A Clinic's Success Rate: What the Number Is Counting

Every page you read on the way to a decision will have a number on it. A clinic’s homepage has one, an agency’s brochure has one, and the destination’s own marketing usually has the best-looking one available.

The number is usually not false. That is the part most explainers get wrong, because it makes the reader’s job sound like catching a lie. The number is normally a real figure taken from a real table. What it almost never says is which population it counted, and that is the only thing that decides whether it tells you anything about yours.

Three things have to be read together with any published success rate: what the numerator counts, what the denominator counts, and who was left out.

What the numerator is counting

Pregnancy and birth are different outcomes, and the gap between them is not small and not secret.

In its report on UK treatment data, the regulator states that birth rates run typically around three percentage points below pregnancy rates, and recommends using pregnancy rates when judging recent years precisely because birth outcomes arrive later and need validating first. A “success rate” that never says which of the two it means has not told you its numerator.

There are then several further numerators in circulation: a clinical pregnancy, a pregnancy continuing past a given week, a live birth, a live birth delivering one living infant, and a singleton live birth. The last two are different numbers, and a table that counts a twin birth as one success and a table that counts two infants are not comparable.

What the denominator is counting

This is where a number moves most, and it moves in one direction: whichever denominator the report chooses, the denominator it picked is the smallest reasonable one.

The UK regulator’s own headline figure is stated per embryo transferred: the average IVF birth rate in 2024 was 30% per embryo transferred. Read what sits underneath that figure and the same page shows the shape of the problem: 38% for patients aged 18 to 34, and 8% for patients aged 43 to 44. One number, three facts. A single average across those two groups would describe neither.

Now change the denominator and the same data becomes a different figure. Common alternatives are per cycle started, per oocyte retrieval, per transfer and per patient, and they differ mostly by what they leave out. A per-transfer figure quietly excludes every cycle that never reached a transfer, which is exactly the group that matters most to anyone asking about their own odds.

This is the same arithmetic that makes a comparison between two providers meaningless before their scopes match, which is why the quote comparison tool refuses to read a gap between two totals as a cost difference. A gap between two rates is no better until their denominators match.

Who is not in the table

This is the section that matters most for anyone planning around a surrogate, and it is stated in the smallest print on the page.

The notes under the UK regulator’s success-rate figures say plainly which treatments have been taken out. For the birth-rate chart in the 2024 report, the exclusions listed are treatments using PGT-A/M/SR, treatments using donor eggs or surrogacy, and cycles where a pregnancy was recorded but no birth outcome was. The 2022 report carries the same exclusion in its chart notes.

So the national figure that a provider may quote at you as “the success rate” does not include surrogacy cycles. That is not concealment; it is a defined dataset with a defined scope. But it does mean the number was never about your path, and anyone presenting it as one has substituted a population you are not in for the one you are.

Two further points from the same reports belong here, because they tell you how much weight an official table can carry:

• The data can be provisional. Birth rates for 2020-21 and 2024 are marked preliminary and not yet validated, and in the earlier report the unvalidated years could not be compared with other reports at all. One clinic’s data was excluded entirely for reporting issues.
• Populations are not interchangeable. The same regulator reports that in 2022-24 the average birth rate per embryo transferred for Asian and Black patients aged 18-37 ran 5 to 6 percentage points below that of White patients, and its chart notes warn that the patient populations behind fresh and frozen transfers differ enough that their rates are not comparable. A rate borrowed from a group you are not in is an estimate of the wrong thing.

What surrogacy-specific numbers can exist at all

If the national tables exclude surrogacy, is there a figure that includes it?

In England, Wales and Northern Ireland treatment is reported to a regulator, so the volume is known: surrogacy accounted for 0.4% of all IVF treatments in the UK in 2022, with the number of people acting as surrogates rising from about 130 in 2012 to about 230 in 2022. Four in a thousand is what a “surrogacy success rate” from a single clinic is usually calculated on: a few dozen cycles, sometimes a handful, gathered over however many years the marketing chose.

The same factsheet states the one surrogacy-specific outcome figure the regulator does publish, and it is not a success rate: the multiple birth rate for surrogacy patients fell from 14% in 2013-2017 to 5% in 2018-2022. It also gives the demographic that makes that figure meaningful: the average age at first IVF treatment among surrogates is about 34. And, importantly for anyone reading a UK number as an endorsement, the regulator says its remit covers how licensed clinics deliver treatment and consent, and does not extend to surrogacy arrangements made outside the UK.

In the United States there is no such regulator, but there is a registry, and it has a different limit. SART CORS holds cycle data from more than 90 percent of US clinics, recording cycle starts, patient demographics and linked cycles. It does not record the agreement, the payment or the program’s outcome. So the registry can support a clinical statistic and cannot support a program-level one, the distinction drawn at length in what 40,177 transfers do not promise you.

One number from that registry does bear on international intended parents directly: of 40,177 embryo transfers to a surrogate from 2014 to 2020, 32% were for intended parents living outside the United States. Read it as a caution rather than a comfort. A national rate for that jurisdiction is an average across a population in which roughly a third of the activity belongs to people who then have to take a child home across a border, whose ages, diagnoses and treatment routes are not yours.

What to ask instead

The useful move is not to find the highest number. It is to find the cell a number came from and ask whether you are in it.

1. Ask for the outcome, not the success. Which is counted: clinical pregnancy, birth after a stated gestational week, live birth, or singleton live birth?
2. Ask for the denominator. Per cycle started, per retrieval, per transfer, or per patient, and ask what each of those excludes. The gap between per-cycle-started and per-transfer tells you how many cycles never reached a transfer.
3. Ask for your age band and your group. Not an overall average, and not the youngest band. If a table reports by age, ask for the band you are in and for the count behind it.
4. Ask what the table excludes. Surrogacy, donor gametes, PGT and frozen-cycle outcomes are commonly separated or removed. If your plan involves any of them, a general table is not your evidence.
5. Ask how current it is and whether it is final. Which year of treatment does it report, when was it published, and has the data been validated? Regulators mark provisional years; a clinic quoting a provisional year as settled is doing arithmetic on unfinished input.
6. Ask for the count as well as the percentage. “60% of 12 patients” is a real number and is not a rate you can rely on. Denominator size is the first question an evidence-based claim should survive.

A provider who answers all six with specifics is not more likely to succeed for you. It is likelier to be telling you something you can use, which is the whole point of asking. These six belong in writing beside the rest of the questions in the agency checklist, and any figure that arrives with a name and a year attached can go onto the same record of published cost figures style you would use for money: source, publisher, year, and whether the publisher sells the service.

If you are holding a specific figure and cannot tell which cell it came from, that is a question worth sending. Say which page it is on, quote the sentence, and name the population you are trying to place yourself in.

What is explained above is how published fertility treatment data is constructed and what its stated limits are. It is not medical advice, and no figure here is an estimate of your own chances or of the cost of your program.