Vehicle marketplaces: the setup, step by step
For the marketplace where the sale closes offline: the tag, the catalog by API, the booking event, two to three weeks of baseline, and the numbers finance will accept.
- What to prepare before day one
- What goes live, in order
- What to measure in week one and week four
Open the setup checklist
Enter your work email and the checklist opens on this page. We also send you the link, so you can come back to it.
Thanks. The checklist is open below.
We use your email to know who reads our research. Privacy policy
What to prepare
One tag on the site
Views, searches, listing clicks and the booking form. The baseline and every result come from these events.
The catalog by API
Listings with status. A sold vehicle must leave the recommendations the hour it is sold.
The booking event
Define what counts: a booked test ride, with its time. Clicks and chats are not results.
Two to three weeks of baseline
Bookings per visitor measured from your own events before anything changes, plus read access to the tools you already run.
What goes live, in order
Measure the baseline
Visitors, listing views and bookings per visitor for two to three weeks. Nothing else changes yet.
Recommendations on live listings
Similar vehicles nearby, ranked per buyer from what they viewed. Sold listings excluded by status.
Triggered banners
A banner for the buyer who viewed several vehicles and has not booked, on the listing and the home page.
Abandoned-browse and abandoned-sell email
For buyers and sellers who left before the step, with the vehicles or the listing they looked at.
Web push from zero
Ask for permission after a listing view, not on arrival. The channel grows from nothing within weeks.
Attribution to the booking, and a holdout
Every result traced from the click to the booked test ride; a holdout group set up automatically turns the result into measured lift.
What to look at, and when
| When | What to look at | What good looks like |
|---|---|---|
| Weeks 1 to 3 | The baseline: visitors, listing views, bookings per visitor, the booking event firing | A stable bookings-per-visitor rate to compare against, in equal windows |
| Week 4 and after | Bookings per visitor against the baseline, the test-ride rate of viewers who saw a recommendation, push subscribers | The pattern from the motorbike marketplace: more test rides per visitor in equal windows, even with fewer visitors |
| Month 3 | The holdout lift, and attributed bookings by placement: recommendations, banners, email, push | A measured lift in bookings, with every booking traced to one placement |
Common mistakes
Counting clicks or chats as results. The conversation goes cold; only the booked test ride counts.
Comparing unequal windows. Twenty-two days against twenty-two days, as the marketplace did; not a quiet month against a busy one.
Recommending sold vehicles. The catalog status must flow by API, not by a nightly file.
Skipping the baseline. Without it, more bookings in a good month look like the agent's work, and fewer in a bad month like its failure.
Asking for push permission on arrival. Most visitors refuse before they have seen a vehicle; ask after a listing view and the channel grows.
See which of last week's visitors you missed
Thirty minutes, with your store and ad accounts open. Then 30 days free. A holdout group decides: if the agent doesn't add orders in 30 days, you don't pay.
Book 30 minutes
You'll talk to one of us.