SharePoint

Before we touch a connector we ask who actually owns the number inside that SharePoint file: pricing, product attributes, supplier cost, promotional calendars. Most retailers can't answer cleanly. The spreadsheet looks tidy because one person has quietly been checking it every morning before it reaches Shopify or NetSuite, catching an error nobody built a rule for.

We map every file, folder and tab currently feeding a live system, then price the actual cost: a missed price update that undersells stock, a stale cost price that understates margin at month end, a product sheet that reaches the warehouse a day late and delays a launch. That conversation usually decides the commercial case on its own.

The question that matters first: does SharePoint stay a working document for buyers and merchandisers, or does it become a feed that other systems trust without a human checking it first? Answer that wrong and the integration inherits every typo in the sheet.

Ownership has to be explicit. Someone in commercial owns the pricing tab and signs off changes; someone in finance owns cost price and margin thresholds; someone in ecommerce owns when an approved change actually goes live. Without that split, a well-meaning edit to a shared sheet overwrites an approved promotional price the night before launch, and the discount runs at the wrong margin for a week before anyone notices.

Build the approval step into the connection itself, not into a person's inbox. A validation and margin check before release matters more here than in almost any other part of the design.

SharePoint was never built to be a transactional feed, and the point-to-point scripts teams build to compensate for that tend to be the first thing that breaks when a second brand or a new warehouse gets added. Patchworks, the integration layer we build this on, treats the file as a trigger rather than a database: it watches for change, runs validation and margin checks, and releases approved data into Snowflake, NetSuite or Shopify using connectors already built for that work.

That makes onboarding a new channel or a new pricing region a configuration change rather than a new script. Checks that used to depend on someone opening the file each morning run on their own, with exceptions raised for a person to resolve rather than silently applied. Cogent² runs that layer; the expertise is in knowing which validation rule actually protects margin, not in the platform itself.

Queue health here means watching how long a change sits in the validation step before anyone approves it, not just whether the file was read. A pricing update that's been waiting three days for sign-off is the same commercial risk as one that never arrived.

Exceptions need a visible home: failed validations, blank rows, mismatched SKUs, all surfaced on a dashboard rather than buried in an email nobody checks on a Friday. Drift shows up as a gap between what's approved in the sheet and what's live in Shopify or NetSuite; spotting that gap before a customer sees the wrong price, or finance closes the month against the wrong margin, is the entire point of watching it.

The runbook is short, which is the problem: a format change to the pricing sheet, a renamed tab, a new approver, can all break the connection quietly. Document who edits the sheet, who approves a release, and who gets paged when a validation fails.

Escalation needs a name, not a department. When pricing stalls in the queue the night before a promotion, someone has to own picking up the phone, not leave it for the next standup.

Hypercare after go-live focuses on the first full pricing cycle: a seasonal markdown, a supplier cost change, a promotional calendar update pushed through the new connection rather than copied by hand. That first cycle tells you whether the approval step survives real pressure or gets skipped because someone's in a hurry.

Peak readiness means checking who's covering sign-off when the usual approver is on leave during the busiest trading weeks of the year. A pricing change stuck in an empty approval queue on a Friday before a sale weekend is a solvable problem if support is watching it, and a missed sale if nobody is.