The actual job the buyer is doing
The mobile session is short and goal-driven. The buyer is on a site, in a van or between meetings. They have a part number, a job reference and roughly forty seconds. They are not exploring the catalogue. They are confirming a price, checking stock at a specific branch, and placing an order against an account that already exists.
Most B2B mobile sites are designed for the session that does not happen. A homepage carousel, a content-led navigation and a checkout that asks for information the buyer already gave the business in 2019.
Fix one: a quick-order surface that respects the input method
Put a quick-order entry on the first screen the authenticated buyer sees. Allow part numbers, customer-specific codes and saved-list shortcuts. Accept paste from the supplier spreadsheet the buyer is already using. Accept barcode scan through the camera with a single permission ask.
The number to watch is the proportion of mobile sessions that reach a basket within ninety seconds. On the sites where this has been fixed properly, that number moves from low single digits into the thirties.
"The trade buyer is not browsing. They are standing on a site, on the clock, trying to confirm a part number. The mobile experience has to be a tool, not a magazine."
Fix two: branch stock that means something
Show stock at the buyer's default branch and the two nearest alternatives. Show the cut-off for collection today and the cut-off for delivery tomorrow. If the stock figure is older than four hours, say so. Buyers underestimate how often they would rather drive to the next branch than wait for delivery, and overestimate how much they trust a stock figure that does not say when it was last updated.
The detail that matters is the cut-off, not the stock. A buyer who knows they have until two o'clock to confirm an order will commit. A buyer who has to guess will phone the branch.
Fix three: a checkout that remembers
The trade buyer has placed three hundred orders against the same job this year. The checkout should know that. Default the job reference, the cost centre, the delivery site and the required-by date to the last used values for that buyer. Let the buyer override any of them in a single tap. Do not ask for a billing address. The account has one.
The checkout that earns mobile orders is the one that asks for the smallest set of decisions the buyer has not already made. Everything else is friction the trade desk will eventually inherit.
What this is not
This is not a redesign. The visual language can stay. The information architecture can stay. The work is in three surfaces: the authenticated landing, the product page and the checkout. Most teams that try to fix mobile by redesigning the whole site discover that the redesign took a year and the trade buyers still do not use it. The three fixes above ship in a quarter and the mobile order rate moves in the next one.
