Why Your Automation Fails: The Missing Foundation

TL;DR: Nick Thompson from WWT's automation practice dropped some hard truths about why enterprise automation keeps hitting walls. His reality check? You can't automate what you don't know exists. Here's why establishing a single source of truth isn't just good housekeeping—it's the foundation that makes or breaks your entire automation strategy.

Quick note: We recorded this conversation with Nick earlier this year, but I'm still catching up on overdue blog coverage. The insights are just as relevant today—maybe even more so as organizations continue to struggle with these same foundational challenges.

Something Nick Thompson said stopped me in my tracks: "There is nothing without the network. There is no application without a network to ride on top of."

Nick's a distinguished solutions architect with World Wide Technology's automation practice, and he's spent the last six years having remarkably similar conversations with clients. Not because the technology hasn't evolved—it absolutely has. But because organizations keep hitting the same fundamental wall: trying to automate infrastructure they can't properly document.

The problem isn't that companies lack automation. Every client has pockets of it scattered everywhere. Individual teams doing their own thing, people automating processes they don't even tell their teammates about. The real challenge? Building enterprise-grade automation capability that actually scales beyond those silos.

The Excel Spreadsheet Reality Check

When Nick mentions "the days of 35 Excel spreadsheets and a Confluence doc over here and someone's Word doc in a text file floating wherever"—he's not talking about ancient history. He recently worked with a client still using Mail Merge out of Excel to render configuration templates for network devices.

Think about that for a second. They're managing critical infrastructure with the same tools we use for holiday card lists.

This isn't about technology sophistication. It's about the fundamental challenge of documentation. You can't automate something that isn't well documented, and you definitely can't scale automation when your documentation is scattered across dozens of systems in different formats.

When Automation Becomes the Problem

Nick shared a scenario that probably sounds painfully familiar: you've got a device that's had manual changes made to it over time. Those changes were never reflected in your documentation. The device fails, someone swaps it out, and automation kicks in to restore it to the "correct" state—except the correct state is now wrong.

Suddenly you've got an inconsistency in your environment, an outage that shouldn't have happened, and a hundred people on a bridge call trying to figure out what went wrong. All because your data wasn't clean.

The stakes here are higher than just operational headaches. Nick works with highly regulated clients who need to prove their infrastructure is in a known good state for compliance. When auditors ask for evidence that security baselines are applied across your entire infrastructure, you better have more than hope and a prayer.

The Three-Step Foundation That Actually Works

Nick walked me through three practical steps that organizations are using to break out of this cycle:

1. Select the Right Platform

Excel isn't cutting it anymore. You need a platform that can handle the scale of modern infrastructure—hundreds of thousands of devices across data centers and branch offices. But it's not just about capacity.

Your platform needs a robust API for both consuming data and keeping it updated. It needs to be multi-user with role-based access control and audit logging. And it needs enterprise support because this becomes the core of your automation infrastructure.

Nick calls these "network sources of truth" or "single sources of truth"—though there's overlap with data center infrastructure management tools since they're all dealing with the same fundamental challenge of asset tracking and lifecycle management.

2. Focus on Data Ingestion and Hygiene

This is where the rubber meets the road. You need to aggregate data from different systems, validate it, and keep it clean over time. The goal isn't just to have the data—it's to have data you can trust.

Nick emphasizes the importance of defining intended state versus actual state. Your source of truth says the infrastructure should look like this, but there's no guarantee it actually does. You need reconciliation processes, configuration compliance runs, and validation that assets in your system actually exist and are online.

3. Operationalize the Source of Truth

This is the big behavioral shift. Instead of logging into a switch to make a port change, you update the source of truth and let automation handle the rest. The data flows down through automated workflows to configure the actual device.

This sets up self-service capabilities where a site contact can request a port configuration through a service catalog. They're not touching the switch—they're updating the source of truth, and automation takes care of the implementation.

Start with the Network (Trust Me on This)

When I asked Nick where organizations should begin, his answer was immediate: the network. "There is nothing without the network. There is no application without a network to ride on top of."

This makes sense from both a technical and practical standpoint. The network team sees everything—they're the glue between all services and systems. Once you've modeled and adopted these principles for the network, you can layer on compute, storage, and applications.

The AI Connection Nobody's Talking About

One thing that struck me in our conversation was Nick's perspective on AI and automation. His team likes to say "there's no AI without automation"—and he's seeing this play out as clients build large-scale facilities for private AI workloads.

But there's opportunity in the other direction. AI can help with code generation and operational queries. Nick mentioned the potential for natural language interfaces to source of truth data, where an operations person can ask questions in plain English and get answers without understanding the underlying systems.

The key is having clean data to feed these AI systems. If you're providing false answers that look truthful, people will run with them and cause problems.

The Leadership Wake-Up Call

Nick's seen a shift in the last year and a half where leadership is finally understanding how important this foundation is. They may not care whether BGP is enabled on the WAN, but they definitely care about having a complete picture of their infrastructure, knowing what state everything is in, and being able to answer questions about security and compliance.

A good source of truth becomes that central point for providing business leaders with answers to the questions they're actually asking—less technical details, more fundamental operational and security insights.

How WWT Approaches This

What I appreciate about Nick's team is that they don't just sell tools—they help organizations understand what the right source of truth is for their specific environment. They analyze existing data sources, sometimes discover teams already using open source tools, and figure out how to elevate those efforts into enterprise capabilities.

They're also integrating this foundation into larger infrastructure refresh programs. Instead of just getting new hardware, clients get a fully automated network backed by a source of truth with a new operational model.

The Bottom Line

Nick's core message is simple but profound: you can't automate what you don't know. Every successful automation initiative depends on having a reliable, clean, centralized view of your infrastructure.

The organizations that get this right reduce risk, increase consistency, and enable the kind of rapid, reliable change that modern business demands. The ones that don't? They keep having the same conversations about automation challenges six years later.

Watch the full episode to hear Nick's complete breakdown of establishing a robust source of truth that actually enables automation success.

Thanks as always!