SEO ROI comes down to five numbers

Most conversations about SEO return start with rankings and traffic. Both matter, but neither is a business outcome. To estimate what SEO is worth you need five numbers, and the honesty to treat the result as an estimate.

The five numbers

1. Organic traffic: monthly visits from search to the pages that sell.
2. Conversion rate: the share of those visits that become a lead (a call, form or booking).
3. Close rate: the share of leads that become customers.
4. Customer value: average contribution after delivery costs, before SEO spend. If you only know revenue, label the result as a revenue-based scenario.
5. Cost: what you invest in SEO each month.

The calculation

• Traffic × conversion rate = leads
• Leads × close rate = customers
• Customers × contribution value = attributed contribution before SEO cost
• Attributed contribution before marketing cost − SEO cost = net marketing return
• Net marketing return ÷ SEO cost = marketing ROI

Our free SEO ROI calculator models revenue less SEO spend. Its output excludes delivery costs and overhead. Use a separate contribution calculation when assessing marketing ROI.

Where the estimate breaks

• Traffic is not all equal. Branded searches and informational visits convert very differently from “service + city” searches. Estimate from the pages that sell, not the whole site.
• Attribution is messy. Many buyers find you organically, leave and come back through another channel. Different attribution rules assign credit differently. A simple model can overstate or understate SEO’s role.
• Time matters. SEO costs arrive before the returns. Choose a period that accommodates your sales cycle, and compare costs with returns from the same cohort.
• Use profit, not revenue, where you can. Revenue overstates the return for businesses with high delivery costs.

How to use the estimate

• Compare scenarios, not single answers: what if conversion improved from 2% to 3% instead of traffic doubling?
• Compare the cost and feasibility of each lever. A conversion repair and a traffic initiative have different costs, evidence needs, and implementation constraints.
• Revisit the inputs with real data every quarter, replacing assumptions with measured numbers.

How do you choose consistent inputs?

Use the same population and period throughout the calculation. Mixing all-site traffic with the close rate of qualified sales leads creates a misleading result.

Define organic traffic before exporting it. Decide whether the model uses sessions, users, or another documented unit. Then use a conversion rate calculated from that same denominator.

For a lead-generation business, distinguish a recorded form event from a unique enquiry. One visitor may submit twice or call after submitting. Deduplicate where the business records allow it, and explain any remaining limitation.

A GA4 key event is an event you mark as important. It does not automatically represent a qualified lead or a customer. Keep the definition visible when presenting the rate.

Calculate close rate from the same kind of lead. If the numerator is closed customers from qualified enquiries, use qualified enquiries as the denominator. Do not multiply that rate by all raw form events without adjusting the model.

For customer value, use an agreed financial definition. Revenue, gross contribution, and lifetime contribution are different inputs. Document delivery costs, refunds, and the value period you include. A lifetime assumption needs evidence about retention and collection, not just an optimistic sales estimate.

How do you handle the gap between a visit and a sale?

Group leads by their acquisition period and follow their outcomes. That makes the sales delay visible instead of mixing today’s costs with customers acquired months earlier.

A monthly traffic report and a monthly sales report may describe different customers. A buyer who arrived in one month and closed in another belongs to a journey that spans both. Decide how your cohort report records that journey.

Keep observed results separate from pending opportunities. A proposal is not collected revenue. A signed contract may still have delivery costs or cancellations. Label the stage that each figure represents.

If you cannot connect enquiries to their acquisition source, report that gap. Analytics estimates can still help compare scenarios, but they cannot establish customer-level profitability without the missing records.

Use Google Analytics 4 to improve the record, and review attribution models before assigning channel credit. Changing the model can change the reported result without changing the business’s sales.

What does attribution tell you about incremental value?

Attribution assigns credit under a set of rules. It does not establish how many sales would disappear if the SEO work stopped.

Some customers may already know the business. Others may discover it through several channels. Brand demand, referrals, sales activity, and organic discovery can overlap. Counting attributed revenue as entirely incremental value ignores that overlap.

Use the attribution result as one view of the evidence. Compare it with lead quality, customer journeys, and changes in non-brand discovery where those records exist. A controlled experiment can answer a narrower causal question, but only if its design and sample support that conclusion.

Avoid a precise incrementality claim when you have only a channel report. State what the report measures and what remains unknown. That makes the estimate more useful for budgeting.

How should you compare scenarios without creating a forecast?

Change one assumption at a time, then check whether the proposed change has a plausible mechanism and cost.

A hypothetical increase from a 2% to a 3% lead rate means more enquiries from the same traffic. It does not prove that a redesign can achieve that increase. It also does not establish that the extra enquiries will close at the existing rate.

A traffic increase may bring a different query mix. Applying today’s conversion rate to every new visit assumes that the new audience behaves like the current one. Test that assumption before presenting the scenario as a plan.

Use lower, central, and upper scenarios only when you can explain their inputs. Do not invent probabilities for them. Record the evidence that would make you revise each assumption.

Compare implementation costs as well as output. Improving a broken enquiry form may require different work from earning visibility for a new service. The cheapest apparent lever in a spreadsheet may have the weakest supporting evidence.

What belongs in the SEO cost figure?

Include the costs needed for the work being evaluated. An agency fee alone may exclude content production, developer time, paid tools, and internal review.

Keep shared costs consistent. If engineering supports several channels, document how you allocated that effort. Avoid changing the allocation between periods just to make a return figure look better.

For a zero-cost input, the ROI division is undefined. Report the contribution and explain the cost basis rather than treating the return as infinite.

A negative marketing return means the modeled contribution is below the included investment for that period. Investigate the inputs, maturity of the cohort, and delivery of the work. It is not proof that SEO cannot work, and it is not a reason to assume future profitability.

The calculation becomes useful when its definitions are visible. Show the period, population, value basis, costs, attribution rule, and unresolved gaps alongside the result.

Sources

• Introducing the next generation of Google Analytics - Analytics Help · accessed October 8, 2026
• Conversions vs. key events in Google Analytics - Analytics Help · accessed October 8, 2026
• Get started with attribution - Analytics Help · accessed October 8, 2026