How to Build a Website Structure Based on Real Demand (Not Opinions)

From client inputs to structure: step by step to a sitemap that fits the market

From Data to Design: Building for Users, Not Guesswork

• Step 1 – Brainstorm (DO NOT SKIP)
• Step 2 – Suggest Seed Keywords
• Step 3 – Clean and De-Duplicate
  • Step 3b – Create Seed Keyword List for Target Markets (International SEO)
• Step 4 – Expand With Keyword Tools
• Step 5 – Clean Expanded List
• Step 6 – Cluster Keywords (ChatGPT prompt)
• Step 7 – Visualize the Structure
  • Hacks for Quick Visualization
• Afterthoughts

Most site structures are made up. A few people sit in a room, argue about what “feels right,” and call it a sitemap. Google doesn’t care. Users don’t either.

This walkthrough shows how to turn a messy keyword dump into a sitemap mockup driven by real search demand. No sticky notes. No brainstorm sessions about “what the customer wants.” Just data, filtered and organized step by step.

Step 1 – Brainstorm (DO NOT SKIP)

This is the most underestimated part. If you do it properly, you’re already ahead of 90% of your competitors.

Most people just sit in front of SEMrush, type a couple of “obvious” keywords, and stop there. They’ll build a whole site structure on top of those five generic terms everyone else is chasing.

Ask your client-facing team. Write down the answers exactly as customers say them:

Rotate this questionnaire among your client-facing team. Gather responses. Use Google Forms or a similar tool.

Step 2 – Suggest Seed Keywords

Now take the raw answers from your client-facing team and turn them into seed keywords.

Rules:

1. Use only the responses. Don’t invent anything, don’t “improve” wording. If jargon shows up in the answers, keep it. That’s part of the data.
2. Keep them short (2–5 words). Seeds are anchors, not long-tails.
3. Write them down raw. Don’t optimize or refine.
4. Extract seeds from all responses. Don’t filter now. You’ll clean later.

Examples:
Response from sales team – a B2B client example:

• “Everyone asks if we can replace their Excel sheets.”
  → Seed keyword: excel replacement

Response from support team:

• “They keep saying ‘catalog’ when they mean product database.”
  → Seed keyword: product catalog

Response from discovery calls:

• “Most leads say they’re struggling with images and PDFs all over the place.”
  → Seed keyword: digital asset management

Response from account managers:

• “We get compared to SAP a lot, but customers actually just mean integration with ERP.”
  → Seed keyword: erp integration

At this stage you’re not curating. You’re just pulling crude anchors out of messy conversations. The cleaning comes later.

Step 3 – Clean and De-Duplicate

Every keyword gets judged. KEEP, APPEND, or DROP.

KEEP → Already clear and specific.

• “sap erp integration” → stays.
• “inventory forecasting software” → stays.
• “multi-currency accounting” → stays.

APPEND → Too generic on its own, but salvageable with a modifier.

• “integration” → erp integration
• “forecasting” → demand forecasting
• “module” → finance module or hr module
• “system” → inventory system

DROP → Too vague, double-edged, or irrelevant.

• “platform” → meaningless without anchor.
• “solution” → could be anything.
• “erp jobs” → wrong intent.
• “oracle vs sap” → competitor battleground, not a structural keyword.

Also: strike duplicates. You don’t need “erp integration” listed ten times in slightly different word orders.

This is where you save yourself from future chaos. Leave “system” in raw form, and you’ll wake up to a sitemap with an entire “System” page. Drop or fix it now.

At this point, you should have a cleaned keyword list that:

• Has no duplicates.
• Uses modifiers to disambiguate vague terms.
• Has dropped anything too generic to ever become a page.

Now you have a list of unique, unambiguous seed keywords. Safe to proceed.

Step 3b – Create Seed Keyword List for Target Markets (International SEO)

If international expansion is on the roadmap, now is the time to create equivalent seed keyword lists in your target languages.

Why now (and not later)?

• Translating raw keyword dumps = junk in, junk out.
• Translating finished clusters = you’ll miss equivalents that only show up earlier in the chain.
• Translating the cleaned seed list = just enough precision without clutter.
• This way you’re not blindly translating everything, but you’re also not waiting until it’s too late.

Important:

• Don’t translate full long-tails.
• Always pull multiple equivalents per seed (different countries/regions often use slightly different terms).

Prompt you can use:

Step 4 – Expand With Keyword Tools

Now – and only now – you have a list that’s safe to feed into SEMrush, Ahrefs, or whatever tool you use. Before this cleaning, dumping raw junk into a tool would just inflate the noise.

I recommend Marketing Miner because it’s fast and pragmatic, but pick your poison.

The goal here isn’t volume – it’s to inflate your options with keywords that are related and still relevant.

Don’t forget to specify your target market here (especially important if you have seed keyword equivalents in multiple languages from the earlier step).

Good sources to pull from:

• Related search (Google’s “searches related to…”).
• Autosuggest (type in your seed and scrape the dropdown, not just from Google but Bing, YouTube, Amazon if relevant).
• Broad match expansions (variants and phrasing).

Don’t just collect everything again. Use your cleaned list as anchors – only expand from those.

Example outputs:

Seed keyword: inventory forecasting

• inventory forecasting in sap
• inventory forecasting software for manufacturing
• demand forecasting vs inventory forecasting

Seed keyword: erp integration

• erp integration with shopify
• erp integration best practices
• erp crm integration examples

Seed keyword: multi-currency accounting

• multi-currency accounting in erp
• multi-currency accounting software for mid-size companies

Seed keyword: hr module

• erp hr module features
• erp hr module vs standalone hr software

Seed keyword: supply chain erp

• supply chain erp systems for manufacturers
• supply chain management module in erp

Step 5 – Clean Expanded List

Now that you’ve inflated your keyword list with tool suggestions, clean it again. This step is faster because your seeds were curated – but some irrelevant or redundant terms always sneak in.

Mostly fine / keep as-is:

• inventory forecasting in sap ✅
• erp integration best practices ✅
• multi-currency accounting software for mid-size companies ✅
• erp hr module features ✅
• supply chain erp systems for manufacturers ✅

Nuanced drops (crossed out):

• erp jobs integration → irrelevant intent (“jobs”).
• sap vs oracle erp → competitor battle, not your structure.
• erp integration pdf → people looking for a file, not a service/product.
• free erp hr module → wrong audience if you sell enterprise solutions.
• inventory forecasting excel template → adjacent but not your offer.

Deduplication:

• If you see both erp crm integration examples and crm erp integration examples → keep one, drop the other.

The list is now unique, relevant, and scoped to real demand.
👉 At this point, you’re sitting on the building blocks of your website navigation structure.

Step 6 – Cluster Keywords (ChatGPT prompt)

This is where your keyword list starts turning into a site structure. Tools can expand lists, but they can’t tell you what belongs together. If you skip clustering, you’ll end up with random pages instead of a coherent sitemap.

Here’s a lean prompt you can drop straight into ChatGPT (or hand to your team):

Clustering is the moment where the noise resolves into structure. After this step, you’re no longer looking at a keyword dump – you’re looking at the blueprint of a website.

Example output of this stage:

Step 7 – Visualize the Structure

Now that you’ve got categories, you can finally see the website taking shape.

Important: the big labels (Integration, Accounting & Finance, Industries) are just organizational buckets. They help you think straight, but they don’t need to show up on the site as standalone pages.

What might become the actual website page structure are the subcategories.

Example Visualization (Outline/Tree):

Use a mindmap or tree tool (Miro, Whimsical, XMind, draw.io) if you want it prettier. But even in plain text. This outline shows your website navigation structure, what’s ready to be turned into content.

Hacks for Quick Visualization

1. Mindmapping / Tree Tools with CSV Import

Whimsical, Miro, XMind let you paste structured text (outline with tabs or bullets) → it auto-renders as a tree/mindmap.

Example:

→ paste into XMind → instant silo structure.

Xmind example: after importing a structured list of topics

2. Graph Visualization (Quick + Dirty)

Export categories/subcategories to a CSV → use draw.io or yEd Live to auto-generate a hierarchical sitemap tree.

Format:

→ import → instant HTML graph.

Afterthoughts

Can it be simplified? Yes – but at a cost. Most competitors skip half these steps, and that’s why their “keyword research” ends up as a random list in SEMrush with no structure. If you want the edge, you do the work.

Can it be automated? Yes – you can throw clustering and cleaning into scripts or tools. But don’t kid yourself: automation saves time, it doesn’t replace judgment.

Is this oversimplified? Also yes. A full keyword audit would layer in search volumes, opportunity scores, CPCs, competitive gaps, and cluster by priority. That’s how you decide not just what belongs where, but what deserves investment first.

But if you’re starting out, this model is enough. It gives you the best market-friendly site structure grounded in demand – not opinions. And that alone already puts you ahead of 90% of your competitors.
And this is true for any business including local, ecommerce or SaaS.

If you want a full keyword audit that uncovers the 90% of untapped opportunities your competitors are missing, get in touch.