Heading 3 styles. If your section titles are just bold text at a larger size, they look like headings to a human and like ordinary paragraphs to everything else — so the split has nothing to follow and falls back on guessing.
Converting bold text to proper heading styles takes a couple of minutes and is the highest-return preparation step available. Use the outline view to check: if a section doesn't appear there, it isn't a heading.
Get the section sizes roughly right
The split usually follows your headings, so the sections become the modules. That means section length determines module length.
Skim the document and look for sections that would take more than ten minutes to read. Those want breaking into subsections with their own headings. Equally, a run of very short sections can be merged — a thirty-second module is annoying rather than efficient.
You're aiming for sections that each cover one complete idea and take a few minutes.
Clean up before converting
Things that produce odd output:
- Unresolved comments and suggestions. Depending on how the document is exported, suggested edits can end up in the text. Accept or reject them first.
- Tracked changes left open, for the same reason.
- Tables of contents. These convert into a strange first module listing headings. Remove before converting; the course has its own navigation.
- Cover pages and version history tables. Useful in a document, meaningless as a module.
- Cross-references like "see section 4.2". Module four doesn't have a section 4.2. Either rewrite or accept a slightly odd sentence.
Check each section stands alone
Because modules get completed separately, sometimes days apart, each needs to make sense without the previous one fresh in mind.
Scan for sections opening with "As described above" or "Building on this". Either add a sentence of context or merge that section into the one before it. This is a five-minute pass that meaningfully improves how the course reads.
Keep the doc canonical
Resist the temptation to edit the generated course. If you do, the doc and the course drift apart, and within a year you'll have two versions of a policy that disagree.
The better arrangement: the Google Doc stays the source of truth, and the course is regenerated when it changes. Google Docs is particularly good for this because the version history and comment thread stay in one place — you keep the collaborative editing workflow your team already uses and the course simply follows.
After converting
Read the generated modules before assigning them. Check the split landed where you expected, that nothing important got absorbed into the wrong section, and that the generated questions test decisions rather than wording.
Then reorder if needed. Documents are often organised by structure rather than by teaching order, and moving modules around after conversion is quick.
Doing it
Poplearn takes the document directly — /courses new splits it into modules and writes the quizzes, and the modules then arrive as Slack direct messages, one at a time, so people finish them where they already work.
/assign #support sends the course to a person, a channel, or every new hire from that date forward, and /progress gives live completion by team, exportable on demand.
Free under 20 seats, installs in about a minute.
