Techosoft
Resource Centre
Keyword research has a problem. It can tell you what people search for, how often they search and which terms appear commercially valuable. But it does not always tell you what people are actually trying to understand.
That gap is becoming more significant as search becomes more conversational. Someone researching a product or service may type a short query into Google, then ask an AI search engine a much longer question containing their situation, constraints, concerns and expectations.
The smarter approach is to combine keyword research with prompt research. Keyword data provides measurable demand. Prompt research adds context around that demand. Used together, they give content teams a stronger basis for deciding which topics deserve investment, which questions belong on existing pages and which gaps justify new content.
AI search has prompted plenty of predictions that traditional keyword research is becoming obsolete. That is overstated. Keyword research remains useful for understanding search demand, search intent, commercial opportunities, query variations, seasonal behaviour, geographic demand, competitive visibility and existing content gaps.
The mistake is treating search volume as the final decision-maker. A keyword with 20,000 monthly searches may attract broad informational traffic but generate little commercial value. A 300-search query might repeatedly appear before qualified enquiries — the second could be far more important to the business.
A good SEO strategist needs to ask more than “how many people search this term?” The better question is: what decision is the customer trying to make when they search it? That is where prompt research adds another layer.
Keyword research usually compresses intent into a few words. Prompt research can expose the context behind those words.
Consider “best CRM software.” Now compare it with: “We’re a 20-person B2B sales team using spreadsheets. Which CRM options would give us automation and reporting without the cost and complexity of an enterprise platform?”
The second question reveals business size, current solution, pain point, desired capabilities, budget sensitivity, buying stage and evaluation criteria — all in one line. Traditional keyword research might identify those subjects separately. A conversational prompt brings them together.
That matters because content should not only target what someone searches. It should help them resolve the decision behind the search.
The two datasets answer different questions:
| Keyword research | Prompt research |
|---|---|
| What are people searching? | How are they framing the problem? |
| How much measurable demand exists? | What context surrounds the question? |
| Which queries recur? | Which concepts appear together? |
| What is the likely intent? | What information is needed to decide? |
| Which terms have commercial value? | What objections and follow-ups emerge? |
Begin with conventional research. Build clusters around core commercial terms, informational searches, long-tail queries, pricing questions, comparison searches, product-specific searches, problem-based searches and location-based searches. Then group them by intent.
For example, a software company might find: Core — project management software; Commercial — best project management software for small business; Pricing — project management software cost; Comparison — Asana vs Monday for small business; Problem — how to manage projects across multiple teams.
That gives you the initial topic territory. Now comes the more interesting part.
Do not ask an AI tool to generate 100 prompts around a keyword. That usually creates a large list with little strategic value. Instead, investigate the decisions surrounding the topic. Ask:
You are not trying to manufacture content ideas. You are trying to understand the information journey around a topic.
Suppose keyword research identifies “enterprise CRM software.” Prompt research may expose questions such as which CRM features actually matter for large sales teams, what an enterprise implementation should include, how long migration takes, what happens to existing customer data, what costs are commonly overlooked, how competing platforms should be evaluated, and when switching becomes worthwhile.
These should not automatically become seven separate articles. They represent a broader decision: how should an enterprise buyer evaluate a CRM? That could become one authoritative resource supported by relevant product or service pages. This is one of the biggest benefits of combining the two research methods.
Content teams can easily create unnecessary URLs when every question becomes a new article. Instead, cluster prompts around the underlying problem. For example:
Supporting questions could cover features, implementation, pricing, migration, alternatives, procurement criteria and common mistakes. Some questions may belong directly within a pillar resource. Others may justify separate pages if they have distinct intent, sufficient depth or strong commercial importance.
The goal is not to maximise the number of URLs. The goal is to cover the customer’s decision thoroughly without creating unnecessary duplication.
A topic with high search volume but no meaningful expertise behind it may be a weaker investment than a smaller topic where your business has something genuinely valuable to contribute.
| Topic | Demand | Prompt depth | Business value | Content gap | Expertise | Total |
|---|---|---|---|---|---|---|
| Topic A | 5 | 5 | 5 | 4 | 5 | 24 |
| Topic B | 5 | 2 | 2 | 3 | 4 | 16 |
| Topic C | 3 | 5 | 5 | 5 | 5 | 23 |
The numbers are not universal. A B2B company may weight commercial value heavily. A publisher may give more weight to audience demand. A specialist service business may give expertise a higher score. The value of the model is that it forces the team to explain why a topic deserves investment.
There is a predictable mistake here. Marketers discover prompts and immediately ask “how many people searched this exact prompt?” In many cases, reliable prompt-volume data simply does not exist. AI prompts can be private, fragmented across platforms and constantly changing.
Not every content gap requires a new URL. Suppose a service page already ranks reasonably well for its primary topic but fails to answer obvious questions around cost, suitability, process, alternatives or expected timeframe. Those answers may belong directly on the service page.
Other topics may deserve standalone resources. A detailed comparison might require its own page. A technical implementation guide could become supporting content. A short suitability question may need only a few useful paragraphs on the main commercial page. Let information architecture follow user needs, not keyword spreadsheets.
A conventional SEO brief often contains a primary keyword, secondary keywords, search volume, competitor URLs, word count and suggested headings. That is useful, but incomplete. A stronger brief should also contain prompt evidence:
Primary keyword cluster, related queries, search intent, SERP observations, commercial relevance.
Common questions, comparison scenarios, customer objections, decision criteria, follow-up questions.
Subject-matter expert, first-hand examples, original data, evidence requirements, claims requiring verification.
Questions that must be answered, gaps competitors have missed, appropriate content format, internal links, supporting pages.
Prompt research is not an invitation to stuff questions into headings. Writing “What is X? Is X worth it? What are the benefits of X? Should you use X?” does not make content more useful.
Understand the questions, then answer them naturally. If customers want to know whether a solution is suitable for a particular business size, explain the factors that determine suitability. If they want a comparison, compare the options properly. If they want pricing, explain what drives the cost. The prompt is research evidence, not necessarily copy.
Keyword and prompt research tell you what people want to know. Your business data can tell you what happens next. Where possible, connect topic performance with organic conversions, lead quality, revenue, product enquiries, demo requests, contact submissions, assisted conversions and sales feedback.
A high-traffic informational topic may have value even if it does not convert directly — it may introduce a new audience or influence a later purchase. But you should know what role it is playing. That is where a disciplined SEO campaign management process becomes valuable: topic priorities should be reviewed against actual performance rather than left untouched after publication.
Keywords are evidence of demand, not automatic content requirements.
This is the same problem in a new format. Cluster questions around meaningful user problems.
Volume tells you scale, not profitability.
AI can identify patterns quickly, but it does not automatically understand your customers, margins, sales objections or competitive strengths.
A strong existing page may need five useful paragraphs rather than a new article.
A ranking is not a business outcome. Track traffic, engagement, leads, revenue and assisted conversions where possible.
A workable process looks like this.
Build the keyword universe. Core keywords, long-tail searches, commercial and informational queries, comparison terms, pricing searches, Search Console data, existing ranking data.
Build the prompt universe. Conversational questions, customer objections, sales and support questions, comparison scenarios, pre- and post-purchase concerns.
Cluster both datasets. Group them around problems, products, services, decisions, audiences and buying stages.
Topic research should not end when the article goes live. Effective SEO campaign management requires ongoing review because priorities change: search behaviour can shift, products change, competitors publish new material, customer objections evolve, AI search changes how people formulate questions, and your own conversion data can also change the value of a topic.
For important commercial categories, a quarterly review is a sensible starting point. Faster-moving industries may need more frequent reassessment.
Combining keyword and prompt research can support stronger E-E-A-T, but research alone does not establish expertise. The content still needs credible evidence of knowledge: first-hand experience, expert review, original research, accurate sourcing, practical examples, transparent methodology, clear authorship, updated information and honest discussion of limitations.
If most answers are yes, the topic has a strong case for investment. If the only justification is high search volume, keep researching.
Keyword research is not becoming irrelevant. It is becoming one part of a larger evidence base: search data tells you what people type, prompt research reveals how they frame complex questions, sales teams explain what prospects worry about, customer-support teams reveal where explanations fail, analytics shows what leads to action, and subject-matter experts provide the knowledge that makes the content credible.
The strongest topic is not necessarily the keyword with the highest volume. It is the topic where demand, intent, business value, information gaps and genuine expertise intersect. That is a much better basis for deciding what deserves your team’s time.
No. AI summaries and chatbot answers still draw heavily on well-structured, well-optimised, crawlable content. Traditional SEO fundamentals remain the foundation that AI visibility is built on top of — not a separate or outdated discipline.
Run your most important target queries through tools like ChatGPT, Gemini, and Copilot directly, and note whether your brand or content gets referenced. Some SEO platforms are also starting to offer dedicated AI visibility tracking.
Pew’s report doesn’t dig into the reasons behind this gap, only the size of it (63% versus 57%). It’s worth noting as a data point, but not something to draw firm conclusions from without more research.
Yes, to a degree. With 38% of employed adults using chatbots for work tasks, it’s reasonable to assume some portion of B2B research and vendor comparison is happening inside chat tools. Brands with no presence in content that chatbots might reference risk being left out of that research phase entirely.
ChatGPT, given its 44% adoption rate and continued growth, is the clear priority. Gemini is a reasonable second focus, with Copilot and Meta AI worth monitoring as well.
