From Keyword Spreadsheet to Website Decisions: What Good Research Changes

A useful keyword study should change what a business does next. Here’s how to turn search-demand evidence into clearer page, architecture, content and commercial decisions.

You’ve been handed a keyword spreadsheet. It has thousands of rows, search volumes, competition scores and perhaps a few coloured tabs that suggest somebody worked very hard.

Then comes the awkward question: what should change on the website on Monday morning?

A large spreadsheet can contain accurate data and still be operationally useless. Good keyword research should help you decide which pages matter most, whether the site reflects how customers search, which language to use, what content is genuinely missing and where commercial attention is justified.

The useful test is simple:

Useful research = evidence + decision + owner + validation.

That is a Plus IQ framework, rather than a Google requirement or scientific law. It gives you a practical way to judge whether a keyword study is ready to guide action. Every important finding should answer four questions:

• What does the evidence show?
• What decision might it change?
• Who owns that decision?
• How will we check whether it was sensible?

Without those answers, research becomes an impressive archive of terms rather than a plan for the website.

Search demand is evidence, not an instruction

A search query can tell you something about what people are trying to find. It does not automatically tell you what page to create, what label to put in the navigation or whether your business should offer the thing being searched for.

Search research has long distinguished between the wording of a query and the user’s underlying goal. The same phrase can mean different things in different contexts, and a query is not a complete specification for a web page. The classic distinction between informational, navigational and transactional goals is useful background, but it does not make the final website decision for you. Broder’s search-goal taxonomy and later research on understanding user goals support that distinction.

Keyword categorisation, intent mapping and clustering can help organise the evidence. They are inputs to the decision, not the decision itself. We cover those analytical steps in more detail in our guide to keyword categorisation and intent mapping.

Search volume has a similar limitation. Google describes keyword-planning figures as historical metrics or estimates used to understand demand and plan advertising activity. They can help compare opportunities, but they are not forecasts of organic traffic, leads or revenue. Google’s historical-metrics documentation and its guidance on Keyword Planner statistics are useful reminders here.

In practice, a high-volume term may be broad, ambiguous, poorly matched to your offer or dominated by search features. A lower-volume term may represent a specialist need that is much more commercially relevant. The number is a comparison point, not a command from the search engine gods.

Decision one: which pages deserve attention first?

Most businesses do not have the time or development capacity to improve every page at once. Research becomes useful when it creates a defensible order of work.

Evidence to gather

Look for a combination of:

• search demand and how consistently it appears across related terms;
• the current page’s visibility, impressions, clicks and position;
• the page’s role in the customer journey;
• commercial relevance, such as product availability, service capacity or strategic importance;
• the likely effort and risk of making the change.

Search Console can provide query, page, impression, click, click-through rate and position data for existing visibility. Google’s Search Console performance documentation explains what these reports contain.

That data is useful, but it is not a complete record of demand. Google documents anonymisation, truncation and other reporting limitations, and query data may be associated with the canonical URL rather than every URL that appeared in a search. Search Console’s data limitations explain why an absent query should not be treated as proof that nobody searches for it.

Possible actions

You might prioritise:

• a commercially important page that already receives impressions but attracts the wrong visits;
• a page sitting just outside strong visibility where clearer content or internal linking may be worth testing;
• a high-value section whose search demand is substantial but whose current pages are thin, outdated or difficult to find;
• a small set of pages where one template improvement could affect many relevant URLs.

This is different from sorting the spreadsheet by volume and starting at the top. A page with 20,000 estimated monthly searches is not automatically the best first project if the business cannot fulfil the demand, the query is ambiguous or the required change carries substantial migration risk.

What changes for the business?

The result should be a prioritised page portfolio, not a promise of traffic. It should show where the organisation is investing attention, where it is testing an improvement and where it is deliberately waiting.

That gives marketing, content, product and development teams a shared list. It also makes trade-offs visible. If a low-volume page supports a strategically important product, it can be prioritised for reasons that search volume alone would never show.

How to validate the choice

Record the page’s starting position, impressions, clicks and relevant business measures before the change. After implementation, review visibility alongside engagement, enquiries, sales or another suitable outcome.

Do not treat a ranking or click increase as proof that the decision was commercially correct or caused the result. Seasonality, promotions, competitors, algorithm changes, stock, service availability and tracking changes can all affect performance. Search Console can help describe what happened in search, but it cannot explain the whole customer journey. Google’s guidance on connecting Search Console and Analytics is a useful starting point for combining those perspectives.

Decision two: does the site structure reflect how people find the business?

Keyword research often reveals that the words used internally are not the words customers use. It can also show that an important topic is buried several clicks away, labelled unclearly or missing from the visible path altogether.

That does not mean search demand should dictate the entire navigation. A site also has to reflect how products work, how customers understand the offer, how the business fulfils it, and what is usable and accessible. Google’s ecommerce structure guidance supports clear relationships between pages, while its SEO Starter Guide cautions against changing a site’s structure unnecessarily. Search evidence should prompt an architecture review, not an automatic rebuild.

Evidence to compare

Set the researched demand alongside:

• the current navigation and category labels;
• internal links and the paths customers use to reach important pages;
• site-search behaviour, if available;
• customer-service questions and on-site feedback;
• the relationship between pages, products, services and supporting information.

Google documents that internal links help it understand the relationships between pages and may contribute to its assessment of relative page importance. Its guidance on ecommerce site structure supports reviewing links and hierarchy, but it does not guarantee a particular ranking outcome from a particular link change.

What might change?

The answer might be a clearer navigation label, stronger contextual links, a revised supporting page or a better route from a broad section to a specific need.

It might also be no structural change at all. If customers already understand the navigation and the pages are easy to reach, rebuilding the hierarchy to mirror a keyword list could introduce unnecessary risk. Our article on how search demand should influence site navigation explores that balance.

What is the business consequence?

A good architecture decision should make it easier for customers to move from a recognised need to the relevant part of the offer. It may also help teams agree which pages are central, which are supporting and which should not be promoted as primary destinations.

Research should make the relationship between demand and the existing site easier to see. It should not turn the navigation into a list of popular phrases.

How to validate the change

Check whether users can find the relevant pages through the revised path. Review internal-link discovery, clicks from navigation, search visibility and behavioural evidence where it is reliable. For a larger change, test a smaller section first rather than moving the whole site in one dramatic afternoon.

Decision three: what language should customers see?

Search data can reveal the phrases customers use when the business uses different words. That can inform headings, navigation labels, explanatory copy, filters and product or service descriptions.

Language matters partly because people use it to judge whether a path is likely to lead to what they need. Information-foraging research describes this as “information scent”: cues help users decide whether continuing along a path is worthwhile. Research on information scent and work on information architecture and findability support the broader principle.

Evidence to check

Compare search wording with customer-service language, sales conversations, site-search terms, existing page copy and user research. Look for repeated differences rather than treating one unusual phrase as a rebrand proposal.

Check what the phrase means in context, too. A popular term may describe a product the business does not sell, a service it cannot provide or an outcome it cannot honestly promise.

What might change?

You might change a label, introduce a plain-English explanation, add a recognised synonym or make a customer-facing phrase more prominent. You do not need to force an exact target keyword into every page. Google’s people-first content guidance emphasises useful, accurate content for people rather than mechanical wording. Google’s helpful-content guidance provides the relevant principle.

What is the business consequence?

Clearer language can reduce the gap between what customers ask for and what the business says it provides. It can also expose a positioning problem: perhaps customers are looking for an outcome while the site talks only about an internal product category.

That does not mean abandoning established terminology. Legal, technical, brand and accessibility requirements may all matter. Sometimes the best answer is to use both terms, with an explanation, rather than replacing one with the other.

How to validate the wording

Review whether users engage with the revised labels or copy, whether site-search refinements change, and whether relevant search impressions and clicks improve. Use qualitative feedback where possible. A phrase that gains impressions but causes customers to expect an unavailable service is not a successful language decision.

Decision four: is there a meaningful content gap?

A keyword absent from the website is not automatically a content gap. The real question is whether customers have a need that the current site does not adequately serve, and whether the organisation can answer it accurately and usefully.

A synthetic example: an online learning platform

Imagine an online learning platform reviewing this illustrative set of terms:

• “project management course”;
• “online project management course”;
• “project management course for beginners”;
• “agile project management certification”;
• “how to become a project manager”.

This is a synthetic example, not client data and not a forecast of performance.

A spreadsheet might assign each term a volume and suggest a page for each. A decision-led review asks different questions:

• Are people looking for a general course, a beginner-friendly route, a recognised certification or career guidance?
• Does the platform actually offer each of those things?
• Is the existing course page clear about audience, format, assessment and outcome?
• Would a career guide answer a genuine need, or would it simply attract people the platform cannot help?

The outcome might be one improved course page with clearer language for beginners, a separate certification explanation because the offer and decision are materially different, and a carefully scoped career guide that explains the available routes without pretending to provide a qualification.

Several terms have informed the plan. None has automatically demanded its own URL. The research has changed the page information, content plan and commercial message.

Assessing the gap

Review the current page against the need implied by the search. Consider the quality and usefulness of existing content, customer questions, competitor results, internal expertise and the business’s ability to maintain an accurate answer.

Possible outcomes

You might improve an existing page, add genuinely useful supporting content, change the content format or decide not to publish. Producing a page for every researched term can create overlap, thin content and maintenance debt.

A gap may also point to a product or service issue rather than a writing task. If customers consistently want an option the business cannot provide, publishing a thin article about it will not solve the underlying problem.

How to validate the work

Define the job the content is meant to do. Then review relevant search visibility, engagement, assisted journeys, enquiries or enrolments, depending on the page’s role. Also check content quality: did the new page answer the need clearly, or did it simply repeat the spreadsheet in prose?

Decision five: what should be commercially prioritised?

Keyword research can surface opportunities, but it cannot independently establish market size, profitability, capacity, customer lifetime value or operational feasibility. Those questions need commercial evidence as well.

Evidence to combine

Consider search evidence alongside:

• strategic importance;
• margin or customer value;
• availability and fulfilment capacity;
• service coverage or geographic reach;
• seasonality and existing demand;
• implementation effort and risk.

Volume and competition can be useful comparison signals, while remaining insufficient on their own. Google’s documentation on historical keyword metrics and Keyword Planner statistics illustrates why these figures should not be treated as complete commercial forecasts. A query with strong demand is not automatically attractive if the business cannot serve it profitably.

What might follow?

The result could be a development project, a content brief, a product-positioning discussion, a stock or service-capacity review, or a decision to monitor rather than invest.

It should also include deliberate exclusions. A strong study can say, “We are not creating pages for these terms because they do not match the offer,” or, “We will monitor this emerging topic until we have a credible response.” That restraint is part of prioritisation, not a failure to find enough opportunities.

What is the commercial consequence?

The organisation gets a clearer reason to invest. It can distinguish a search opportunity from a business opportunity and avoid promising traffic or revenue figures that the evidence cannot support.

How to validate the priority

Agree the commercial measure before implementation. It might be qualified enquiries, completed purchases, applications, product engagement or a service-capacity measure. Then review search performance alongside that measure, with enough context to account for promotions, seasonality, availability and other changes.

What a useful research output should contain

If you are reviewing an existing keyword study, ask whether each major finding contains the following:

• Evidence: Which queries, pages, trends or customer signals support the finding?
• Interpretation: What might customers be trying to do, and how confident are we?
• Decision: What should change, stay as it is or be investigated?
• Non-decision: What are we deliberately not changing, and why?
• Owner: Is this for content, UX, development, product, commercial or another team?
• Dependencies: Does it require stock, legal review, analytics work, engineering time or stakeholder agreement?
• Risk: Could the change create confusing navigation, overlapping pages, migration problems or unsupported promises?
• Validation: What will we review, when will we review it and what would count as a useful result?

A recommendation such as “create a page for keyword X” is incomplete. A more useful recommendation might say: “The current course page does not clearly address beginners, although several related searches indicate that audience. The content lead owns a revised section and clearer navigation label. We will review visibility, engagement and enrolment quality after launch, while keeping the existing URL unless evidence shows a separate user need.”

That recommendation is more modest than a spreadsheet full of page assignments. It is also far more likely to survive contact with the website.

When research points in different directions

Good evidence will not always agree with itself. Search demand may suggest one label while customer research favours another. A commercially valuable service may have little visible search demand. A popular query may describe an offer the organisation cannot fulfil.

That is normal. Search research is one input into a decision, not a replacement for customer understanding, product logic, accessibility, legal review or commercial judgement.

In those situations, record the conflict rather than hiding it. State which evidence carries more weight for this decision and why. You may decide to use a customer-friendly label with a search-recognised phrase in supporting copy, run a small test or monitor demand without changing the architecture.

The same principle applies to low-volume and emerging demand. A universal minimum-volume threshold can remove valuable specialist or regional opportunities. Equally, a handful of unusual terms should not create a forest of pages nobody can maintain.

The real test: what changes next?

A keyword study has done its job when people can use it to make and record decisions. That might mean:

• prioritising a set of commercially important pages;
• reviewing navigation and internal links without rebuilding the site unnecessarily;
• using customer language more clearly and accurately;
• addressing a genuine content or product gap;
• deciding which opportunities justify investment;
• monitoring, consolidating or declining work that does not have a strong case.

The spreadsheet may still be useful. It can hold the underlying terms, metrics and supporting analysis. But it should sit behind a decision record, not stand in for one.

For a small site, an in-house team may be able to make these calls quickly. At larger organisations, the difficulty is usually not finding more keywords. It is reconciling conflicting evidence, assigning ownership and implementing changes safely across content, UX, development and commercial teams. That is where specialist support can help: diagnosing the evidence, prioritising the decisions and carrying them through to validation.

Useful keyword research does not leave you with more data and the same website. It leaves you with agreed actions, clear owners and a sensible way to find out whether those actions were worth taking.