Is SEO Worth It in 2026? Do the Arithmetic First

Ten results rank for this question. Two are Reddit threads, one is a Quora page, three are published by companies that sell SEO, and one of the higher-ranking articles still carries "2024" in its title. Between them they offer a flowchart, a list of statistics, and a verdict of "almost always yes".

None of them gives you a number you can check against your own business.

The question is not philosophical. Organic search either produces more gross profit than it costs you, inside a period you can survive, or it does not, and each term in that sentence is something you can measure or bound before spending anything. This article gives you the formula, a script you can paste, three worked examples, and the conditions under which the answer is no.

One clarification first, because the two questions get mixed. This page is about whether to do organic search at all. If you have already decided you want customers from search and you are choosing between paying for clicks and earning them, that is a crossover calculation against ad spend, and we ran it in SEO vs Google Ads.

One thing to settle before the arithmetic, because it constrains everything that follows. Google's own hiring documentation states: "No one can guarantee a #1 ranking on Google." Nothing here predicts what your budget will earn. The point of the model is to tell you what would have to be true for the investment to pay, so you can judge whether that is plausible in your market.

The whole question in one line of arithmetic

SEO is worth it when the gross profit produced by organic clicks exceeds what those clicks cost you to earn, within a period you can survive.

monthly gross profit from organic = C × R × M

C = organic clicks per month at steady state
R = click-to-customer rate
M = gross profit per customer

Break-even is the month where the running total of C × R × M passes the running total of what you spent, in money and in billable hours.

Three inputs. That is the entire decision. Everything else in this article is about not getting each one wrong, and the first one is where almost every published ROI model breaks.

Input 1: clicks, not searches

C is clicks. Not search volume, not impressions, not "monthly searches" from a keyword tool. The gap between those numbers got wider in 2026, and the published models have not caught up.

Take the query this article targets. It draws 421 US searches a month. Above the organic results sits an AI Overview, and two of the ten positions are forum threads. The pages holding page-one organic spots carry, by Ahrefs' estimate on 2026-09-03, somewhere between 33 and 194 monthly organic visits each, and that figure counts every keyword those pages rank for, not just this one.

A four-hundred-search keyword does not deliver four hundred clicks. It rarely delivers a hundred.

The second half of the gap is newer. Google now reports your visibility in AI Overviews and AI Mode inside Search Console, and the report has one metric. From Google's own help page: "Impressions are how many times links to your site were shown to a user in a generative AI feature on Google Search." The same page states: "As of August 31, 2026, we've rolled out these insights to all websites worldwide."

Impressions. No clicks, no click-through rate, no average position, no queries.

That is a fact about your model, not a complaint about Google. The surface that every vendor now tells you to optimize for cannot be entered into a click-based ROI calculation, because the clicks are not reported. So build the model on the clicks you can count in the standard Performance report, and treat AI-feature impressions as upside you are not pricing. If you want the mechanics of that surface rather than its measurement, we covered them in answer engine optimization and LLM SEO.

How to bound C before you have any data

You do not need a forecast. You need a ceiling and a floor.

• Ceiling: sum the monthly volume of the ten to twenty queries you could realistically target, then take 10 to 25 percent of it. That band is where a set of pages holding decent positions tends to land once AI Overviews, ads, forums and video carousels have taken their share.
• Floor: zero. An Ahrefs study of roughly 14 billion pages, published in 2023, found that 96.55 percent of pages get no search traffic from Google at all. That is the base rate you are betting against, and it is the reason to model first.
• Replacement: after 90 days, delete both numbers and use your own. Search Console gives you clicks, impressions, click-through rate and average position per query. Your measured click-through rate at position 4 in your SERP beats any industry average, because it already accounts for the AI Overview and the ads sitting above you. Once you have a few months of it, the same per-page pull is the input to an SEO content audit.

Input 2: what one click is actually worth

M is gross profit per customer, not revenue, and not the price on your pricing page.

Revenue flatters the model by whatever your delivery costs are. A contractor booking a $4,200 job at a 35 percent margin earns $1,470 of gross profit from it. Running the model on $4,200 makes SEO look three times better than it is.

For anything with repeat purchase or a subscription, use gross profit across the customer's lifetime, and be conservative about lifetime. A $49 per month product at 85 percent margin with a 14-month average life is worth about $583 of gross profit. Use the average life you have measured, not the one in your plan.

Then multiply by R, the rate at which an organic click becomes a customer. For most small businesses this is a chain, not a single number:

R = (click → lead) × (lead → customer)

A contractor at 3 percent click-to-lead and 30 percent lead-to-job has R = 0.009. Nine customers per thousand clicks.

The output of this section is one number: gross profit per organic click, R × M. For the contractor, $1,470 × 0.009 = $13.23 per click. For the SaaS at 1 percent, $583 × 0.01 = $5.83 per click. Write it down. It is the only number that makes a cost quotable or absurd.

Here is where the arithmetic kills some businesses outright. A $30 one-time product at 50 percent margin, converting at 1 percent, is worth 15 cents per click. No ranking rescues that. The problem is not the channel.

Input 3: what it costs, in both currencies

Cost has two currencies and most models only count one.

Money. If you are buying the work, a quote is not a price. It is six line items bundled into one number, and we priced each of them separately in how much SEO costs. What matters for the break-even model is the monthly figure and any setup fee, both of which a proposal will give you.

Hours. If you are doing it yourself, your time is the cost, and pricing it at zero is the single most common error in DIY plans. A contractor who bills $85 an hour and spends 4 hours on an article is spending $340 on that article, whatever the invoice says. Price your hours at what you would have billed, and the two options become comparable.

Google's own guidance on hiring is blunt about which side of that trade a small operator sits on: "If you run a small local business, you can probably do much of the work yourself." The doable half is what we mapped in DIY SEO for small business.

One more line in the cost column that people forget: the months where you spend and earn nothing. That is not overhead, it is the shape of the investment, and the script below makes it visible.

The calculator, pasteable

Save this as breakeven.js and run it with node breakeven.js. No dependencies, no install.

// breakeven.js: replace every number with yours.
const inputs = {
  steadyClicks: 100,        // organic clicks/month once pages mature
  rampMonths: 8,            // months to reach steadyClicks (linear ramp)
  clickToCustomer: 0.01,    // R = (click→lead) × (lead→customer)
  grossProfitPerCustomer: 583, // M, gross profit not revenue
  monthlyCost: 600,         // agency retainer, or your hours × your rate
  setupCost: 0,             // one-off: migration, technical fixes, tooling
  horizon: 24,
}

const perClick = inputs.clickToCustomer * inputs.grossProfitPerCustomer
let cumProfit = 0
let cumCost = inputs.setupCost
let breakEven = null

console.log(`Gross profit per organic click: $${perClick.toFixed(2)}`)
console.log('month  clicks  profit    cumProfit  cumCost')

for (let m = 1; m <= inputs.horizon; m++) {
  const clicks = Math.round(
    inputs.steadyClicks * Math.min(1, m / inputs.rampMonths)
  )
  const profit = clicks * perClick
  cumProfit += profit
  cumCost += inputs.monthlyCost
  if (breakEven === null && cumProfit >= cumCost) breakEven = m
  console.log(
    String(m).padStart(5),
    String(clicks).padStart(7),
    ('$' + profit.toFixed(0)).padStart(8),
    ('$' + cumProfit.toFixed(0)).padStart(10),
    ('$' + cumCost.toFixed(0)).padStart(8)
  )
}

console.log(
  breakEven
    ? `Cumulative break-even in month ${breakEven}.`
    : `No break-even inside ${inputs.horizon} months at these inputs.`
)

Run it once with your honest numbers and once with your pessimistic ones: half the clicks, two thirds of the conversion rate, twelve ramp months instead of eight. If the pessimistic run never breaks even inside a horizon you can fund, you have your answer and it cost you nothing.

The ramp is the input people fake. Set it long. Nobody can tell you when pages start earning, and any model that assumes month-one traffic is selling you something.

Three businesses, same arithmetic, different verdicts

The interesting part of that table is the bottom row. Three businesses with almost nothing in common, and all three need roughly 80 to 100 organic clicks a month to cover their cost. The variable that decides the outcome is not the business model. It is whether 100 clicks a month is achievable in their SERP.

That reframes the decision usefully. Given the click volumes measured earlier, a single article holding a page-one position for one mid-volume keyword does not produce 100 clicks a month. A set of ten to thirty pages, each earning a handful, does. SEO is a portfolio question, and break-even is a portfolio number.

It also explains the failure mode people describe as "we tried SEO and it did not work". Four articles is not a trial of the channel. It is a sample too small to reach the break-even row of the table above, and if the doubt is really about which channel deserves the hours, that is a separate measurement against social.

The check that kills good-looking keywords

Volume and difficulty tell you whether you can rank. They say nothing about whether ranking would be worth anything, and that is a separate check most people skip, whether they run it by hand or hand it to an AI SEO agent.

Open the actual search results for your target query and read who holds the top spots, and for what. Two patterns decide it:

• Wrong reader. If the ranking pages are written for enterprise marketing teams or for people shopping for an agency, and you sell to owner-operators, the traffic will arrive and not convert. We keep a rejected list of exactly this kind of keyword: one commercial family in our own niche adds up to more than 12,000 monthly searches and every page ranking in it exists to get a firm hired. Good volume, wrong reader, skipped. The measured version of that call, on a keyword with a difficulty of 7, is in SEO keywords for photographers.
• Wrong format. The page types Google rewards tell you what it thinks the query means. For this article's keyword, the top ten is tutorials, opinion pieces, Reddit and Quora. Google reads the query as peer advice. A service page would not hold there no matter how well it was built.

Run that check before writing, not after. It is the cheapest step in the whole process and it is the one that prevents the outcome everyone fears, which is traffic that produces no customers. What it looks like applied to one trade, query by query, is in SEO for contractors.

When the honest answer is no

The published articles on this question almost all conclude yes, which is a poor sign given who publishes them. Here are the conditions where the model says no, each with a test you can run today.

Nobody searches for what you solve. Test: check your ten most obvious queries in any keyword tool. If nothing clears 50 searches a month in your country, the demand is not in search. Word of mouth, partnerships or outbound will beat it.

You need revenue inside 90 days. Nothing in this article promises a timeline, and that is precisely why this is a disqualifier. If the business does not survive a quarter of spending without return, the model's ramp column is fatal regardless of its final row.

Gross profit per customer is under $20 with no repeat purchase. At 1 percent conversion you are earning 20 cents per click. You would need thousands of clicks a month to cover any real cost, which puts you in the highest-competition part of any SERP.

Your buyers do not discover this way. Procurement lists, app-store search, marketplace placement, referral networks: if your last twenty customers came from none of them through Google, check before assuming the twenty-first will.

The SERP is locked and you have nothing unique. Test: look at the domain ratings of the top ten. If the lowest is 79 and you are starting near zero, and your page would say the same things theirs do, the arithmetic is not the problem. Some SERPs have a low-authority page in the top ten, which is the fastest signal that the door is open. The one measured for this article has pages at domain rating 36 and 45 in the top ten, which is why it was worth writing.

You cannot sustain it for six months. The compounding argument for SEO requires compounding. Three articles and a pause is a cost with no asset at the end.

If two or more of those apply, the answer is no today. It may be yes in a year, with the same arithmetic and different inputs.

What changed in 2026, and what did not

The changes are real and they are on the measurement side.

• AI Overviews appear above the organic results on informational queries, including this one. Fewer clicks reach the pages below.
• Search Console's AI-features insights reached all sites worldwide on August 31, 2026, and report impressions only.
• Forums hold positions on advice-shaped queries that used to go to blogs. Two of the top ten here are Reddit.

What has not changed is the cost side, and Google says so directly. Its documentation on AI features states: "There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary", and adds: "You don't need to create new machine readable files, AI text files, or markup to appear in these features."

That matters for the model. The arrival of AI search did not add a line item. Anyone selling you one for a new acronym is selling the same work at a higher price, a point we made in more detail in LLM brand visibility and in the piece on getting cited by ChatGPT.

The same page of Google's documentation is equally direct about what tools can promise: "Third-party tools don't have access to our internal ranking data. They can't guarantee performance." That was last updated on 2026-06-05, and it applies to every dashboard and every forecast, including the one you just pasted.

Test the answer for $0 before committing

You can run a real experiment on this question before spending a retainer.

1. Verify the site in Search Console. Free. Wait 28 days for a baseline if you have never had it.
2. Read your own click-through rate by position in the Performance report, filtered to queries where you already appear. Use that instead of an industry average, because it already prices the AI Overview above you.
3. Pick five target queries. Check volume, difficulty, and who ranks, using the two tests from the section above.
4. Publish three pages against the three weakest SERPs of the five. Track the hours. Those hours are the real cost input.
5. Read the results in order at 8 to 12 weeks: impressions first (are you shown at all), then average position (are you competitive), then clicks. Impressions rising while position sits at 40 means indexed but not competitive, which is a content problem, not a technical one. Nothing at all after 12 weeks is usually an indexing problem, and the technical checklist is where to start.
6. Feed the measured numbers back into the script and re-run it. Now the model is yours instead of an estimate.

The experiment costs three articles. The decision it informs is a year of spending, and results vary between sites for reasons no tool can see.

FAQ

Is paying someone to do SEO worth it?

It changes one input, the cost, and leaves the other two alone. Run the script twice: once at your hourly rate for the hours you would spend, once at the quoted retainer. The version that breaks even inside a horizon you can fund wins.

On choosing a provider, Google's guidance carries the warnings worth memorizing: "No one can guarantee a #1 ranking on Google. Beware of SEOs that claim to guarantee rankings, allege a 'special relationship' with Google, or advertise a 'priority submit' to Google." It also flags firms that are "secretive or won't clearly explain what they intend to do", and unsolicited email pitches. We went through the same evaluation from the buyer's side for SaaS SEO agencies and B2B SEO agencies, and for the consultant option.

Is SEO dead now with AI?

No, but what you can prove about it changed, and for a model built on measurement that is the part that matters. Your appearances in AI Overviews and AI Mode now arrive as impressions with no clicks, no click-through rate and no queries attached, so a growing share of your visibility is real and unpriceable at the same time.

The work did not change. Google states there are no additional requirements and no new files or markup needed to appear in those features, so nothing new belongs in the cost column either. The keyword-selection consequence, which is a separate argument, is worked through in SEO vs Google Ads.

Is SEO dead or evolving in 2026?

Evolving, and specifically on the surface rather than the mechanism. Pages still have to be crawlable, relevant and better than what ranks, and that part reads exactly as it did three years ago.

What is new is that a growing share of your visibility is reported to you as impressions with no clicks attached, which makes attribution harder and makes measuring what you can measure more important, not less.

What is the 80/20 rule in SEO?

It is the observation that a small fraction of your pages will produce most of your organic traffic, and it is consistent with the base rates: the 2023 Ahrefs study of about 14 billion pages found 96.55 percent of them getting no Google traffic at all.

The practical use is in the review, not the plan. Measure clicks per page every quarter, keep publishing in the shape of whatever is working, and stop maintaining pages that have produced nothing after two quarters. The portfolio logic from the worked examples applies here too, and it is why single-article trials tell you nothing. We wrote up how that plays out for a product business in SEO for SaaS, and for a store in content marketing for ecommerce.

How long does SEO take to work?

Nobody can tell you, and any specific figure you are quoted is a sales number rather than a measurement. What you can do is decide in advance what you will check and when: indexation at two weeks, first impressions at four to eight, position movement at eight to twelve, and clicks after that.

Set the ramp in the calculator to the pessimistic end of whatever you believe. If the model only works at the optimistic end, it does not work.

Is SEO worth it for a small business?

It depends on the same three inputs, and small local businesses often have the best version of two of them: high gross profit per customer and searchers with clear local intent. The constraint is usually the hours.

Google's own answer to the related question is that a small local operator "can probably do much of the work yourself", which changes the cost input more than anything else on this page. Price your hours honestly, run the script, and let the arithmetic decide rather than the flowchart.