The Trip Your App Doesn't Show You

The study behind the map, in plain language.

The fastest bus trip from Providence Hospital to Woodward & Warren arrives at 1:42 PM. Google never showed it. Moovit never showed it — not first, not fifth, not anywhere on the list.

Three buses: the 10, the 3, the 8. Leave at 12:45, arrive at 1:42. Google's best answer got there at 1:55. Moovit's at 2:01. And route 3 — the middle bus, the one that makes the whole trip work — appeared in none of the 25 itineraries the outside planners returned between them.

One trip could be a fluke. So we tested it properly.

First, the rules

We wrote the plan before we ran a single check. Every trip picked by rule, not by us. No cherry-picking after seeing the answers.

We built every best journey we could construct among 15 key places — transit centers, busy corners, a hospital, campuses. 630 journeys. 88 of them needed two transfers. A fixed rule picked 24 of those 88 to test. For comparison, we added 8 best journeys needing only one transfer, matched as closely as we could.

Then we asked Google and Moovit for every trip, at the exact day and time the plan called for. 64 checks in all.

The score

Of the 24 best two-transfer journeys, Google showed 2. Moovit showed 5.

Scale that up to all 88, and the best estimate is stark: about 89 of every 100 of these journeys are invisible to Google. About 78 of every 100 are invisible to Moovit.

The test group is small, so here is the honest range, not just the headline. Google's true figure sits somewhere between about 56 and 97 of 100. Moovit's, between about 44 and 94. At minimum: more than half of these trips go unseen. At best estimate: the large majority.

One transfer? No problem

Now give the same apps a journey with one transfer. Two buses, one change.

Both showed 7 of 8.

Same apps. Same city. Same test. The failures are not random. They pile up exactly where a journey needs two transfers.

Blame the middle bus

Look at which trips went missing. Some have a middle bus doing the real work: take it out of the network and the trip falls apart, or gets much slower. Others have a middle bus you could delete for free.

Google showed none of the first kind. Not one of 14. Its only two hits were the free-to-delete kind — and even those showed up only on trips arriving 27 and 28 minutes later than the best one.

The rider never knows

Picture the rider. No car. Phone in hand. She asks the app how to get across town.

If the fastest trip needs two transfers, chances are the app never shows it. She is not offered a slightly worse trip. The trip itself is hidden — and she never learns it existed. She just concludes the bus takes forever. Or doesn't go there at all.

For a rider who depends on the bus, that is the difference between a city that works and a city that doesn't.

Checked, re-checked, reviewed

We wrote the plan before testing. We picked every trip by rule. When a few early checks turned out to have run at the wrong time or the wrong day, we threw them out and ran them again. Two re-runs changed answers in the apps' favor — those are the answers in the results. Then three reviewers who had no part in the study re-checked every number in it.

The home team does its own math

Detroit's own planner — built into the Live Transit Detroit map on this site — does its own math, straight from the bus schedules. Google's answers and Moovit's answers are never fed into it. It found the 1:42 trip because it read the timetable, not the apps.

The full technical record — every trip, every number, every method — is Installment 07 of Live Transit Detroit, "The Invisible Middle Leg."

Full technical record: Live Transit Detroit, Installment 07 — "The Invisible Middle Leg," at detroitautomationacademy.com/live-transit-detroit/.

The Signal is the newsletter of Detroit Automation Academy — where Detroit builds what's next. If you know a young builder who should be in the room, send them our way.