Which channels produced started cycles?

Fertility patients research for months, consult more than one clinic, and often wait for funding or a life event. Almost none of that is visible to click attribution, and the campaign that produced the patient looks like it produced nothing.

Matching on the patient rather than the session credits the started cycle to the original inquiry whenever it arrives. For a segment with this cycle length, that is the difference between measuring marketing and guessing.

This is among the most sensitive data any clinic holds, which is exactly why the match takes only a contact detail, an amount and a date. No diagnosis, no protocol, no outcome.

Nothing is installed on the site either, so there is no script on a page where a patient is reading about their own fertility. That property matters more here than almost anywhere else.

Fertility pricing rarely resolves into one clean figure. A patient may buy a single stimulated cycle, a two- or three-cycle package at a discount, or a program with a refund attached if no birth results. Medication is frequently billed by a separate pharmacy and never touches the clinic's ledger at all. Anesthesia, genetic testing and storage may each arrive as their own line. So the question of what a channel produced has an answer that depends entirely on which of those files somebody exported, and the difference between the narrowest and broadest reading can be a factor of two.

The defensible figure is what the clinic itself collected, on the date it was collected. Money that passes through to a pharmacy or a laboratory is not the clinic's revenue and including it inflates every source, unevenly, as soon as two campaigns produce different treatment mixes. A package bought up front should be exported on the day it was paid, not spread across the cycles it funds, because the payment date is the one that sits near the decision an advert influenced.

Packages complicate the counts more than the revenue. One patient buying three cycles is one decision and three episodes of care, and every row in the sales file is counted as a sale, so a package exported as one row per cycle makes a channel look like it produced three patients. Exporting one row per package payment keeps the count honest, which matters here because clinics are usually comparing sources on cost per new patient rather than cost per procedure, and a threefold error in the denominator is not survivable.

A growing share of fertility demand is routed by an employer's benefit administrator. Somebody is told their plan covers treatment, given a network of clinics, and arrives having already chosen from a list. In the clinic's own data that patient appears as a direct inquiry or a branded search, because that is literally how they got to the website, and every advertising report will therefore credit whichever campaign happened to be running on the clinic's own name. The clinic's marketing lead sees a rising line on a branded campaign and reads it as demand it created, when it is demand a human resources department created on its behalf.

The consequence is a budget that drifts toward branded search on the strength of patients who were never available to any other channel. It is a self-confirming error: the more of the routed demand branded campaigns intercept, the more efficient they appear, and the more they are funded. Clinics can spend years bidding against themselves this way without anything in the data suggesting they should stop, because nothing in the click record distinguishes a patient who was sent from one who was persuaded.

Asking at intake and carrying the answer into the file is the only reliable fix. Record the benefit program or employer as a source, list it alongside Google, Meta and physician referral, and the report ranks it on the revenue actually collected. The routed share may turn out to be the largest single source in the practice, which would be genuinely good news about the contract and rather bad news about the advertising budget that had been taking credit for it.

Elective freezing behaves nothing like treatment cycles. The patient is usually younger, the decision is made without a diagnosis in the room, and the money arrives in two very different shapes: a retrieval paid for once, and a storage fee paid every year afterwards, sometimes for eight or ten years. Beyond that sits the possibility of a thaw and a transfer at some undetermined point, which may be the largest revenue event of the whole relationship and may never happen at all.

No advertising measurement ever built survives that timeline, and clinics know it, which is why freezing campaigns tend to be judged on retrieval revenue alone and quietly compared against treatment campaigns that resolve within months. That comparison is not close to fair. It ignores a decade of storage fees and the eventual cycle, and it systematically undervalues the campaign that recruited a patient who will be in the practice's system longer than most staff members. It also means the two campaign types are being funded from one budget on the strength of a comparison that was never valid, and the imbalance grows each year the freezing cohort ages.

Because every storage invoice carries the same patient's contact details, exporting them as ordinary revenue rows credits them to the inquiry that produced the retrieval, however long ago that was. Running the report across several completed years is the honest way to compare freezing against treatment, and the records to do it already exist — they were kept to bill the storage. Export them as plain rows with a contact detail, an amount and a date, and nothing marking what they were for: what a payment bought is health information, and CloseRev signs no business associate agreements.