Turning Messy Document Bundles Into a Fast, Confident Editing Workflow
Overview
Staple AI is a document processing platform used by finance and operations teams to manage high volumes of documents, invoices, purchase orders, delivery notes, and more. Documents arrive from multiple sources (email, cloud storage, direct upload), often bundled together as multi-page files. Teams need to review, organize, and prepare these files before they move into the processing pipeline. The Document Editor is the tool that makes this possible, helping operations teams split, merge, move, and export pages across document sections.
Challenge
The challenge required turning a fragile, one-page-at-a-time editing process into a bulk-friendly workspace that all three user types could trust.
No bulk actions: Selecting and acting on multiple pages at once wasn't possible, slowing down high-volume users
No undo: Split and merge were permanent with no recovery, a critical risk for all users, especially less experienced ones
Weak page visibility: Users couldn't identify document types at a glance without opening each page
Broken export confidence: No clear summary of what would actually be exported
Meaningless auto-names: System-generated section names gave zero context, creating extra manual renaming work
No workflow continuity: Editing required jumping out to external tools
Objective & Goal
Primary goal: Help finance and operations teams split, merge, and reorganize messy multi-page document bundles into clean, correctly classified files — fast, without leaving the pipeline. Bulk operations, multi-select, copy, and move across sections need to feel natural and fast, keeping users in flow.
Secondary goals:
Reduce cognitive load and use visual cues (color coding, labels, previews) so users can classify pages at a glance during high-volume processing
Build export confidence and give users previews, section summaries, and clear naming so they don't have to second-guess their work
Auto-generate names for new documents
Process
Starting from the pain points and personas, I framed the redesign as testable hypotheses before moving into wireframes:
If we introduce bulk page selection with a persistent action bar, high-volume users can act on many pages at once instead of one at a time.
If we add an undo stack for destructive actions like split, merge, and delete, users across all experience levels act with more confidence, reducing hesitation and error-recovery time.
If we use AI to identify key information about documents, we can enable more powerful bulk operations.
If we show a clear export summary screen listing every section, page count, and filename before confirming, users will feel more confident and make fewer errors.
If we auto-generate meaningful section names based on detected document type and date, users spend less time on manual renaming and the workspace stays organized by default.
If we add a confirmation prompt before any destructive action like section delete or merge, users make fewer accidental errors and trust the tool enough to work faster.
If we introduce a fixed view with clear status tags like Edited, Ready to Export, and In Review, all users can orient themselves in a complex workspace instantly without losing track of progress.
I then mapped the end-to-end editing flow search or upload a document, open it, review its pages, select pages, choose an operation (split, merge, copy/move), update the workspace, review the new document, and export and designed each screen to resolve a specific point of failure in that flow: corrupted-file and search handling on upload, silent preview failures and pagination on review, scroll-reset and shift-click gaps during page selection, no-undo risk during split, type-mismatch checks during merge, and duplicate-name and mid-export failure handling in the export flow.
The resulting editor workspace has a persistent document library on the left, a working canvas in the center with page-level thumbnails and status tags, and a bottom action bar (Split / Merge / Copy / Move to) that stays available across bulk selections. The Export modal was designed around hypothesis 4 directly: a format picker (PDF / DOCX / ZIP), a section checklist showing page count and detected document types per section, a duplicate-name warning, and a confirm button that updates its own label ("Export 2 sections as PDF") to reflect exactly what's about to happen.
Result
Redesigned the onboarding experience for a global CDN SaaS platform to improve user activation and retention. By simplifying the flow, refining plan categorization, and integrating customer segmentation, the redesign achieved 84% more direct purchases, 33% fewer inactive users, and a 78% boost in monthly revenue — proving the power of data-driven UX design in SaaS environments.
Learnings
Designing for three personas with very different risk tolerances a power user who wants speed, a verifier who wants certainty, and a novice who wants safety showed that the same bulk-action system can serve all three if destructive actions are reversible and status is always visible. Tying every screen back to a specific pain point in the original flow (rather than a general "make it easier" brief) made trade-offs like defaulting to confirmation prompts an easy call, since the alternative was named explicitly in research.
Additional learnings:
Undo and confirmation prompts aren't redundant; undo protects speed-focused users after the fact; confirmation protects careful users before the fact
Auto-naming only builds trust if it's paired with an easy way to catch and fix conflicts, which is why the duplicate-name warning sits inside the export modal rather than after export
Benchmarking against AI-native competitors (Document AI, Rossum) clarified that the goal wasn't full automation it was making manual review fast enough to compete with automation
You’ve seen the results now see
Figma File
I document everything in Figma the ideas, the iterations, the experiments.
Take a peek, it’s all there.
CLICK TO READ
How Delivery Hero streamlines marketing reporting across all their brands with Clarisights. How Delivery Hero streamlines marketing reporting across all their brands with Clarisights. How Delivery Hero streamlines marketing reporting across all their brands with Clarisights. How Delivery Hero streamlines marketing reporting across all their brands with Clarisights. How Delivery Hero streamlines marketing reporting across all their brands with Clarisights. How Delivery Hero streamlines marketing reporting across all their brands with Clarisights. How Delivery Hero streamlines marketing reporting across all their brands with Clarisights. How Delivery Hero streamlines marketing reporting across all their brands with Clarisights. How Delivery Hero streamlines marketing reporting across all their brands with Clarisights. How Delivery Hero streamlines marketing reporting across all their brands with Clarisights. How Delivery Hero streamlines marketing reporting across all their brands with Clarisights