Turning purchase order chaos into a simpler workflow.

How I moved from warehouse visits to an AI-assisted purchasing workflow by understanding the real problem before designing the solution.

RoleProduct Designer II

ContributionDiscovery, workflow, interaction, prototyping

SurfacesScanner + Web

OutcomeReleased to capped beta

01 · Observe

I visited warehouses to understand the real problem.

In order to understand what was happening in the real world, I visited warehouses and watched how people received stock.

I found that even with digital products, many important steps were still manual. Teams checked paper documents, compared information by hand, and moved data between different systems. The real challenge was much bigger than the feature we planned to build.

Warehouse visit showing the people, documents, and manual work behind receiving stock

02 · Structure

Mapping the current and ideal receiving workflow.

I mapped the current receiving workflow and the ideal flow with my PM. This helped us agree on where to start and what to test first.

Current and ideal receiving workflow mapping

03 · Test the first assumption

We tested the fastest idea first.

We started with document capture in Scanner because it was the fastest idea to test.

Early prototypes helped us learn quickly, but continued research showed we were solving only one part of the workflow.

Early lo-fi document capture concept tested in Scanner

Creating Purchase Orders—not receiving them—was where merchants spent the most time.

05 · Understand why

Why was this happening?

We mapped the wider ordering workflow to understand why Purchase Orders were often created outside Lightspeed.

Purchase order workflow mapping across different businesses

This changed how we saw the problem.

The map helped us compare different businesses using the same view, even when the people and processes were different.

Insight 01 · Jobs

Jobs helped us compare different businesses.

Instead of comparing job titles, we compared the jobs people needed to get done.

Insight 02 · Variation

The workflow changed from business to business.

Industry, suppliers, team size, and store setup all changed how people worked.

Insight 03 · Why POs were skipped

This explained why Purchase Orders were often skipped.

Many buyers skipped Purchase Orders because Lightspeed wasn't part of their ordering process.

06 · Evidence

Scanning documents solved only part of the problem.

The data showed that the bigger effort happened earlier, when merchants created Purchase Orders.

5–23 hours spent creating Purchase Orders manually

The burden increased sharply for merchants handling large catalogues and high-volume orders.

70% created Purchase Orders when stock arrived

Only 30% prepared them in advance—revealing how much friction existed before receiving even began.

The manual workflow

Supplier
document

Purchase
Order

Search / Add
products

Enter quantities
and costs

Receive
stock

Every Purchase Order started with manually recreating information from a supplier document.

If we could reduce the effort to create Purchase Orders, we could improve the workflow for most merchants—not just the final receiving step.

07 · Decision 01 · Expand the solution

I took on the Web workflow—not just the mobile feature.

Once we understood where merchants spent the most time, the solution could not stop at document capture. I proposed supporting the full workflow on Web, where merchants could review matches, resolve exceptions, and create an editable Purchase Order.

Rather than reuse an interface designed for a simpler task, I redesigned the experience around matching, catalogue changes, and unresolved items—the decisions merchants actually needed to make.

Whiteboard exploration of the end-to-end automated purchase order workflow — Workflow exploration

Cross-functional design review with feedback on the Web workflow — Design reviews

AI-assisted prototype exploration showing the prompt and generated interface — AI-assisted iteration

Usability testing session for the automated purchase order workflow — Customer validation

The decision: design the Web experience around the workflow we discovered, rather than force the problem into an existing UI.

08 · Decision 02 · Connect the whole journey

The Web workflow also changed what mobile needed to do.

When I stepped back and looked at the end-to-end experience, Web alone was not enough. Merchants still needed a simple way to start the workflow where the physical work happened—using Scanner to create an order manually or capture a supplier document.

I designed the mobile flow around that hand-off: start the job in Scanner, let the system process the document, then continue the complex review on Web. I used AI-assisted prototyping to test native behaviour faster, but the product decision was about giving each surface a clear role.

AI-assisted native mobile prototype built to explore the purchase order entry flow — Validate the mobile experience faster

Updated Scanner flow presented during a cross-functional walkthrough — Define the full Scanner journey

09 · Decision 03 · Choose the scalable direction

The most familiar concept was not the strongest product decision.

We chose Concept B because it created a stronger foundation for future workflows.

Separating matching into its own workspace gave the team clearer technical boundaries, kept the Purchase Order focused, and created room to support more document-based workflows over time.

Direction A

Keep review inside the Purchase Order

Pros

Fewer transitions, familiar context, and a direct path from imported data to an editable order.

Cons

Matching states, exceptions, and catalogue changes crowded the core Purchase Order experience.

Selected direction

Direction B

Separate matching into a dedicated workspace

Pros

Clear ownership of matching, better support for exceptions, and room to reuse the capability across future workflows.

Cons

Added one more step and required a clear hand-off back into the Purchase Order.

10 · Outcome

From concept to real work.

Merchants used the connected workflow to turn supplier documents into editable Purchase Orders.

Outcome

From a supplier document to a ready-to-review PO

One connected workflow across Scanner and Web—from capture to matching and review.

70%+of Purchase Orders processed through OCR by key beta merchants

10 minfrom hours of manual work, according to merchant feedback

Liveend-to-end workflow shipped beyond the prototype

Voice from merchants

“Absolutely game changing. It’s cut down hours of my work to about 10 minutes.”

“Worked perfectly as designed. Order loaded exactly.”

“This has been extremely useful and we hope that it will continue to be offered even after beta.”

“Very easy and eliminates me from having to convert a PDF into CSV or Excel to import.”

Understanding the workflow shaped every design decision that followed.