Skip to content
All insights

General

Build, Buy or Connect? One Supplier Delay, Three Ways to Handle It

By Shanshan "Sasha" Ma

The build vs. buy debate is easier when you run it on one real piece of work. A single supplier delay, handled three ways: build a tool, buy a tool, or connect the systems you already have. What each takes, who keeps it running, and what it still misses.

The build vs. buy debate never gets old. It usually runs in the abstract: features, licenses, headcount, roadmaps.

At LevelOneAI, we think it's easier to decide on one real piece of work. So here's one, run three ways.

The work

PO 4512 for machined housings is confirmed for Oct 12. On Oct 5, at 7:48 AM, the supplier emails the buyer: tooling problem on line 2, new date Oct 26.

To act on that email, someone has to know four things that live in four places:

  • The new date is in the supplier's email.
  • The old date is in the ERP, which still says Oct 12.
  • The date it's needed is in the production plan: assembly run A-88 needs the housings on Oct 20.
  • What it puts at risk is in the customer orders: Brightline ($138,000, ships Oct 28) and Keller ($76,000, ships Nov 2).

Then someone has to decide. Wait and ship $214,000 late, expedite for $9,400, or ask the supplier to split the shipment for $1,800. And someone has to make sure the answer actually comes back, the ERP gets updated and the parts arrive.

In most companies, that someone is the buyer, with three tabs open and a phone. Whether the work gets stuck depends on how busy the buyer is that week.

Here are three ways to fix that.

Option 1: Build it

Your team, or a capable person with an AI coding tool, builds a tracker. It reads supplier emails, picks out the PO number and the new date, compares it with an ERP export and flags the difference.

What it takes: a first version can come together in days. Getting it right takes longer. Supplier emails reference POs in a dozen different ways, threads mix three orders, and the ERP export changes format after an upgrade. Each of those is a fix someone has to make.

Who keeps it running: whoever built it. When they're busy, or leave, the tool quietly gets out of date and the buyer goes back to the tabs.

What it still misses on PO 4512: usually the last two steps. A tracker that flags "date changed" doesn't know run A-88 needs the parts on the 20th unless someone also wires in the plan, and it doesn't keep the issue open until the split shipment is confirmed and received.

When it's the right answer: the workflow is unique to you, it's part of why customers choose you, and you have people whose job it is to maintain it. (More on that trade-off in Vibe Coding Your Own Business System Is Easy. Keeping It Running Is the Hard Part.)

Option 2: Buy a tool for it

You buy a supplier portal, a supply chain visibility tool or an ERP add-on, so suppliers confirm dates in a structured way and changes land in the system.

What it takes: a subscription, an implementation, and the part nobody quotes: getting suppliers to use it. Some will. Many keep emailing, because email is what they use with every other customer.

Who keeps it running: the vendor maintains the software. Your team maintains the adoption.

What it still misses on PO 4512: the tool sees the date change if the supplier enters it there. It usually doesn't know about run A-88 or the Brightline and Keller orders, because those live in other systems. The buyer still connects the dots and still chases the close.

When it's the right answer: the gap sits inside one function, your suppliers will adopt it, and you need what the tool does well, such as structured confirmations, supplier scorecards or documents in one place.

Option 3: Connect the systems you already have

Keep the ERP, the email and the plan where they are. Add something that reads across them and holds the work while it's moving.

What it takes: read access. The supplier email is matched to PO 4512, compared with the ERP date and the need date, and linked to the customer orders it affects. The buyer sees one item: what changed, what it puts at risk, the three options and what each costs. They approve the split on Oct 5, the day the email arrived. The item stays open until the supplier confirms in writing and 60 housings are received for run A-88.

Who keeps it running: the matching, the rules and the follow-through are maintained as a product, not as a side project. Your team owns the decisions.

What it still misses: it doesn't replace a supplier portal if you need suppliers to self-serve, and it can only connect what it can read. If the production plan lives in someone's head, that's the first thing to fix.

When it's the right answer: the work crosses systems, the information already exists, and the problem is that nobody owns the connection between them.

Side by side, on PO 4512

  • Sees the supplier's email. Build: if you build it to. Buy: only if the supplier uses the tool. Connect: yes.
  • Knows run A-88 and the customer orders. Build: if you wire in the plan and orders. Buy: usually not. Connect: yes, from the plan and order data.
  • Shows the options and what each costs. Build: rarely. Buy: rarely. Connect: yes.
  • Keeps the issue open until it's resolved. Build: if you build it to. Buy: usually not. Connect: yes.
  • Who maintains it. Build: your team. Buy: the vendor, plus supplier adoption. Connect: the vendor.
  • Time to a first result. Build: days for a prototype, months before you can rely on it. Buy: months, including the supplier rollout. Connect: the same day for one buyer, weeks for the team.

None of the three is always right. Sometimes the answer is better software. Sometimes it's connecting the gaps between the systems you already have. Often it's both: buy the tool your suppliers will use, and connect what it knows to everything else.

Three questions to decide

Before you compare vendors or start a build, take one real exception from last month and ask:

  1. Where did the information live? If the answer is one system, improve that system. If it's four, the gap is between them.
  2. Who connected it, and how long did it take? That's the work you're actually trying to support, and the cost you're trying to remove.
  3. What did it cost when it was late? One exception, one number. If you can't name it, you can't size the fix or prove it worked.

The goal isn't to make people adapt to the software. It's to make the software support the work, whether you're starting from scratch or building on what you have.

Try it on your own work

You can see what option 3 looks like on your own email and an open-PO export with Operations Workbench Starter, free for 30 days. When the whole team needs it, with live ERP data and approvals, that's the Operations Workbench. For the decisions to make before any of this, see Five Decisions That Decide Whether Your Software Supports the Work.

The PO 4512 example is illustrative, with a sample company and data.