How to Find Keywords That Drive Sales: Conversion Focused SEO Keyword Research

Most keyword lists I inspect share one flaw: they rank words by search volume and hope sales follow. In my experience since 2012, that hope rarely pays. Conversion focused keyword research works differently, because it weighs buying intent, reachability and page fit before it looks at volume. In this guide I explain the exact method I use with clients in Istanbul and abroad. Follow it step by step and apply it to your own site.

How do you do keyword research that drives sales?

Sales driven keyword research follows four moves: classify every query by intent, measure how reachable the first page is for your domain, map each keyword to one page type and funnel stage, then track conversions and prune what fails. In short, money intent multiplied by reachability multiplied by page fit predicts revenue; volume is only a tie breaker.

I run this process for every SEO consulting engagement before a single page gets written. However, the order matters more than the tools. First, I collect the language customers actually use. Then I expand it with tool data, verify intent on the results page, score the list and assign pages. Finally, I measure for 90 days and cut what does not convert.

• Collect customer language from sales calls, support tickets and site search.
• Build a seed list and expand it with a research tool.
• Tag every keyword with an intent class.
• Verify intent by reading the live results page.
• Score and rank by money intent, reachability and page fit.
• Map each keyword to a page and schedule production.
• Measure conversions, then prune or promote.

Why do high volume keywords so often fail to bring sales?

High volume keywords usually sit far from the purchase decision. Someone searching "what is web design" wants a definition, not a quote. As a result, even a first place ranking for that query brings readers who leave without contacting anyone. The traffic looks good in a report and does nothing for revenue. Volume led keyword research produces exactly this outcome.

The numbers around traffic itself are sobering. Ahrefs studied about 14 billion pages in its Content Explorer index and found that 96.55 percent of pages get no traffic from Google, while another 1.94 percent get one to ten visits a month. The three main causes: no search demand, no backlinks, and content that mismatches search intent.

Position also compounds the problem. Backlinko's analysis of four million search results shows the first organic result takes 27.6 percent of clicks, roughly double the second. Therefore a broad keyword where you land in seventh place delivers a fraction of what a narrower keyword in first place delivers. The narrower one usually converts better too.

What are the types of search intent, and which ones make money?

Search intent falls into four classes: informational, navigational, commercial and transactional. Tools such as the Semrush Keyword Magic Tool label each keyword with one of these letters, I, N, C or T. You can then filter by them. For conversion focused keyword research, commercial and transactional queries take priority, because the searcher already compares options or wants to buy.

Informational queries still matter, but their job differs. They build trust and feed remarketing audiences instead of closing sales directly. Navigational queries, meanwhile, belong to brands, so chasing someone else's brand name wastes effort. In practice I keep roughly this ratio in the first plan: two thirds C and T keywords, one third I keywords that support them.

Why is keyword research more profitable when it starts with customer language?

Customers rarely search the way marketers write. A tool will hand you "ecommerce consulting", yet the person on the phone asks "why does my Shopify store get visits but no orders". That second phrase carries a buying problem and almost no competition. For this reason my keyword research starts with words spoken by real customers, not with a tool.

The sources are already inside your business. For example:

• Sales call notes and the objections that repeat.
• WhatsApp and email questions from the last six months.
• Internal site search logs, if your site has a search box.
• Support tickets, especially the first sentence of each one.
• Reviews on Google and marketplaces, including competitors' reviews.

Once I have thirty to fifty such phrases, I group them by problem. Each group becomes a seed keyword. Then the tool expands those seeds. This sequence keeps the research anchored in money problems. So the final list rarely fills up with definitions nobody pays for.

How do you pull sales driving queries out of Search Console?

Search Console holds the cheapest keyword research data you own, because it shows queries you already rank for. However, you see only part of the picture. Ahrefs examined 22 billion clicks across 887,534 properties and found that 46.77 percent of clicks fall under anonymized queries in the April 2025 data. Roughly half of real demand stays hidden in the long tail.

Google's own help documentation confirms that some queries leave the report for privacy reasons, that the Performance report keeps 16 months of history, and that a branded versus non branded filter exists since March 2025. For conversion work, the non branded view is your opportunity list. Here is the five step routine I use.

1. Open Performance, set the range to the last 12 months, and apply the non branded filter.
2. Filter positions between 4 and 15, because those queries already have relevance but not clicks.
3. Sort by impressions and export the top 500 rows.
4. Tag each row with an intent letter and drop pure informational rows for now.
5. Match the remaining queries to the page currently ranking and check whether that page can actually sell.

The final check matters most. If a blog post ranks ninth for a transactional query, the fix is often a dedicated service page, not more words on the post.

How should you read Google Keyword Planner data for conversions?

Keyword Planner tells you more about money than about volume, which makes it useful for keyword research aimed at conversions. According to Google's help page, average monthly searches appear only for exact match and include close variants, while click and cost forecasts account for match type, bid, budget and seasonality. Without active spend, volumes show as ranges such as 100 to 1K, so treat them as bands, not facts.

The column I read first is the top of page bid. Advertisers pay high bids only where clicks turn into revenue. Therefore a high bid is a money intent signal that no free tool replicates. In practice a query with a 1K to 10K band and a strong bid outranks a 100K query with a bid near zero on my scoring sheet.

Also compare the low and high bid estimates. A wide gap usually means a mixed auction with both cheap informational clicks and expensive buyer clicks, so the keyword needs a modifier before it becomes useful. A narrow, high range signals a clean commercial query.

Which modifier words signal high commercial value?

Modifiers are the small words that reveal where a searcher stands in the buying journey. Adding them to a seed keyword turns a vague topic into a sales driving query. For example, "SEO" is a topic, while "SEO agency pricing" is a request for a quote. I keep a running modifier list per language for keyword research and add to it after every customer interview.

The strongest English modifiers I see in client data:

• Price signals: pricing, cost, how much, quote, rates, packages.
• Comparison signals: best, top, vs, alternative, review.
• Provider signals: agency, consultant, company, freelancer, expert.
• Location signals: near me, in London, Istanbul, local.
• Urgency signals: same day, fast, emergency, available now.
• Specification signals: for small business, for Shopify, B2B, enterprise.

In Turkish the equivalents include fiyat, ücreti, ne kadar, en iyi, firması and teklif al. German buyers type Kosten, Preise, Anbieter, Agentur and Angebot. Because these words differ by market, never translate a modifier list; instead, rebuild it from local searches and local sales conversations.

Do long tail and zero volume keywords bring sales?

Yes, often, and the reason is structural. Google has repeatedly stated that 15 percent of daily searches have never been seen before, a figure it has confirmed since 2013. Tools cannot estimate demand for phrases nobody has typed yet. So a large share of buyer language sits outside every database.

Ahrefs tested this directly with its zero volume keyword experiment in October 2022. Keywords the tool estimated at ten or fewer monthly searches often produced more impressions in Search Console than predicted. One example, "hreflang tag seo", showed an estimate of ten yet got 20 impressions in 28 days. That said, fewer than one percent of the tested low volume keywords passed 100 impressions, and the author calls the sample small.

My rule: target a zero volume keyword when it matches a real customer question, carries a commercial modifier, and fits an existing page with a few added paragraphs. Skip it when it needs a whole new page and shows no buyer signal.

How do you find the keywords your competitors make money from?

Competitors reveal their profitable keywords in three places, and competitor keyword research there requires no paid tool. First, their Google Ads copy. Nobody keeps paying for a headline that fails, so recurring ad phrases mark proven demand. Search your core service terms in a private window and note every repeated phrase in the ads.

Second, their service page titles and H1 headings. Agencies and shops that rank well usually align each page title with one commercial keyword, and you can read those titles straight from the results page. Third, the results page structure itself, which I cover in the next section.

Paid tools then add depth. A competitor gap report shows queries where two or three rivals rank and you do not. However, I filter that list by intent before I trust it, because a rival ranking for "what is SEO" tells me nothing about their revenue. Instead, I look for their transactional and commercial rankings and compare page fit against mine.

How do you confirm search intent by reading the SERP?

The results page is the only intent source Google shares openly, so SERP reading belongs in every keyword research routine. Before I commit a keyword to a page, I open it in an incognito window. Then I record what the top ten contains. If eight of ten results are blog guides, Google treats the query as informational. A tool label cannot override that. If service pages and ads dominate, the query is commercial.

Specifically, I note these signals:

• Page type of each top ten result: guide, service page, category, product, directory.
• Presence of text ads above the results, which confirms advertiser interest.
• Shopping, map pack or local results, which reveal product or local intent.
• An AI overview at the top, which tends to absorb simple definition queries.
• The wording of titles, because they show how ranking pages frame the promise.

After that, I draft the title and meta description for my planned page and check them in the Google SERP preview tool. If my snippet reads like a lecture while the top results read like offers, the page will underperform even if it ranks.

How do you judge keyword difficulty and reachability?

A keyword difficulty score in a keyword research tool is a rough estimate of backlink strength among ranking pages. It measures nothing more. It ignores your content quality, your topical authority and how well your page matches intent. Consequently, a KD of 30 can be unreachable for a new domain and a KD of 60 can be realistic for an established one.

I judge reachability with a three part check instead. First, domain strength: are the ranking sites in my league, or are they global brands? Second, content quality: could I write something clearly more useful than the current top three? Third, links: how many referring domains does the weakest top ten page have, and can I match that within six months?

Remember the Ahrefs traffic study, where missing backlinks ranked among the three main reasons pages get nothing. If the weakest competitor still has 80 referring domains and you have five, the keyword belongs in a later quarter. Meanwhile, pick the queries where the weakest ranking page is genuinely weak.

How do you build a conversion focused keyword scoring table?

A scoring table turns keyword research opinions into a ranked list you can defend. I score five factors from one to five: intent, bid signal, reachability, page fit and volume band. Then I weight intent and page fit double, because they predict conversions, and volume single. The table below is an example calculation with illustrative numbers, not client data.

Example calculation: suppose the first keyword sits in the middle of its band at 3,000 monthly searches. A first place ranking at the 27.6 percent click rate from the Backlinko study yields about 830 visits. With an assumed conversion rate of 2 percent, an assumption for illustration only, that is roughly 17 enquiries a month from a single page. The definition keyword, by contrast, would need ten times the traffic to match that.

Which keyword goes on which page in keyword research?

Every keyword in the final list needs one owner page, and the page type should match the funnel stage. Keyword research without this mapping step produces a spreadsheet, not a plan. I use a simple three stage model and assign an internal link target and a success metric to each stage, so the team knows what "working" means.

Two tools help at this stage. I check draft pages in the keyword density checker to make sure the target phrase appears naturally, and I build clean URLs with the slug generator so the address reflects the keyword without stuffing.

How do sales driving keywords differ for local businesses?

Local buyers add a place to almost every commercial query, so the pattern becomes service plus city or district. "Dentist" is a topic; "dentist in Kadıköy open Saturday" is a customer. For local keyword research I therefore build the seed list around districts, neighbourhoods and landmarks. The service alone is not enough.

"Near me" searches behave differently again, because Google resolves the location from the device. You cannot put "near me" in a page title sensibly. Instead, you win those queries through a complete Google Business Profile, consistent address data and a service page that names the areas you actually cover.

Page structure follows the same logic. A single web design service page can rank for a city, but district level queries usually need their own sections or pages with unique local details. Also watch review language; the phrases customers use in reviews are excellent local keyword sources.

How do you separate category and product keywords in ecommerce?

Online stores lose sales when a product page competes with its own category for the same phrase. The rule I apply in ecommerce consulting projects is simple: plural and generic terms belong to categories, while specific model, size, colour and brand terms belong to products.

For example, "running shoes for flat feet" is a category or a filtered listing, because the buyer still compares. "Asics Gel Kayano 31 size 44" is a product query, because the buyer already knows. Mixing those levels creates two weak pages instead of one strong one.

Filtered listings deserve special attention. Combinations such as material plus colour often carry strong buying intent with modest volume. I check Search Console for filter combinations that already collect impressions, then decide which ones earn an indexable landing page and which ones stay blocked. Above all, keep one clear owner per keyword.

How does Google Ads data speed up SEO keyword selection?

The search terms report in Google Ads is the fastest conversion evidence available, because it shows the exact queries that produced leads or orders, not estimates. When I run Google Ads management alongside SEO for a client, I export converting search terms every month and feed them straight into the organic keyword plan.

The workflow takes an hour. First, filter the report to terms with at least one conversion in the last 90 days. Second, group them by theme and check whether an organic page already targets each theme. Third, prioritise the themes where the cost per conversion is high, because organic rankings there save the most budget.

The ROAS calculator helps me frame the decision for clients. If a converting search term costs a great deal per click, an organic ranking for the same phrase has a clear monetary value. That number persuades faster than any difficulty score.

How does keyword cannibalization lower sales, and how do you prevent it?

Cannibalization happens when two or more of your pages target the same keyword, so Google alternates between them and neither gains stable rankings. The commercial damage is worse than the ranking damage, because the page that wins the alternation is often the weaker converter, such as an old blog post beating a service page.

Prevention starts in the mapping step. One keyword, one owner page, recorded in the plan. Detection comes from Search Console: filter a query and look at the pages tab. If two URLs share impressions for a commercial term, you have a conflict. Then choose the page that can sell and act:

1. Redirect the weaker page if it has no unique value.
2. Merge its useful content into the owner page.
3. Otherwise, reframe the weaker page around a distinct keyword and link it to the owner.

Internal links settle most conflicts quickly. Point supporting articles at the owner page with anchor text that contains the keyword, and stop linking the reverse way with the same phrase.

How do you measure whether your chosen keywords bring sales?

Measurement closes the loop, and without it keyword research is a guess. I track three layers. Rankings and clicks come from Search Console, sessions and conversions from GA4 at page level, and lead quality from the CRM. Rankings alone mean nothing, therefore the page level conversion number decides whether a keyword stays.

The routine runs on a 90 day cycle. Each quarter I export the landing page report from GA4, join it with the Search Console page report, and rank pages by conversions per click. Pages with strong clicks and weak conversions get an intent review, because they usually attract the wrong stage of buyer. Pages with few clicks but strong conversions get more internal links and a content refresh.

Attribution needs discipline as well. Tag every campaign link with the UTM builder so paid and social traffic never contaminates the organic numbers. Finally, ask sales which pages the good leads mention; that qualitative signal often explains a number the dashboards cannot.

What are the most common keyword research mistakes?

The same errors appear across industries. Most of them come from treating keyword research as a one time spreadsheet task. Here are the mistakes I correct most often when I audit an existing plan.

• Sorting by volume and ignoring intent, which fills the plan with definitions.
• Trusting tool intent labels without opening the results page.
• Assigning several keywords with different intents to one page.
• Chasing a competitor's brand or informational rankings instead of their commercial ones.
• Dismissing zero volume phrases that customers repeat on every call.
• Writing to a target word count instead of answering the query fully.
• Never checking conversions per page after publishing.

The word count point deserves emphasis. Google states plainly that it prefers no particular length; the question is whether the reader learns enough to reach their goal without searching again. So once the keyword is right, the content must satisfy the intent completely, or the ranking will fade.

How did AI overviews and chat search change keyword selection?

AI overviews and chat style search absorb many simple informational queries, because the answer now appears before any link. As a result, definition keywords lose even more of their limited commercial value. Commercial and transactional queries, on the other hand, still send people to pages, since a buyer needs a provider, a price or a checkout.

Query shape has shifted as well, and keyword research must follow it. People type longer, question style and comparative phrases, such as "which is cheaper for a small shop, Shopify or WooCommerce". Consequently I now add a "question and comparison" pass to every keyword plan. Those phrases get direct answers under matching headings. I do not chase the trend beyond that; the scoring logic stays the same.

How often should you refresh your keyword list?

Review the full keyword research list every quarter and the top twenty commercial keywords every month. Seasonal businesses need an additional check before each peak season, because buyer language shifts with promotions and product cycles. New customer questions from sales and support should enter the list continuously, not at review time.

Each quarterly review answers three questions: which keywords converted, which lost rankings, and which new phrases appeared in Search Console and the Ads search terms report. Then the scoring table gets updated and the content calendar follows it. If you would like a second pair of eyes on your current list, you can contact me. My references show the kind of work this method produced.