D365 F&O at 90% Complete - What I Wish I Knew at 10%

We're 90% through our Dynamics 365 Finance & Operations migration at Julius Silvert. After three years of parallel running VAI S2K, building integrations, migrating data, training users, and weathering approximately 47 moments where I questioned every decision that led us here, we're almost at the finish line.

This post is for the CIO who's at 10%. Or 0%. Or maybe you're at 50% and wondering if the light at the end of the tunnel is actually an oncoming train. I'm going to tell you what I wish someone had told me when we started.

The Setup: Where We Started

Julius Silvert is a $110M food distribution company. Two locations - Philadelphia and Capital Heights, MD. 60 trucks. 5,100+ SKUs across 154 categories, heavy on primal cuts, fresh vegetables, and specialty items. When I joined in January 2023, we were running VAI S2K on an AS/400 that had been chugging along since the early 2000s.

S2K worked. That's the thing. It wasn't pretty, it wasn't modern, but orders got processed, inventory got tracked, and invoices got sent. The problem was everything around it:

• Reporting required manual exports to Excel
• Integration with BFC Dakota (our WMS) was held together with SFTP scripts and prayers
• E-commerce integration with Adobe Commerce was a nightmare of custom WebSphere proxies
• Finding developers who understood RPG was getting harder every year
• The AS/400 hosting costs were climbing while capabilities stayed flat

The business case for modernization was clear. The execution would be anything but.

Lesson 1: The Implementation Partner Matters More Than the Software

I've written before about ERP selection, but here's what I underestimated: once you've picked Dynamics 365 (or NetSuite, or SAP, or whatever), the software is the software. What determines success or failure is who implements it.

We went through five implementation partners during our evaluation. Three of them gave us essentially the same slide deck with our logo photoshopped in. They talked about "best practices" and "industry templates" without ever asking about our actual business.

The partner we chose asked different questions:

• Walk me through your catch weight process for primal cuts
• How do you handle substitutions during picking when a product is out of stock?
• What happens when a truck breaks down mid-route?
• How do your sales reps currently see customer order history?

They understood food distribution. They'd done it before. That expertise cost us about 15% more than the cheapest option, and it was worth every penny.

Lesson 2: Data Migration Is Where Projects Go to Die

Everyone talks about data migration. Nobody talks about it enough.

Our VAI S2K system had 22 years of data. Customer records with multiple ship-to addresses that had been entered inconsistently. Vendor records with contacts who'd retired a decade ago. Item records with descriptions that meant something to someone in 1998 but were incomprehensible now. Pricing matrices that had been adjusted so many times they contradicted themselves.

We initially budgeted 3 months for data cleanup. It took 11.

Here's what I'd do differently:

Start data cleanup before the ERP project officially begins. Like, a year before. Make it a separate initiative. Clean your customer master. Standardize your item descriptions. Audit your pricing. This isn't ERP work - it's business hygiene that you should be doing anyway.

Accept that you won't migrate everything. We eventually decided to archive anything older than 5 years. If we needed historical data, we could access it in a read-only archive. This reduced our migration scope by 60% and nobody has ever complained.

Automate validation ruthlessly. We built Power Automate flows that ran every night comparing S2K data to D365 data during parallel operations. Any discrepancy generated a ticket. This caught hundreds of issues that manual testing would have missed.

Lesson 3: The Catch Weight Problem Is Real

If you're in food distribution, you know what catch weight is. If you're not: it's when you sell by one unit of measure (like cases) but price by another (like pounds). A case of ribeyes might weigh anywhere from 45-55 pounds depending on the cut, and you need to track both the case count and the actual weight for pricing and inventory.

D365 F&O handles catch weight. It's a supported feature. What nobody tells you is that it handles catch weight the way D365 thinks you should handle catch weight, which may or may not match how you actually handle catch weight.

Our specific challenges:

• Two-step receiving: We receive products in one weight (as delivered by the vendor) but sometimes repack them into different weights for customers. D365's catch weight doesn't love this workflow.
• Weight tolerance: When a customer orders 50 pounds of ground beef, they'll accept 48-52 pounds. D365 wanted exact matches by default.
• Pricing tiers: Some customers have price breaks at weight thresholds. D365's pricing engine needed significant customization to handle this.

We ended up with about 40 hours of custom development specifically for catch weight workflows. Our implementation partner had done this before (see Lesson 1) so they knew where the pitfalls were, but it was still more complex than the standard demo suggested.

Lesson 4: Integration Is the Whole Ballgame

D365 F&O is not an island. It has to talk to:

Each of these integrations had its own project plan, its own testing phase, and its own set of surprises. The BFC Dakota integration alone took 6 months to get right because of the real-time inventory synchronization requirements.

What I'd tell my past self: Budget 40% of your total project effort for integrations. Not 20%. Not 30%. Forty percent. And that might still not be enough if you have complex real-time requirements.

Lesson 5: Change Management Is Not Optional

I'm a technology person. I think in systems and processes and data flows. What I consistently underestimate is that technology changes require people to change, and people don't change just because you tell them to.

Our warehouse team had been using S2K terminals for 15+ years. They knew every shortcut, every workaround, every quirk. Telling them "here's a new system that's better" didn't make them happy - it made them anxious. And anxious people make mistakes.

What worked for us:

Super users, not training classes. We identified 2-3 people in each department who were genuinely curious about the new system. We trained them deeply, then let them train their peers. People trust their colleagues more than IT.

Parallel running longer than we wanted. We ran both systems for 8 months. Yes, that meant double entry for some transactions. Yes, that was painful. But it also meant nobody felt like they'd lost their safety net until they were genuinely confident.

Acknowledging what was lost. S2K had things people liked. A specific report format. A particular workflow shortcut. Instead of dismissing these as "old thinking," we acknowledged that some things would be different, and different isn't always immediately better. That honesty built trust.

Lesson 6: Microsoft's Update Cadence Is Relentless

D365 F&O gets service updates continuously. Major releases twice a year. You can't stay on an old version forever - Microsoft won't support it.

This is actually a good thing for security and features, but it creates operational overhead that legacy systems didn't have. Every update needs to be tested. Every customization needs to be validated. Every integration needs to be verified.

We've built a standing "update team" that dedicates one week per quarter to testing and deploying updates. We've automated as much regression testing as possible. We've also learned to be conservative about customizations because every customization is technical debt you'll pay interest on forever.

Lesson 7: The ROI Is Real, But Not Where You Expect

When we built the business case, we projected savings from reduced manual processes, better inventory accuracy, and lower hosting costs. Here's what actually happened:

What delivered ROI:

• Real-time inventory visibility reduced stockouts by 23%
• Automated pricing eliminated billing errors that were costing us $180K/year in credits
• Power BI integration gave sales reps insights that increased average order value by 26%
• Single sign-on and modern security reduced IT support tickets by 40%

What didn't deliver (yet):

• Hosting costs are actually higher than S2K (but we're getting way more capability)
• Staff reduction didn't happen - we redeployed people to higher-value work instead
• Implementation timeline was 18 months longer than projected

The total ROI is positive and substantial, but it came from different places than we predicted. The business case was right that modernization would create value; it was wrong about which specific value streams would materialize first.

Lesson 8: Your Team Will Surprise You

When we started this project, I was worried about my team. Would they be able to handle a modern cloud platform? Would the learning curve be too steep? Would I lose people who couldn't adapt?

I was wrong to worry.

People who'd spent years working around S2K limitations suddenly had tools that did what they needed. The warehouse supervisor who used to spend 3 hours a day on manual reports now spends that time actually supervising. The AP clerk who reconciled invoices by hand now reviews exceptions from automated matching.

We didn't lose anyone because they couldn't adapt. We lost two people who decided they didn't want to - and honestly, that was probably the right outcome for everyone.

The 96% IT staff retention I mentioned isn't just a metric. It's proof that people will embrace change when they see how it makes their work better.

What's Left in the Final 10%

We're not done. The remaining 10% includes:

• Final cutover of the Capital Heights location (Philly is already live)
• Decommissioning the AS/400 (emotional, honestly)
• Completing the advanced warehouse management features we deferred
• Building out the remaining Power BI dashboards
• Documenting everything for the team that will maintain this going forward

We'll cross the finish line in Q1. Three years after we started. Eighteen months later than originally planned. Within budget (barely). And with a system that will serve the company for the next decade.

Would I Do It Again?

Yes. Without hesitation.

VAI S2K was a dead end. Maybe not immediately - we could have run it for another 5-7 years - but eventually, we would have been forced to modernize under worse conditions. Better to transform while you're strong than to be forced to transform while you're struggling.

D365 F&O isn't perfect. The licensing is complex. The update cadence is demanding. Some features that seem standard in other systems require customization. But it's a platform we can build on, integrate with, and extend. That was never true with S2K.

If you're at 10%, keep going. If you're at 0% and considering starting, start. If you're at 50% and questioning everything, that's normal - push through.

The view from 90% is pretty good.