Before the First URL: Designing the Information Architecture of a Manufacturing Marketing Agency

There is a stage in website development that is often invisible to the client. No pages have been designed yet, there are no URLs, there is no navigation menu, and there may not even be a single line of code. And yet one of the most important parts of the project is already happening: the information architecture is being designed.

For a simple website, this can look like an unnecessary exercise. For a serious manufacturing company, or a marketing platform built specifically for manufacturers, it’s something very different. It’s the foundation on which the entire information system will eventually operate.

Start with the model, not the pages

When I started designing the new Manufacturing Marketing Agency project, I deliberately didn’t begin by asking what pages the website should have. I started with a different question: what does the manufacturing marketing domain actually contain? This is a conceptual modeling problem.

In software engineering, we don’t normally begin by creating thousands of individual objects without first defining the entities and relationships that make up the system. Information architecture works in a similar way: before we create pages, we need to understand the structure of the information those pages will represent.

Manufacturing marketing is not a simple domain. A manufacturer can be described simultaneously by its industry, its production technologies, its business objectives, its products, its applications, its capabilities, and the marketing services relevant to its situation. Trying to force all of this into one hierarchical tree quickly becomes problematic, so I started with four independent dimensions.

The four dimensions

Industrial answers the question of what kind of manufacturing or industrial business we’re talking about: Automotive Manufacturing, Aerospace Manufacturing, Metal Fabrication, Industrial Equipment, Electronics Manufacturing, Medical Device Manufacturing, Construction Products, HVAC Manufacturing, Furniture Manufacturing, Contract Manufacturing. This dimension describes the industrial context.

Technology answers the question of how the product is manufactured: CNC Machining, Laser Cutting, Laser Welding, Additive Manufacturing, Robotics, Industrial Automation, Injection Molding, Sheet Metal Fabrication, Welding, IoT and Industry 4.0. This is a completely different dimension, because a technology can exist across multiple industries. CNC machining, for example, is relevant to automotive, aerospace, medical devices, industrial equipment, and many other manufacturing sectors, so Technology cannot simply be treated as a child category of Industrial.

Solutions answers the question of what business problem the manufacturer needs to solve: Lead Generation, OEM Customer Acquisition, Market Expansion, New Product Launch, Distributor Growth, RFQ Generation, Engineer & Specifier Reach, Sales Growth, Competitive Displacement, Digital Sales Growth. These aren’t industries and they aren’t technologies, they represent business objectives and problems.

Services answers the question of what the agency can actually do to address those problems: Manufacturing SEO, Technical Content Marketing, Google Ads / PPC, LinkedIn Marketing, Industrial Web Design, AI Search Optimization, Conversion Rate Optimization, Marketing Automation, Account-Based Marketing, Analytics & Attribution. Again, this is a different dimension.

These are not four levels of one hierarchy

This distinction is probably the most important part of the model. I’m not building something like Industrial leading down to Technology, then Solution, then Service, because that would imply every item belongs somewhere below another item, and it doesn’t. Instead, I treat the four dimensions as independent coordinates of the same conceptual space.

For example, a company could be described as Industrial: Automotive Manufacturing, Technology: CNC Machining, Solution: OEM Customer Acquisition, Service: Manufacturing SEO, all at once. The manufacturer exists in one industrial domain, uses certain technologies, has specific business objectives, and different services can be applied to those objectives. This is much closer to a multidimensional model than to a traditional website menu.

But the four dimensions are not enough

This is where another layer becomes important: relationships. Categories by themselves are not the system, the connections between them are.

Consider a CNC machining company serving the automotive industry. There may be a relationship linking Automotive Manufacturing, CNC Machining, Precision Manufacturing, OEM Customer Acquisition, and Manufacturing SEO, and these relationships aren’t necessarily parent-child, they’re semantic. One concept is relevant to another, and one entity can be associated with several others. A case study can demonstrate a particular technology, that technology can matter across several industries, a specific business problem can occur in those industries, and several services can address that problem. This is where the architecture begins to move from a taxonomy toward a semantic graph.

The website becomes a network, not just a collection of pages

Imagine a future website with 200 service pages, 300 industrial pages, 200 technology pages, 100 solution pages, 500 blog posts, 100 case studies, and hundreds of FAQs, applications, guides, and other resources. If these pages are created independently, the result can become a very large collection of disconnected documents. But if they’re built around a common conceptual model, the same content can participate in multiple relationships.

A case study about laser welding for an automotive manufacturer could connect to Industrial: Automotive Manufacturing, Technology: Laser Welding, Solution: OEM Customer Acquisition, and Service: Manufacturing SEO, as well as related applications, products, technical articles, FAQs, and other case studies. Now the website has an internal structure that reflects the actual domain. The content is no longer simply tagged, it’s related, and that distinction matters.

Why this matters for SEO

This is also where information architecture stops being an abstract exercise and becomes a practical SEO tool. Search engines don’t see a website exactly as a human does: they crawl pages, follow links, analyze page content, headings, anchor text, and surrounding context, and they also analyze how pages relate to one another through the site’s internal linking structure. So the architecture of a large website influences how its information can be discovered and understood. A logical hierarchy helps, meaningful internal links help, and clear relationships between related content help. But there’s another important idea here: before keyword research, there’s a conceptual space.

Building the keyword space before keyword research

Keyword research is usually described as the starting point of SEO. In practice, I see it slightly differently. Before asking what keywords people search for, it helps to first understand what concepts actually exist in the domain.

Suppose we identify Automotive Manufacturing, CNC Machining, Precision Manufacturing, OEM, RFQ Generation, Engineer Engagement, and Manufacturing SEO. We haven’t yet performed keyword research, we don’t yet know search volume, keyword difficulty, or the exact wording potential customers use. But we’ve already created something valuable: a semantic space in which keyword research can take place. I think of it as a keyword space, and keyword research then becomes the process of mapping real search behavior onto that conceptual space.

The sequence becomes concept, entity, category, relationship, search terms, content, URL. That’s very different from collecting a spreadsheet of thousands of keywords and then trying to figure out what they mean afterward.

And this becomes increasingly important for AI search

Search is changing. Users are increasingly asking questions in natural language, and AI systems and AI-assisted search interfaces are increasingly retrieving, combining, summarizing, and presenting information from large collections of documents. In that environment, simply having a large amount of content isn’t necessarily enough, the system needs to understand what the content is about and how different pieces of information relate to each other.

Consider concepts like laser welding, automotive manufacturing, OEM supplier, production capacity, RFQ generation, and contract manufacturing. A page containing these words is one thing. A website whose architecture consistently establishes meaningful relationships between them is something else.

The objective isn’t to trick AI, and it isn’t to add artificial tags everywhere. It’s certainly not to assume that a particular taxonomy automatically produces rankings or AI citations. The objective is more fundamental: to make the information structure explicit. The more complex the website becomes, the more that principle matters.

This is not just categorization

This is probably the biggest misconception about this stage of the project. Someone looking at a taxonomy spreadsheet might ask why spend so much time on categories. The answer is that categories aren’t the final product, they’re the model of the domain. They determine what concepts the website is capable of representing, create the framework future content can fit into, provide the structure for internal linking, help organize entities, create the conceptual space for keyword research, and provide a consistent framework through which content can be connected.

Without this model, hundreds of pages can become hundreds of isolated URLs. With the model, those same pages can become parts of a larger information system.

Designing for a manufacturing company is different

Manufacturing websites are particularly interesting because the underlying domain is often much more complex than a typical B2B service website. A manufacturer may have multiple production technologies, multiple industries served, multiple products, multiple applications, different buyer personas, engineering requirements, technical specifications, certifications, production capabilities, geographic markets, OEM relationships, distributors, RFQ processes, and long sales cycles.

The website therefore has to represent a multidimensional business. A simple Home, Services, About, Contact structure isn’t enough for a serious industrial knowledge base. The architecture has to be capable of growing with the business.

Designing for years, not months

This is why I’m approaching the Manufacturing Marketing Agency project as a long-term information system, not a collection of 20 or 30 pages, not a temporary SEO project, and not a pile of keywords converted into URLs. The model needs to support hundreds, potentially thousands, of interconnected pieces of information over time: new industries, new technologies, new applications, new solutions, new services, and new case studies that create new relationships between existing entities. The architecture should allow all of this without having to rebuild the entire system every time the content grows.

The technical implementation comes later

Only after the conceptual model is defined do I move toward the technical implementation. In this project, that means Next.js, React, and Payload CMS. The CMS will eventually represent the entities, content types, categories, relationships, and reusable structures defined by the model, but the technology isn’t the starting point. The starting point is understanding the domain.

The sequence is domain, conceptual model, entities, taxonomy, relationships, keyword space, content architecture, URLs, and only then technical implementation. That’s the order I’m interested in, because a website isn’t just a collection of pages. A serious website is an information system, and for a serious manufacturing company, the information system itself becomes part of the company’s digital infrastructure.

The interesting part is what happens next

Once the four dimensions, Industrial, Technology, Solutions, and Services, are connected through relationships, something much larger starts to emerge. The website is no longer simply organized, it begins to represent a model of the manufacturing domain. That model can become the foundation for content, SEO, internal linking, structured data, search visibility, AI-assisted retrieval, and eventually a much broader digital knowledge system.

The four dimensions are only the beginning. The real architecture is in the relationships between them.

Marian Polyak
Web Developer | SEO & AI Search Optimization | Digital Marketing
Manufacturing & Industrial Marketing