• by Chuck Martin , Staff Writer, July 25, 2012

How well can a mobile phone estimate when you’ll arrive somewhere? 

As a long-time user of location tracking, it always bothered me that my car GPS would tell me I would arrive at a different time than my smartphone did. 

Admittedly not great at either directions or time management, I look to technology to assist in any trips of any complexity or distance.

On a local level, mobile apps have come through for me on more than one occasion. Yelp and Poynt have frequently guided me to favored eating locations while My Car Locator (Android) and Take Me to My Car (iPhone) have guided me back to my parked car.

For finding things at locations and locations of things, apps rule. But for longer car trips, I frequently want an accurate estimate of my arrival time to more effectively plan on the other end, such as setting up a meeting time.

One thing that irks me is that my car GPS always provides a significantly different ETA than my smartphones. Earlier this week I had the opportunity to test once and for all which is most accurate and why. 

I had to take a business trip from Portsmouth, NH to New York City for some interviews and business meetings, so I decided to monitor the entire trip of roughly 260 miles and measure the accuracy of all the devices. 

I entered the same address into my car GPS (a Garmin Nuvi), Google Navigation on my Samsung Galaxy Nexus on Verizon and MapQuest on my iPhone 4S on AT&T. 

All three have turn-by-turn directions with audio, so I turned on all three and started the trip at 12 p.m.

All three devices suggested essentially the same route to New York -- but the projected arrival times, as usual, were not in sync.

GPS said I would arrive at 4:03 p.m., Android at 4:43 p.m. and iPhone at 4:54 p.m. Viewed another way, Garmin was the optimist and iPhone was the pessimist. I was looking for the realist.

I logged the continually changing projected arrival times every 15 minutes for the entire trip. Aside from one five-minute rest stop, the only variables were traffic and my driving speed, in general with the flow of traffic.

When approaching one particular suggested exit, all three devices recommended taking it despite a visible two-mile backup entering. I overruled all three, which all instantly readjusted, then continued to the next designated exit and lost no projected time at that point.

The GPS remained relatively constant until it added 10 minutes during a traffic jam at 1:30 p.m. and another four minutes by 1:45 p.m., although it deducted those four minutes by 2:45 p.m., projecting arrival by then at 4:11 p.m. Still optimistic -- but inching closer to reality.

While GPS was adding minutes to its projection, iPhone was deducting them. By 1:15 p.m., iPhone had deducted 17 minutes from its initial arrival time projection but added another 11 minutes back to that during the same traffic jam the GPS noted. 

Android was quick to add six minutes initially, and then deduct 12 in the next quarter, before becoming somewhat stable for the rest of the trip. Fifteen minutes before we hit the traffic jam, during which GPS and iPhone added 10 minutes, Android added six. During the traffic jam itself, when the other two devices added 10 minutes, Android decreased them by one.

The traffic jam had apparently already been factored in by Android before we hit it.

For the entire trip, Android was closer than iPhone to arrival time at virtually every 15-minute interval. By 3:45 p.m., Android accurately projected my exact arrival time, while the other two never did. 

Arrival at my New York destination was exactly 4:19 p.m. Here are the final results: 

GPS projected 4:03 p.m., so it was off by 16 minutes, meaning that I arrived 16 minutes later than projected. I was destined to be late from the start with no hope of catching up.

iPhone projected arrival at 4:54 p.m., so it was off by 35 minutes -- meaning that I arrived 35 minutes earlier than initially projected.

Android projected arrival at 4:43 p.m., so it was off by 24 minutes, meaning I arrived 24 minutes earlier than projected.

Some observations:

• Viewing heavy traffic ahead on Android Navigation provided options to seek alternate routes before reaching the actual traffic.
• Leaving turn-by-turn on requires very heavy battery use, which can’t be done for any length of time without being connected to power.
• The Nexus battery gets quite warm during the process.
• Having turn-by-turn directions being spoken by three separate female system voices is like having three backseat drivers all issuing commands and suggestions you can follow or ignore with no real in-car consequences.
• GPS is continually an optimist to arrival time, appearing to base your speed on traveling the maximum speed limit, and then some, leaving little leeway to ‘beat the system.’
• Android and iPhone seem more realistic in speed limit estimates, allowing somewhat of a cushion within a lengthy trip and accounting for traffic along the way.

On the return trip, I conducted the same experiment, noting projected arrival time every 15 minutes. Although I did stop longer on the trip back, I arrived 84 minutes later than GPS projected and 13 minutes earlier than both Android and iPhone projected, which were identical at the start.

Just leaving New York, I called home to my human GPS Teri and mentioned that I was going to run this experiment.

“Why bother -- you’re going to be home at about 9:30?” she said without hesitation or any technological assist. Her estimate turned out to be the most accurate to within just a few minutes.