What is happening on go.mypregnancychoices.com, what its conversion figures actually measure, and what we need from the agency.
The go. subdomain is a set of paid-search landing pages run by the advertising agency. It is not part of the website rebuild, but it accounts for nearly everything analytics records — so any conversation about performance has to start here.
| Page | Sessions | Conversions | Rate |
|---|---|---|---|
| /ultrasounds | 25,433 | 15,775 | 62% |
| /abortion-info-ppc | 5,456 | 560 | 10% |
| /unexpected-pregnancy-ppc | 2,213 | 174 | 8% |
| /abortion-pill-ppc | 1,923 | 109 | 6% |
| /thank-you | 982 | 1,214 | 124% |
Four campaign pages and a shared confirmation page. One of them — the ultrasound page — is doing most of the work: it accounts for 71% of sessions and 88% of conversions on the subdomain.
Listing the recorded conversions by name settles what they are.
17,311 of roughly 18,000 recorded conversions are a visitor tapping a phone number. Form submissions are 1,068 — under six per cent. “Get directions” has fired four times in two years.
A tap is a real signal of interest and worth counting. It is not an appointment. Some taps never become calls, and some calls never become bookings. The headline number should not be reported as appointments.
The device split shows the same thing independently. A form works identically on a phone and a laptop; a tap on a phone number only works on a phone.
Three of the four campaign pages behave normally — roughly 21% on mobile and 17–23% on desktop. Consistent across devices, which is what a form looks like.
The ultrasound page does not. It records 65% on mobile against 1.3% on desktop, a fiftyfold gap on the same page in the same campaign. That page’s primary call to action is the phone number, and the numbers show it.
The confirmation page reinforces the point: 1,227 conversions against 1,044 views. More conversions than visits means more than one event fires per completion, so even the form figure is not a clean one-per-enquiry count.
Plotting the four conversions separately resolves this. Text and Call collapse in June 2025. Form submissions continue at the same steady level throughout, before and after.
So this was never a demand story, and never a problem with the enquiry forms. What stopped was the counting of taps.
Call-tracking software is installed on the site. It rewrites phone numbers on the page as each visitor arrives, so that calls can be attributed to their source. If the analytics tracking was configured to watch for one specific number, that rewriting would silently break it — the links continue to work perfectly for visitors, they simply stop being recorded.
Three different phone numbers are currently in circulation across the website and the Google Business Profile, which is consistent with number-swapping being active.
Asking the agency when the call-tracking software was installed would settle this in one answer. If it was mid-2025, the calls are not lost at all — they are being recorded in that system rather than in analytics. Connecting the two would restore the measurement and improve on it, because that system counts calls that actually connected rather than taps on a link.
Nothing has deteriorated. The enquiry pipeline has been steady throughout; only its measurement changed. The form figures are usable across the whole period, and once call tracking is reconnected the phone figures will be too.
Text recorded 8,656 and Call 8,655 across two years. Two supposedly independent actions landing within one count of each other warrants opening the tags to check they are not both firing on overlapping triggers. The people counts do differ — 3,217 against 2,999 — so it is not a straightforward duplicate, but at 2.7 and 2.9 events per person it is higher than most people would tap.
For a walk-in clinic whose main invitation is to come in, that is a missing trigger rather than a low number.
465 desktop sessions produced 6 conversions. With the phone number as the only call to action, desktop visitors have nothing to tap. A visible form would recover that traffic.
Versions of each page with a trailing slash appear separately — /abortion-info-ppc/, /abortion-pill-ppc/ — with visits but zero conversions. Small numbers, but it indicates conversion tracking does not fire on that variant.
147 sessions to the ultrasound page recorded 170 conversions with an engagement rate of zero. More conversions than sessions, and no engagement at all. That combination is not a real person.
Alongside English there are Spanish page views, and smaller numbers in Russian, Persian, German, French, Indonesian and Romanian. Volumes are tiny, but Spanish is meaningful and consistent with the demand we see on the main website.
None of these are challenges. Each one materially improves what we can do on the rebuild.
It has attracted more than 25,000 visits and been refined over years. Whatever it has learned about how an anxious visitor on a phone responds to a free-ultrasound offer should shape the new service pages. That costs nothing but a conversation.
Form submissions are measured consistently across the whole period and can be used as they stand. Call and text figures are only reliable up to June 2025, and will be reliable again once tracking is reconnected. Mixing the two into a single “conversions” number would mislead in both directions.
Roughly nine in ten sessions are paid. If the new website is measured on total traffic, the paid programme will mask whatever the rebuild achieves — and any change in ad budget will read as a website result. Organic sessions and organic-attributed enquiries need to be reported on their own.