The channel with no invoice, no click ID and half its keywords hidden

Ask a marketing team which channel brings in the most customers and a surprising number of them will say organic search. Ask them what it earned last quarter and the room goes quiet. They can tell you sessions. They can tell you rankings, down to the position of individual phrases on individual days. They can usually tell you how many form fills analytics credited to it. What they cannot tell you is the figure they could give in a heartbeat for Google Ads: closed revenue, by the channel that produced it.

That is not because the team is careless. It is because organic search is missing the two things every other channel's reporting is quietly built on. Nobody sends an invoice for it, so there is no spend to anchor a return. And it carries no click identifier, so there is no thread that runs from the visit to the platform's own record of it. Paid media reporting is a system for reconciling those two things. Take them both away and most of the machinery has nothing to hold.

Organic search is the channel most businesses depend on most and can defend least, because it arrives without an invoice, without a click ID, and with half its keywords withheld.

What organic search does not give you

An organic visit arrives with less information attached than any paid click. There is no campaign, no keyword, no cost and no identifier, only a referrer saying the visitor came from a search engine. Everything a report says about organic search beyond that has been reconstructed, not received.

It is worth being precise about the gaps, because they are different in kind and they call for different responses.

• No cost line. Paid channels hand you a denominator. Organic search makes you build one from retainers, salaries and tools, and most teams never do, which is why SEO is defended with traffic charts instead of returns.
• No click identifier. A paid click carries a gclid or an equivalent, which lets a conversion be reported back to the platform and lets the platform tell you about the click. An organic click carries nothing of the sort.
• No keyword. The query that produced the visit has not been passed to websites since 2011 for signed-in users, and for practically everyone since. What you know is the page they landed on.
• No platform-side report of outcomes. Google Ads will tell you, imperfectly, what it thinks converted. Search Console will tell you what was clicked and stops there. There is no organic equivalent of a conversion column.

None of this makes organic search unmeasurable. It makes it unmeasurable by the methods that work for paid media, which is a narrower statement and a more useful one.

What Search Console knows, and the half it keeps

Search Console reports queries, impressions, clicks and average position for your site in Google Search, in aggregate and without any notion of an individual visitor. It also withholds queries made by too few people, which in practice is close to half of all clicks.

Search Console is the best first-party view of organic demand that exists, and it is routinely asked to do a job it was never built for. It is a report about searches. It knows that a query was shown your page four thousand times and clicked ninety. It has no idea who any of those ninety people were, whether any of them rang you, or whether the one who did is now a customer. There is no join key in it, by design.

It is also incomplete in a way that matters more than most teams realise. Google omits what it calls anonymised queries: those not issued by more than a few dozen users over a two to three month period. They are counted in the chart totals and left out of the table, which is why the rows never add up to the headline. For a business with a long tail of specific, local or technical searches — which describes nearly every business that sells something considered — that long tail is exactly what is hidden.

Search Console is a report about searches, not about customers. It cannot tell you what organic search earned, and it was never trying to.

Why analytics over-credits and under-credits organic at the same time

Web analytics assigns a session to organic search when the referrer is a search engine and no campaign parameters are present. That rule hands organic search credit for demand other channels created, and strips it of credit for sales that close somewhere analytics cannot see.

The over-crediting is mostly branded search. Somebody hears about you from a neighbour, sees the van, gets the postcard or watches the video, and a week later types your company name into Google and clicks the first unpaid result. Analytics calls that organic search. In any sense that matters for budgeting, it was a referral, or direct mail, or video. Organic search was the road they drove in on. A healthy share of what reports as organic is the echo of everything else you do.

The under-crediting is everything that happens after the visit. The person who found you through a genuinely non-branded search — "roof leak repair near me", "orthodontist for adults", "commercial HVAC maintenance contract" — reads two pages and then picks up the phone. Analytics records a session that ended without a conversion. The call is answered, a quote goes out, and a contract is signed three weeks later in a system that has never heard of the session. The channel did its job perfectly and appears, in the only report that tracks it, to have done nothing.

Read down the right-hand column and the pattern is plain. The tools that know about search know nothing about money, and the system that knows about money knows nothing about search. Every honest answer to "what did SEO earn" is a join across that gap.

Capturing the source when there is no click ID to capture

The source of an organic lead has to be recorded at the moment of first contact, on the lead record itself, because nothing downstream can reconstruct it. For a form that means hidden fields; for a call it means the call tracker's session data. The landing page stands in for the keyword.

With paid media you can be lazy about this, because the platform remembers the click for you and an identifier ties the two together. With organic search there is no second copy of the record. If the source is not written down when the person first makes themselves known, it is gone, and no amount of analysis later will bring it back.

1. Capture the first landing page, the referrer and any campaign parameters when a session starts, and hold them for the length of the visit.
2. Write them into hidden fields on every form, so the lead arrives in the CRM already carrying its origin. A lead whose source is a dropdown filled in by a salesperson afterwards is not a measurement.
3. Use call tracking with dynamic number insertion and make sure organic sessions get their own number pool. The call record then carries the source and the landing page the same way a form does.
4. Store the landing page as a field of its own. With the query withheld, the page a person arrived on is the best available evidence of what they were looking for: a visitor who lands on the emergency repairs page was not researching your company history.
5. Separate branded from non-branded at this point if you can, by landing page. Homepage and contact-page landings from search are overwhelmingly branded; service, location and article pages mostly are not. It is a proxy, and a decent one.

That last step deserves defending, because it sounds cruder than it is. You will never recover the actual query. But the structure of a site does a lot of the work for you: people searching for your name land on the homepage, and people searching for a problem land on the page about that problem. Grouping organic leads by landing-page type gives a split between navigational and discovery traffic that is good enough to budget on, which is more than the query list ever offered.

Matching organic leads to closed sales

Once leads carry their source, attributing revenue to organic search is the same operation as for any other channel: match each closed sale to the lead that preceded it, on a normalised phone number or email address, and let the sale inherit the lead's source. No click ID is involved at any point.

This is the part that ought to be reassuring. The reason organic search looks unmeasurable is that people try to measure it the way ad platforms measure themselves, by following an identifier from click to conversion. That thread does not exist for an unpaid click and never will. But a match between a lead file and a sales file never used that thread in the first place. It uses the person: the phone number they called from, the email they typed into the form, compared after both have been normalised to one format.

So an organic lead and a paid lead are, to this method, the same kind of object. Each is a row with a contact detail, a date and a source. When a sale closes with a matching contact detail, the revenue lands on whichever source the lead carried. Organic search stops being the special case and becomes one more value in a column — which is what lets it be compared, for the first time, against the channels you are paying for.

This is the comparison CloseRev exists to produce. Export your leads and calls with the source each one carried, export your closed sales, and it matches the two on normalised phone and email so that organic search sits in the same revenue-by-channel table as your paid campaigns. Sales that match no lead stay in Direct / Unknown rather than being handed to whichever channel looks most plausible.

Two cautions, both learned the hard way by teams that did this and then misread the result. First, the match window has to be long. Organic search skews towards considered purchases with slow cycles, and a thirty-day window will credit the fast, cheap jobs and miss the large ones. Second, first contact is not always first touch. Somebody may have clicked a paid ad in March, not enquired, and come back through an organic result in May to fill in the form. The lead's source will say organic. That is a true statement about how they arrived when they finally made contact, and it is worth knowing that it is not the whole story.

Giving a channel with no invoice a return

Organic search has a cost even though nobody bills you for clicks: the agency, the writers, the tools and the salaried hours. Dividing attributed revenue from non-branded organic search by that total, over a year rather than a month, gives a return that can sit beside paid media in the same report.

The absence of an invoice has an odd effect on how SEO gets discussed. Because there is no spend figure in a dashboard, it is treated as free, and because it is treated as free, nobody asks what it returns. Then budget season arrives, the content retainer is the line nobody can defend with a number, and it is cut in favour of a channel whose dashboard produces one. The channel was not underperforming. It was under-measured.

• Add up what organic search really costs: retainers, freelance and in-house content, technical work, tools, and a fair share of the salaries involved.
• Use non-branded attributed revenue as the numerator. Including branded search flatters SEO with demand that other channels bought.
• Measure over twelve months. Work done in the spring earns in the autumn, and a monthly return on SEO is noise.
• Report the match rate alongside it. A return built on a sixty per cent match rate is a floor, not a point estimate, and saying so is what makes the number survive a finance review.

The figure you end up with will be rougher than the return on a search campaign. It will also very often be the best return in the table, which is the whole reason to go to the trouble. A channel that compounds, that does not stop the day you stop paying, and that turns out to close more revenue for what it costs than anything else you do is a channel worth being able to prove.

What you will never know, and should say so

Some things about organic search are not recoverable by any method: the exact query, the visits that never became a contact, and the influence of a search result on somebody who later arrived another way. An honest report names those limits instead of modelling over them.

There is a temptation, once the match is working, to push it further than the data goes — to apportion the unmatched sales across channels in proportion to the matched ones, or to infer keywords from landing pages and present them as fact. Resist both. The value of a number built from your own sales records is that it can be checked, row by row, by somebody sceptical. The moment part of it is estimated, all of it is open to argument.

What you can say is already a great deal. This many customers who paid us first made contact through a non-branded organic search result. They landed on these pages. They were worth this much, against this much spent. This share of our sales we could not trace to any source, and we have not guessed. Most businesses have never been able to say the first sentence of that about the channel they depend on most. Being able to say all five is a different conversation with whoever holds the budget.

You will never get the keyword back. You can still know which customers organic search brought you and what they paid, and that is the number the budget meeting actually needs.

Questions people actually ask

Why is organic search harder to attribute than paid search?

Paid search attaches a click identifier to every visit and sends you an invoice, so there is both a thread to follow and a cost to divide by. Organic search does neither. A visitor from an unpaid result arrives with a referrer that says only that they came from Google, the query is withheld, and nobody bills you for it, so the two numbers most attribution is built from are both missing.

Can Search Console tell me which keywords produced revenue?

No. Search Console reports queries, impressions and clicks in aggregate and has no concept of a person or a sale. It also omits a large share of queries for privacy reasons: Ahrefs measured 46.77% of clicks going to queries Google did not report, across 22 billion clicks in April 2025. It is a good tool for understanding demand and a poor one for understanding money.

How do I attribute a closed sale to organic search?

Record the source when the lead is created, on the lead itself: the landing page, the referrer and any campaign parameters captured into hidden form fields or the call tracker's session data. Later, match the closed sale back to that lead on phone number or email. The sale inherits the lead's source. That works for organic search exactly as it does for any other channel, because it never depended on a click ID.

Is branded organic search really organic?

It arrives through an organic result, but the demand was usually created somewhere else: a radio ad, a van, a referral, a paid campaign the week before. Somebody typing your company name is navigating, not discovering. Report branded and non-branded organic separately if you can, and treat the branded share as a measure of how well everything else is working rather than as a win for SEO.

What is a sensible return figure for SEO if there is no ad spend?

Use what you actually pay: agency retainers, content production, tooling and the salaried time of whoever does the work. Divide closed revenue attributed to non-branded organic search by that figure over a period long enough to be fair, which for most businesses is a year rather than a month. It will be rougher than a paid-search return and it is still far better than not having one.

Does AI search change any of this?

It makes the missing data more missing. Answers that resolve a question without a click leave no visit to attribute, and visits that do arrive from an AI assistant often carry even less referrer information than a search result does. The method that survives is the one that never relied on the referrer being rich: capture what you can at the moment of contact, and match people rather than sessions.

Found this useful?

• Share on X
• Share on LinkedIn
• Share on Reddit
• Share on Email