We run gosim.ch, a small shop that sells travel eSIM data plans. In June we turned on Google Ads for the first time. About 490 clicks later we had one paid order, a $7.99 plan. Here's what our analytics showed when we went back through it, and what we changed.
Where the clicks actually came from
We sell to people travelling to Turkey, Japan, Thailand and Switzerland. Our ad traffic in June and July looked like this (Umami, by visitor country):
- India: 2,662 visits
- Indonesia: 270
- Cambodia: 244
- Vietnam: 203
- Laos: 124
- United States: 30
- Switzerland: 13
So roughly 9 out of 10 visits came from countries where almost nobody is buying a travel eSIM for a trip abroad from a Swiss shop. 94% were on mobile. That pattern most likely means the campaign was showing on cheap display and app placements, not search results. We hadn't turned those off, and the automated campaign type happily spent there.
The funnel
Across about 3,900 visits in that period:
- 3,475 landed on the homepage
- 407 picked a destination
- 41 clicked through to checkout
- 1 paid
The homepage was the landing page for almost every ad. Someone searching "esim italy" got a generic page and had to search again. The few ads that pointed to a destination page (Italy got 68 entries) at least landed people on a plan list.
What we changed
- We stopped the ads. With one order, there was nothing to tune.
- We put the effort into pages that answer one trip each: "best eSIM for Turkey", "best eSIM for China" and so on, with real prices next to Airalo, Nomad and Saily, and a short section on what actually works there (for China, which apps load without a VPN).
- We added funnel events (destination picked, plan clicked, checkout started) so the next test can be judged in days, not months.
Those guide pages are now our top landing pages, with no ad spend. Over the last three days the China guide was number one, and China plans were our best seller.
If we run ads again
- Search campaigns only, no display or partner network placements.
- Location targeting set to "people in" the countries we sell to, not "people interested in".
- Every ad goes to the matching destination page, never the homepage.
- A kill rule written down before starting: a fixed number of clicks with zero orders and it stops.
None of this is new advice. We just had to watch 490 clicks go to the wrong people to take it seriously.
GoSIM team
Top comments (2)
The kill rule written down before starting is the part I'd underline. Without it, 490 clicks turns into 900 because each week feels like it might be the one. Did excluding by location alone fix most of it, or did you also have to cut search partners and display placements to get the country mix sane?
Thanks for sharing the real numbers. The funnel from your figures: 3,475 homepage visits out of about 3,900 (89%), 407 destination picks (10.4%), 41 checkout clicks (10.1% of picks), 1 purchase (2.4% of checkout clicks). One thing I could not reconcile: the title says about 490 ad clicks, but the funnel uses about 3,900 visits. Were the other visits non-ad traffic, or do ad clicks and analytics visits count differently? If you can split the funnel by campaign type (Search vs. display-style placements) and by country, that would show whether the 94% mobile and foreign traffic came from placement or from targeting.