The short answer
Compare the value of organic conversions against the cost of the work, using the client's own conversion values and a consistent baseline. State assumptions and avoid claiming credit for unrelated changes.
Agree on what a conversion is worth
Use the client's real average order value, lead close rate or customer value. Never invent figures.
If data is missing, say the return cannot yet be calculated.
Compare like with like
Use matching periods and note seasonality, promotions and tracking changes that could explain movement.
Organic growth during a big ad campaign may not be caused by SEO alone.
Show the working
Clients trust numbers they can follow. ARIA connects visibility and conversion data so agencies can present return calculations with clear sources.
A simple SEO ROI formula
SEO ROI = (value of organic conversions minus cost of SEO) divided by cost of SEO. Value comes from the client's own figures: average order value for stores, or lead to customer rate multiplied by customer value for service businesses.
Example: if organic search produced 40 qualified leads in a quarter, 25% became customers and each customer is worth $1,500, organic value is $15,000. If SEO cost $6,000 for that quarter, ROI is ($15,000 minus $6,000) divided by $6,000, or 150%. Every number in that example must come from real records before it appears in a client report.
Attribution: what you can and cannot claim
Analytics tools assign credit differently depending on the attribution model. A customer who first found the site through search, then clicked an ad and finally returned by email may be credited to any of those channels.
Be explicit about which model you use, and avoid claiming that SEO caused growth that coincided with a large ad campaign, a price change or a seasonal peak. Stating the uncertainty builds more trust than an inflated number.
Time horizons matter
SEO work often takes months to show its full effect, while costs start immediately. Report ROI over a period long enough to be meaningful, and show the trend rather than a single month.
Agree the measurement method with the client at the start, so the conversation at renewal is about results rather than about how they were counted.
Check the measurement chain before calculating return
Follow one enquiry from the landing page to the sales record before trusting a report. Submit a clearly labelled test enquiry, confirm that the event fires once, and check that the sales team can identify its source. Then remove or exclude the test. Repeat this for phone enquiries and bookings if those are important routes to purchase. A form submission and its confirmation page must not become two separate leads.
Write down where each number comes from and who maintains it. Analytics might supply the channel, the customer database might supply the sales outcome, and finance might confirm collected revenue. Agree how records are joined without putting personal information into analytics. If consent choices or missing identifiers leave gaps, describe those gaps. A tidy spreadsheet cannot recover information that was never collected.
Choose revenue or profit deliberately
Revenue return and profit return answer different questions. Revenue describes sales associated with the channel. Profit considers what remains after the relevant costs. Ask finance which basis the client wants before presenting either as the investment case. Do not compare a revenue calculation from one quarter with a profit calculation from another. Label forecasts separately from money already received.
For example, a hypothetical store might record $10,000 in organic sales and $4,000 in contribution after product costs, fulfilment and returns. With $3,000 in SEO costs, a calculation using contribution would show $1,000 remaining before other overheads. That is a different story from treating every sales dollar as value available to pay for marketing. The example is a planning illustration, not a benchmark or a reported result.
Reconcile costs and sales on a repeatable schedule
Include the costs that belong to the agreed scope: agency fees, specialist development, commissioned content and any dedicated tools. Decide how shared costs are allocated rather than quietly excluding them. For internal work, agree whether staff time is included and how it will be recorded. Keep that rule stable so a change in accounting does not look like improved marketing performance.
At each reporting close, ask sales which enquiries became customers and ask finance whether cancellations, refunds or unpaid invoices change the value. Keep an initial estimate and a later settled view when sales take time to close. For example, a hypothetical consultancy could review enquiries by the month they arrived, then update their outcomes after the normal sales cycle. This prevents recent, unresolved leads from looking like failures.
Turn uncertainty into a decision rather than a stronger claim
Separate observed return from incremental return. Observed return assigns value using your chosen reporting method. Incremental return asks what would not have happened without the work. A normal analytics report does not establish that counterfactual. Where practical, compare a defined rollout with an unchanged group of similar pages, while recording differences in demand and promotion. Treat the comparison as imperfect evidence, not proof.
End the review with a decision threshold agreed with the client. Continue work that supports a credible commercial case, investigate weak lead quality, or repair tracking before increasing spend. Record the reason and the next review date. Avoid filling missing sales data with optimistic lifetime value assumptions just to make the retainer appear profitable. An honest measurement repair plan is more useful than a precise number built on guesses.
Keep a separate note of measurement changes, including event renaming, consent settings and sales pipeline definitions. Review it before interpreting any sudden movement. If a new tracking setup captures more enquiries, explain that coverage improved rather than describing the whole increase as new demand. Ask the client to confirm the final calculation and its limitations before the report circulates to other stakeholders.
Common questions
What is a good SEO ROI?
It depends on margins and timeframe. Focus on whether the return justifies continued investment for that client.
Can SEO ROI be measured exactly?
Rarely. Present a reasonable estimate with stated assumptions.
How should refunds affect SEO ROI?
Use net sales or another finance-approved value and apply refunds consistently. If refunds arrive after the reporting period, update the earlier estimate or show a clearly labelled adjustment in the next report.
Can I calculate return without access to sales records?
You can report tracked enquiries and acquisition costs, but a revenue return needs verified sales values or explicitly labelled assumptions. Request a limited, aggregated sales export rather than presenting lead volume as profit.