To add B-roll to a video: review the complete primary edit first, mark the moments that need visual support, define what each visual should explain or demonstrate, choose the strongest available source, verify accuracy and usage rights, place it close to the relevant sentence, set its duration, and inspect the complete export. Do not add B-roll to every available gap.
The principle behind the order: begin with the message, not with a folder of visual assets. And begin late enough; B-roll placed on an unstable first cut gets redone when the structure moves.
Step 1 to 3: find the moments and name their purpose
Watch the whole edit once without touching anything, and resist the assumption that every talking-head section needs to be covered. Mark moments with one of six labels: explain, demonstrate, prove, context, transition, emphasis. A marker is not yet a decision to add anything.
Then force each marker through one sentence: "This visual helps the viewer understand ______." If the honest completion is "that the video is dynamic", it fails. Purpose first, footage second.
Step 4 and 5: choose the source, then get the asset
Per marker, pick the strongest source: original footage for anything real and specific, a screen recording for software, a screenshot or still when movement adds nothing, a diagram for processes and abstractions, stock for generic settings, an AI-generated visual for concepts that are clearly illustrative, or no B-roll at all when the speaker is the best available image.
Search specifically. "Business technology" returns wallpaper; "creator reviewing multiple video takes on laptop" returns something usable. For generated material, one boundary is absolute: do not generate a realistic customer, office or result and present it as genuine evidence.
Step 6: verify before placing
Four checks per asset: factual fit (a generic graph does not prove a specific result), privacy (names, faces, customer data, notifications), rights (the licence covers this use), and misleading implications (what does this image claim when it appears over these words?).
Step 7 and 8: place and time it
Place the clip close to the line it supports. Too early spoils the point, too late explains something the viewer has already given up on; usually the visual lands once the complete idea has been spoken. Then set duration by comprehension, not by rule: long enough to understand, gone when the subject changes. Ignore any advice with a number in it ("change the visual every two seconds"); the sentence decides.
Step 9 to 12: transitions, captions, audio, pacing
A clean cut is often sufficient; do not add transitions simply because the software offers them. Check every caption group that crosses the clip: readable, and not stacked on top of a text-heavy recording, because nobody reads two dense layers at once. Keep narration dominant and let B-roll audio through only when the natural sound adds context. Then watch the affected minute for load: speech plus captions plus visuals plus music draw from one attention budget.
Step 13: the removal pass
Walk the timeline once more, asking of every clip: what breaks if this goes? Selective B-roll is often stronger than constant B-roll, and the removal pass is where average edits become clean ones.
Step 14: inspect the export
Watch the exported file, not just the editor: crop, caption overlap, timing, rendering quality. A visual that works inside the editor may render differently in the final file, and the file is what viewers get.
Where ReadyForm fits
ReadyForm does the first twelve steps' groundwork while making the complete edit from your takes: it finds stock for the sentence it supports, uses your own Library uploads, and times each placement to the words. Every clip remains yours to move, trim, replace or delete per scene, which is exactly steps 13 and 14. See how the edit is made or the AI B-roll feature.