Book a call

Is SEO Actually the Right Growth Channel for Your Business?

SEO needs four conditions to survive together. Preserve query categories, observed click outcomes, signal limits and the bounded indexed attribution comparison. Four constant-offset bent bands travel across bright blue in yellow, orange, coral, and cyan beneath a white FIT identifier.

SEO deserves to lead your growth plan only where searchable demand, credible answer–offer fit, earnable visibility, and incremental post-click value overlap; if any condition fails, search may remain useful infrastructure without being the right primary channel. Summary

A software company ranks first for a phrase that attracts 12,000 searches a month. The traffic report looks triumphant. Sales asks why none of the visitors resemble the buyers it can close.

Nothing is technically broken. The page loads. The keyword appears in all the expected places. The ranking is real. The growth is not.

This is the trap inside the question, “Should this business invest in SEO?” Search volume makes the opportunity visible, so it receives the first and often the last word. Yet volume describes only one condition: people typed something. It does not show whether those people need what the business sells, whether the site can earn their attention, whether the result page will send a visit, or whether that visit creates value that would not have arrived anyway.

SEO is not one opportunity. It is the overlap of four conditions: searchable demand, answer–offer fit, earnable visibility, and incremental value.

If all four exist, search can compound into an unusually durable growth asset. If one is missing, a company can publish for years and still build little more than an expensive reporting habit.

Channel-fit boundary

SEO works only inside the overlap

Demand, answer–offer fit, earnability, and incremental value can each be strong while the complete growth route remains absent.
Reading note

Editorial synthesis of the paper's search-demand, relevance, authority, result-page, conversion, and incrementality evidence.

Search is excellent at capturing intent that has already become language. A person recognizes a need, frames it as a query, and asks a search system to reduce uncertainty. That makes SEO powerful near decisions that people naturally research: how to solve a known problem, which alternatives exist, what a product costs, what requirements apply, or whom to trust.

It also creates a hard boundary. Search cannot capture demand that has not become a query.

A novel category may solve an urgent problem while its buyers lack the words to find it. Enterprise work may begin through procurement lists, analyst relationships, referrals, conferences, partner ecosystems, or direct access to a narrow group of decision-makers. A local emergency service may depend on proximity and immediate availability. A fashion object may spread through taste and social proof long before a buyer searches for its functional description.

In those settings, the first growth job is not ranking. It is creating recognition, trust, distribution, or a vocabulary for the problem.

This distinction is easy to lose during keyword research because the tools produce numbers even when the strategic unit is unclear. Google explains that Trends data are sampled, normalized within the selected time and place, then scaled from 0 to 100. A value of 100 marks peak relative interest inside that comparison. It is not the number of buyers. Low-volume terms can show zero even when some searches exist.

That makes Trends useful for pattern detection and dangerous as a stand-alone market estimate.

The first channel-fit question is therefore not “How much search volume exists?” It is: Which customer situations become queries, and how close are those queries to a decision the business can serve?

Query opportunity map

A smaller query can occupy the stronger commercial position

Decision proximity and earnable visibility define four different jobs before reported volume enters the decision.
Reading note

Editorial synthesis of Nagpal and Petersen, 2021, and Google Trends methodology. The cells are diagnostic categories, not measured market segments.

One peer-reviewed SEO study makes the point concrete. Mayank Nagpal and J. Andrew Petersen examined the first 30 results for 1,791 queries across an online retailer, a culinary school, and an urgent-care provider. Query popularity mattered, but it interacted with competition, specificity, intent, content relevance, and online authority. The same keyword rule did not apply across the three markets.

That finding changes the research task. Instead of collecting the largest terms, map search situations:

  • What happened immediately before the query?
  • Is the person learning, comparing, validating, buying, repairing, or returning?
  • Can the business complete that task in public?
  • What action would indicate real progress after the answer?

A query with modest volume and clear commercial consequence can be more valuable than a broad term with an audience that only wants a definition. This is not an excuse to dismiss scale. It is a way to stop mistaking a crowd for a market.

The page must complete a real task

Once demand exists, the second question is whether the business has a credible answer–offer fit.

This is stricter than “Can the marketing team write about it?” Almost any organization can produce an article about a popular subject. Far fewer can answer the query with distinctive evidence, connect the answer to an offer, and fulfill the expectation created by the page.

Nagpal and Petersen found that online authority was especially important for informational searches, while content relevance became more important for transactional searches later in the customer journey. The result is intuitive once the two jobs are separated. Early research asks, “Is this source worth believing?” A transaction asks, “Does this result match what I need to do now?”

Both tests can fail.

A new site may publish a precise medical explanation but lack the authority needed to earn attention in a high-stakes result set. An established brand may rank for a product category but send the visitor to a generic page that does not match the query's requirements. A consultant may attract thousands of people who want a template while selling a service whose buyers need procurement evidence, integration detail, and risk reduction.

The practical unit is not a keyword. It is a query–answer–offer contract:

  1. The query exposes a real task.
  2. The result promises a specific reduction in uncertainty.
  3. The page fulfills that promise with useful content or functionality.
  4. The offer provides a credible next step when the task requires more help.
  5. The business can deliver what that next step implies.

If the contract breaks at step four, the page may become a good publication with no commercial route. If it breaks at step five, SEO can amplify disappointment faster than the business can repair it.

Technical SEO belongs inside this contract, but it is not a substitute for it. Google's SEO Starter Guide explains crawlability, indexing, descriptive content, link structure, and user value. It also makes two useful cautions: no method guarantees that Google will crawl, index, or serve a page, and some changes can take several months to show their effect.

Technical eligibility opens the door. It does not make the room interested.

Visibility must be earnable, not merely imaginable

An SEO forecast often multiplies search volume by a standard click-through rate for a target rank. The spreadsheet is elegant. The assumption underneath it is doing dangerous work.

Michael Baye, Babur De los Santos, and Matthijs Wildenbeest studied organic product-search traffic and found that retailer-name prominence was as important as screen position in predicting clicks. Models that omitted name prominence and the endogeneity of position substantially inflated the apparent effect of rank.

In plain language: a familiar result at position three and an unknown result at position three do not own the same opportunity. They may have earned that position through different underlying strengths, and the rank itself does not cause every click credited to it.

The result page adds more variation. Ads, maps, products, videos, forums, featured answers, local packs, comparison widgets, and AI summaries can occupy the same nominal position space. The query may reward a tool, a product feed, a directory, a known publication, or a first-party source instead of another article. Competitors may have years of references, brand searches, expert contributors, proprietary data, and links that cannot be reproduced on a content calendar.

Earnability therefore needs its own model. For each query family, inspect:

  • who currently earns attention and why;
  • which result formats the search system prefers;
  • what evidence and authority the leading sources possess;
  • whether the business can offer a meaningfully better answer;
  • which technical or organizational dependencies must exist first;
  • how long the opportunity can wait before another channel becomes more valuable.

This is where “SEO takes time” becomes too vague to guide a decision. The delay is not one fixed waiting period. Some pages can surface within days. A new body of authority can take years. The relevant duration is the time required to produce a credible result in this result environment—and the value of everything displaced while that work matures.

The search page may keep the click

Even an earnable result does not guarantee a visit.

In July 2025, Pew Research Center published an analysis of 68,879 Google searches made by 900 U.S. adults. A traditional result received a click in 15 percent of searches without an AI summary. When an AI summary appeared, that rate was 8 percent. Links cited inside the summary received clicks in only 1 percent of summary searches.

Result-page click shift

An AI summary cut traditional-result clicks nearly in half

Traditional-result clicks were observed in 15 percent of searches without an AI summary and 8 percent with one; summary source links received 1 percent.
Reading note

Pew Research Center, 2025. Observational data from 68,879 Google searches made by 900 U.S. adults.

Sessions also ended more often after a page with an AI summary: 26 percent, compared with 16 percent without one. AI summaries appeared in 18 percent of the observed searches, so the effect did not apply to every query. It did, however, affect enough behavior to invalidate the assumption that visible demand reliably becomes website traffic.

The obvious response is to avoid queries that can be answered inside the result page. That is directionally useful but incomplete.

Some no-click visibility can still create value. A trusted citation can reinforce recognition. A concise answer can reduce support burden. A useful definition can shape category language. A local result can produce a call without a website session. The problem is not that value requires a click. The problem is that traffic reports cannot measure value that happens without one.

If no-click value belongs in the plan, define the outcome separately. Measure branded demand, assisted discovery, calls, saved support time, citation visibility, or another observable consequence. Do not preserve the old traffic goal and call its disappearance brand awareness after the fact.

A visit must survive the economics after the click

The click is the narrowest point in most SEO reporting and the widest point in business reality.

Anindya Ghose and Sha Yang analyzed several hundred paid-search keywords over six months. Higher sponsored positions were not always the most profitable. Landing-page quality was associated with higher conversion and lower cost per click. Although the study concerns paid search, the lesson transfers carefully: attention has no fixed value outside the page and offer that receive it.

Evidence ledger

Every familiar SEO metric stops before growth

Rank, click, and conversion each observe a useful transition while leaving a different part of causal and economic value unresolved.
Reading note

Editorial synthesis of Baye et al., 2016; Ghose and Yang, 2009; Google Search Console; Blake et al., 2015; and Lewis and Rao, 2015.

For organic search, the full route is:

impression → result attention → click or no-click action → task completion → qualification → conversion → retained economic value

Every arrow can change by query. A broad educational article may create many impressions and weak commercial qualification. A technical comparison may create fewer visits but materially influence a high-value decision. A support answer may prevent a ticket rather than create a lead. A local page may produce a phone call that browser analytics never joins to the result.

Google Search Console provides impressions, clicks, click-through rate, average position, and query-level filters. Those are diagnostic inputs, not a growth verdict. Join them to landing-page events, qualified actions, sales outcomes, margin, retention, and the cost of production and maintenance.

Then segment. Branded and nonbranded queries answer different questions. Informational and transactional searches carry different tasks. New and returning visitors may have different causal relationships with the channel. Mobile and desktop result pages can offer different interaction paths. A blended total can rise while the valuable cohort stays flat.

This is the point at which SEO stops being a content program and becomes a channel model.

Search attribution can flatter itself

Search receives credit at a convenient moment: after intent already exists.

A person sees a recommendation, hears a founder on a podcast, discusses a vendor with a colleague, then searches the company name. Analytics records organic search. The search result mattered, but it did not necessarily create the demand. Without a counterfactual, the report cannot distinguish navigation from acquisition.

Randomized paid-search research makes the attribution problem visible because ads can be turned off for selected users. In large eBay experiments, Thomas Blake, Chris Nosko, and Steven Tadelis found that observational returns overstated experimental effects. Brand-keyword ads created no measurable short-run benefit in that setting, while nonbrand effects were concentrated among new and infrequent users.

That is not proof that search never works. It is proof that averages conceal who needed the channel.

In another experiment, Joseph Golden and John Horton compared a search-advertising shutdown with the loss that observational data predicted. The experimental customer loss was only 63 percent of the observational estimate. A 2025 study by Sarah Moshary found sponsored ads taking clicks from organic results and reducing transactions in its setting, even while the ads remained profitable for the platform selling them.

The brand-search experiments by Andrei Simonov, Chris Nosko, and Justin Rao add an important qualification. Without competitor ads, brand advertising produced only about 1 to 4 percent lift. When competitors advertised, the brand could lose 18 to 42 percent of clicks by staying absent. The same query changed economic meaning when the result environment changed.

Attribution gap

Observed attribution predicted more customer loss than the experiment found

When the observational estimate is indexed to 100, the experimentally measured customer loss is 63 in this marketplace study.
Reading note

Golden and Horton, 2021. One experimental marketplace context; ratio shown to expose attribution bias, not to forecast another business.

SEO cannot be switched off as cleanly as an ad, and its effects can persist. That makes causal measurement harder, not optional.

Randall Lewis and Justin Rao examined 25 field experiments representing $2.8 million in advertising spend. The median confidence interval for ROI was wider than 100 percentage points. Some informative tests could require more than 10 million person-weeks. Their message is uncomfortable and useful: a dashboard can report attribution to two decimals while the incremental return remains commercially uncertain.

The answer is not to stop measuring. It is to match certainty to the decision.

Use operational cohorts to diagnose the funnel every week. Use phased publishing, geographic variation, query portfolios, content holdouts where feasible, interruption designs, or matched comparisons when an investment decision requires incremental evidence. Keep branded navigation separate from discovery. Compare new and existing buyers. State what the test cannot prove.

Compare SEO with the channel buyers actually use

A channel can pass all four search tests and still not deserve first priority.

The opportunity cost is the strongest alternative, not doing nothing.

If buyers discover the category through professional communities, a partnership program may reach them earlier. If trust transfers through referrals, customer advocacy may outperform another hundred articles. If product use spreads inside teams, integrations and invitations may create a stronger loop. If the market is small and identifiable, direct access may produce evidence before search authority matures. If a marketplace controls the purchase path, listing quality and marketplace reputation may be the binding work.

This comparison prevents a common strategic error: evaluating SEO against its own historical traffic instead of against another use of the same people, money, and time.

Give each candidate channel the same decision frame:

  • What customer state does it reach?
  • Does it create demand or capture demand?
  • What credible asset does the business already possess?
  • How long until the first decision-changing evidence?
  • What does a qualified outcome cost?
  • Which value persists after spending slows?
  • What work must stop to fund it?

SEO often performs well on persistence. A useful page can keep earning discovery after publication. It can also decay, require updates, lose result features, or become obsolete when the search system answers the task directly. Compounding is a possible property of the asset, not a guarantee attached to the channel's name.

Run a channel-fit test before building a content machine

The cleanest test begins with a bounded query portfolio, not a company-wide promise to “do SEO.”

Choose one customer situation that matters. Map the query family around it. Inspect the current result environment. Define the answer–offer contract. Publish the smallest set of pages or tools that can demonstrate technical eligibility, relevance, authority-building potential, result-page attention, and downstream qualification.

Then stage the evidence.

Stage one: demand and eligibility. Confirm that the query family appears in real customer language. Ensure the pages can be discovered and indexed. Stop if the category depends on demand creation that another channel can perform sooner.

Stage two: impressions and relevance. Track which queries and pages receive impressions. Read unexpected queries as evidence about language and intent. Stop if visibility consistently arrives for the wrong task.

Stage three: result attention. Examine click-through rate in the actual result layouts. Note AI summaries, maps, ads, products, forums, and other features. Stop treating rank as a traffic forecast if the result page completes the task.

Stage four: qualification. Join queries and pages to meaningful actions. Separate branded navigation, existing customers, job seekers, students, support users, and other nonbuyer cohorts. Stop scaling if the qualified route remains empty.

Stage five: incremental value. Compare the result with a credible baseline and the next-best channel. Include production, engineering, expert time, authority-building, maintenance, delay, and displaced work. Scale only when the opportunity survives that comparison.

Channel-fit test

Scale only after all four conditions survive

Each condition has a pass state, a stop state, and a next piece of evidence so SEO cannot advance on traffic optimism alone.
Reading note

Editorial synthesis of the paper's complete qualified evidence base.

A 90-day pilot can reveal language, indexing, impressions, early result attention, and operational feasibility. It may not reveal the mature revenue effect of a competitive SEO program. The test should promise an evidence horizon, not a magical payback date.

That distinction protects both sides of the decision. It stops a weak program before it becomes a publishing factory, and it stops an impatient company from killing a credible search asset before the necessary evidence can mature.

The right answer is a boundary, not a belief

SEO is not dead. It is not free. It is not mandatory. It is not a trick performed on a page.

It is a conditional growth system that begins when demand becomes a query and succeeds only when the business can earn attention, complete the task, connect the answer to real value, and show that the value is more than redirected intent.

That makes the decision less glamorous and more useful.

If buyers search, the business can answer credibly, visibility is earnable, and downstream value survives a counterfactual, SEO can be the right growth channel. If buyers do not search, if the result page absorbs the click, if authority is uneconomic to build, or if the conversions would arrive anyway, lead with the channel that fits the market's actual behavior.

The question is not whether SEO works.

It is whether this search terrain contains a viable route from a real query to incremental value for this business.

References

Summary

Treat SEO as a conditional demand-capture system: prove that buyers search for the need, that your business can satisfy the query credibly, that visibility is realistically earnable, and that the resulting attention creates incremental value after cost and delay.

  1. Map the queries that express a real customer situation, decision, or task—not only terms with reported volume.
  2. Connect each query family to a public answer, offer, proof point, and next action that your business can genuinely support.
  3. Estimate earnability from result-page competition, authority, format, technical eligibility, and the time required to build them.
  4. Model result-page click yield separately from landing-page qualification, conversion, margin, and retention.
  5. Compare SEO with the channel buyers already use to discover, trust, and purchase this category.
  6. Run a staged test with leading indicators, downstream cohorts, and a credible counterfactual before scaling production.

What Are the Formulas for Virality in Social Media? (Part 1)

Expand the Offer or Improve the Core: Which Move Should Come Next?