Measure your agency on revenue, not rankings

The monthly report says rankings are up, visibility is up and organic traffic is up. In a different meeting, the finance director asks why online revenue is flat. Both can be true at once, and that is the problem.

Most agency reporting is built around SEO KPIs the agency can move and you cannot spend: positions, impressions, sessions. They are useful to the people doing the work. They are poor evidence that the work is paying back.

Rankings are an input. Revenue is the outcome. This piece sets out why so many agencies report the first, what a marketing director should measure instead, how to set targets without pretending attribution is solved, and what to write into the contract.

Why rankings and traffic make poor SEO KPIs

A ranking report answers a question nobody on the board asked. Positions vary by location, device and search history, so the number in the report is one sample of many. A top three position for a phrase that never converts is worth nothing. Google's own guidance on hiring an SEO is blunt about it: "No one can guarantee a #1 ranking on Google."

Traffic is not much better on its own. For an established brand, a large share of organic visits comes from people searching for the brand by name, who would have found you anyway. Much of the rest is informational and will never buy. Our piece on how brand traffic flatters agency reports shows how to split it out.

AI answers have loosened the link further. When AI Overviews and AI Mode answer more questions on the results page, the same visibility can produce fewer clicks. A traffic target can be missed while the commercial picture improves, or hit while it gets worse.

A ranking is a means. Nobody ever banked one.

Why agencies prefer vanity metrics

It is not always cynical. Revenue depends on things an agency does not control: price, stock, the website, the checkout, the sales team, even the weather. Rankings and traffic move faster, more often and more visibly, so they make for a more comfortable monthly call.

But comfort is the point. A metric the agency can always find good news in protects the retainer. A metric tied to your revenue exposes it. If every month's report is green and the business is not growing, the report is measuring the wrong thing.

There is a pricing reason too. When an agency is paid by the hour or as a percentage of ad spend, activity is the product, so activity is what gets reported.

The SEO KPIs that actually matter

Five measures cover most of what a board needs from search. None is perfect. Together they are much harder to game than a ranking chart.

• Non-brand organic revenue: orders and revenue from organic search, excluding searches for your brand name. It is the closest single number to what SEO adds.
• New customers from search: first orders from paid and organic search, so you are not paying to win back people who already buy from you.
• Blended cost per acquisition: total search cost (agency fees, ad spend, content and tools) divided by orders from paid and organic search together. It stops one channel looking cheap by taking credit from the other.
• AI referral conversions: orders and leads from ChatGPT, Perplexity, Gemini and similar referrers. Small for most brands today, and worth its own line so you can see the trend.
• Share of search: your brand's share of branded searches in the category. When Les Binet presented the measure for the IPA in October 2020, it correlated with market share in all three categories he studied and tended to move ahead of it.

Keep rankings, visibility and technical health underneath as diagnostics. They tell the team where to look. They do not tell the board whether to keep paying.

For an online retailer, the e-commerce visibility playbook shows how these measures fit together across SEO, PPC and AI search.

If you sell to businesses, swap revenue for pipeline

The principle holds when the sale happens offline. Replace orders with qualified pipeline and closed revenue from your CRM, and replace blended CPA with cost per qualified opportunity. Form fills are the B2B version of rankings: easy to count, easy to inflate.

The B2B lead generation playbook and the SaaS playbook cover connecting search to CRM data and trial-to-paid conversion.

How to set targets with honest attribution

Attribution is not solved, and anyone who says otherwise is selling something. Google retired first click, linear, time decay and position-based models in Google Ads and GA4 in 2023, leaving data-driven and last click. Those two attribution models can give different answers for the same sale, and neither sees everything.

So set targets the way a finance team would:

1. Agree a baseline. Take at least 12 months of data, split by brand and non-brand, paid and organic, so seasonality and existing demand are visible before anyone claims credit.
2. Write down the definitions. What counts as a brand search, a new customer and a conversion, and which attribution model the report uses. Change them only by agreement.
3. Set a range, not a number. A cautious, central and strong case, each with its assumptions. Judge the agency on whether results land in the range and whether it can explain why.
4. Test what you can. For paid search, incrementality testing with holdouts shows what would have happened anyway. For organic, compare the pages and categories the work touched with those it did not.
5. Review revenue quarterly and inputs monthly. Organic revenue moves too slowly and noisily for a monthly verdict. Leading indicators can be read monthly, as long as everyone knows that is what they are.

What a one-page report should show

The order of a report tells you who it was written for. If the first page is a ranking chart, it was written for the agency. A report written for you reads in this order:

• Revenue first: non-brand organic revenue and new customers from search, against the agreed target range.
• Cost next: total search cost and blended cost per acquisition, with the direction of travel.
• Then the leading indicators that explain the movement: non-brand visibility on commercial terms, AI share of voice and technical health.
• Then what changed, what the team is doing about it, and what it needs from you, whether that is developer time, budget or a decision.
• Finally the caveats: the attribution model used, any tracking changes and anything unusual in the period, such as a promotion or a stock problem.

What to put in the contract

If the KPIs only live in a slide, they will drift. Put them where the money is agreed.

• The outcome KPIs, with definitions, baseline and target range, in the statement of work.
• Data access both ways: the agency gets read access to revenue or CRM data, and you own every account it works in.
• A fixed fee for a defined scope. Paying by the hour or as a percentage of spend rewards activity, not results.
• A paid research phase before any retainer, so targets are set from your data rather than a pitch deck. We make the case for that in our piece on feasibility reports.
• A quarterly review against the KPIs, ending in a written decision: continue, change scope or stop.
• Notice terms that let you act on that decision.

But revenue is not in the agency's control

That is true, and it is the strongest argument against this approach. A price rise, a stock problem or a broken checkout can wipe out good search work in a week.

The answer is not to pay purely on results. Payment by results pushes an agency towards the cheapest wins, which usually means brand terms and existing customers. The answer is to measure on revenue, pay a fixed fee, and have an honest conversation every quarter about what moved the number.

Measured on revenue does not mean paid on revenue. But an agency that cannot explain its contribution to revenue cannot really explain its fee either.

Questions to ask your agency

• What share of the organic revenue you report comes from brand searches?
• Which of your KPIs would you be happy to have written into the contract?
• What would have happened to revenue without your work, and how do you know?
• What is our blended cost per acquisition across paid and organic, and which way is it going?
• What will you stop doing if a tactic is not moving revenue?

How we work

We measure our own work on revenue: non-brand revenue, new customers and blended acquisition cost across integrated search, for a fixed, scoped fee, starting with paid research. If you want reporting that survives a finance director's questions, book a growth plan call.

Frequently asked questions

What are the most important SEO KPIs?

The SEO KPIs that matter most to a business are outcomes: revenue and orders from non-brand organic search, new customers acquired through search, and blended cost per acquisition across paid and organic. Share of search and conversions from AI assistant referrals add a view of brand demand and emerging channels. Rankings, traffic and technical health are useful diagnostics underneath, not targets in their own right.

Should you measure SEO on rankings?

Rankings are useful for diagnosing problems but poor as a target. Positions vary by location, device and search history, a top position for a phrase that does not convert earns nothing, and Google says no one can guarantee a number one ranking. Measure SEO on non-brand revenue, new customers and acquisition cost, and keep rankings as a supporting signal.

How do you separate brand and non-brand organic revenue?

Analytics does not report revenue by organic search query, so the split is an estimate. Classify queries in Search Console as brand or non-brand, work out which landing pages each group sends visits to, then apply that split to organic revenue by landing page in analytics. Agree the method and the brand term list once, and keep them fixed so trends stay comparable.

Should search agencies be paid by results?

Pure payment by results usually backfires, because it pushes an agency towards the cheapest wins, such as brand searches and existing customers, which would have converted anyway. A better model is a fixed fee for a defined scope, with revenue-based KPIs, baselines and target ranges written into the contract and reviewed every quarter, so the agency is measured on revenue without being paid on it.

How often should SEO results be reported to the board?

Quarterly for revenue and new customers, because organic revenue moves too slowly and noisily for a monthly verdict. Leading indicators such as non-brand visibility, AI share of voice and technical health can be reviewed monthly by the marketing team. A board report should fit on one page, lead with revenue, and state the attribution model and assumptions behind the numbers.

Sources

• Google Search Central: Do you need an SEO? (last updated 5 June 2026)
• IPA: Binet presents fast, cheap, predictive Share of Search metric (14 October 2020)
• Google Ads Help: First click, linear, time decay, and position-based attribution models are going away (2023)