Because, despite Google’s mapping and data crunching prowess, it wasn’t designed to get the ETA right for logistics.
Supply chain logistics management isn’t just about finding the shortest route between point A and point B anymore.
Fleet usage, fuel consumption, and meeting delivery timetables are important KPIs, and they’re getting more aggressive.
Logistics departments have a hard time planning their pickup or delivery timelines to hit tighter windows of delivery. It’s getting tougher, because there are too many parameters to consider. Hundreds of orders, dozens of destinations spread out far and wide, and hitting different windows of delivery using as few vehicles and drivers as possible.
Toss in things like changing traffic and road conditions, multiple pickups or deliveries, varying loading/unloading times, warehouse space availability, or congestion at a sorting facility, and you’ve got a mix that’s hard to get on top of every day.
Predicting ETAs has never been harder — or more vital — than ever before.
How Does Google Calculate Your ETA?
Google Maps ETAs are calculated based on a variety of things.
Google gathers and crunches massive amounts of data to determine an ETA when you look up directions in the Google Maps app. In a nutshell, Google relies on:
- Crowdsourcing — Constantly gathering anonymous location data from millions of mobile phones across the world to track movements, giving it a treasure trove of past movement patterns using GPS and cellular triangulation data. That gives Google an idea of congestion (on a road, or at a traffic signal for example) and allows it to determine average speed to and from those points, either during the entire day or at a specific time.
- Data aggregation & analyzing past patterns for future predictions — When it comes to anything involving crunching large amounts of data with numerous variables, few are as good at it as Google. Google estimates ETAs using a variety of things, depending on what data is available for a particular area - city plans, proposed road maintenance or diversions, new signals going up, urbanization plans or anything that could cause more congestion. Google draws data from whatever sources it can to come up with the best predictions that Google Maps can make.
Estimating ETA based on past performance isn’t easy.
It depends on the data available for an area, which includes factors like official speed limits, the likely speed a vehicle can hit depending on road or traffic conditions, and actual travel time logged from previous trips, all of which gets averaged out and sorted based on time of day or weather conditions to give you the best guesstimate.
And a guesstimate is the best that Google can give you.
It’s not that far off the mark mind you; Google Maps’ ETA guesstimates were good enough for Uber when the taxi company started out. But Uber is moving away from Google Maps to calculate trip ETAs.
Why Google Maps ETA Won't Work for Logistics Planning
Transit time on Google Maps isn’t always accurate.
When Uber launched in NYC, it used Google services to predict travel times. Unfortunately, they found that Google’s ETA predictions were “on average, off by 3.6x the actual pickup time” during the first week. Uber’s engineering team thinks Google Maps APIs are “great for almost everything else, such as geocoding street addresses into latitude/longitudes and then back again. In other words, we love them for everything except for accurate ETAs.”*
While Google’s ETA projections work well as a general estimate of arrival time, the accuracy isn’t great, which is probably why Uber is investing $500 million in a global mapping project to develop its own system to get more accurate transit and arrival times.
It’s not that Google’s algorithms don’t work well, it’s just that they’re not designed with supply chains and logistics in mind.
Google needs to know a lot more about things like live traffic patterns, disruptions and enough historic data to make accurate current estimates. That’s a tall order, especially in parts of the world where detailed maps, street signs, or even proper roads aren’t always available.
Here are some other reasons why Google Maps’ ETAs aren’t always the best bet, especially for supply chain logistics management, route optimization, or delivery planning.
- Location Accuracy May Not Always Be Bulletproof
Google relies on crowdsourced location data, from which it derives information on traffic movement, congestion, and transit time. The data is drawn in anonymously from mobile devices that are tracked using cellular triangulation or the built-in GPS hardware that’s commonplace in most modern mobile phones. The accuracy of the data depends on the quality of the signal or hardware that generates it, which means it isn’t always accurate. Although Google does have the ability to scrub or filter out bad data, it could throw Google Map’s ETA algorithms off by hours if not implemented properly. - Google Maps’ ETA Isn’t Accurate for Long Distances
The transit time estimate that Google Maps gives you at the start of a trip is based on the road and traffic conditions along the route at that time. That’s usually reasonably accurate for shorter distances. Most shipments aren’t short distance however, and the longer you’re on the road, the more those conditions change. Google doesn’t accurately factor in traffic signals, stops along the way, and unplanned route deviations over longer distances. The lack of real-time calculation and updates mean there’s no dynamic route optimization, which affects overall accuracy. - Map Apps Don't Take Physics Into Account
The type of vehicle and its load determines a shipment’s its average speed. Navigation apps like Google maps weren’t designed to account for things like vehicle size, engine displacement, or factor in its average speed depending on the load aboard. - Pit Stops and Other Constraining Factors Aren’t Accounted For
Google doesn’t factor in the time it takes to stop for fuel, rest, halts at a toll-house, warehouse or sorting facility, the time needed to load/unload goods, or other constraints like limited night-time driving. - Google Maps’ ETA Doesn't Work for Multimodal Shipments
Google ETA doesn’t take train schedules, air congestion, queues to load/unload containers at ports, or the time it takes to inspect cargo into account. - ETAs Aren’t Accurate On the Road Less Traveled
Most long-distance surface transport happens over remote highways, where pedestrian or other vehicular traffic is usually at a bare minimum. Since Google relies on anonymous crowdsourced transit time data to make more accurate estimates, Maps has no way to gauge accurate ETAs for less traveled routes, especially in remote regions, and more so if they have bad cellular coverage or aren’t well mapped. There simply isn’t enough data on which to base ETA estimates on. - Google Maps’ Doesn't Take YOUR History into Account
Crowdsourced data isn't relevant to supply chain logistics management. Shipping works on different milestones, and your transit time depends on your own routes, past performance, and current transport corridors. Google doesn’t have that information, which makes it unlikely that Google can give you an ETA tailored to your specific circumstances. Even if Google starts tracking cargo movement and develops algorithms tailored to supply chain movements, it’d be a few years before they’ve gathered enough enterprise data to give us a reasonably accurate ETA for long-haul cargo or shipment deliveries.
So What Do You Need?
Google’s Maps app wasn’t meant (or designed) for businesses to run supply chain operations, it’s more tailored to consumers that need help with directions or picking the fastest route to where they’re going.
Even if you had to rely on Maps for fleet route optimization in a pinch, the only way it would be more effective is to make sure the driver or dispatch personnel keep a constant eye on current conditions, making corrections or modifying the route to compensate for unexpected delays.
If you’re running large-scale operations, the need for human intervention means the setup isn’t scalable without some serious investment in human capital; not exactly an ideal solution.
Logistics companies that measure their performance and compare their predictions against real-world outcomes stand a better chance of predicting more accurate ETAs as compared to Google Maps.
It’s not something that’s possible overnight though, and even the best data-driven predictions need time and plenty of fine-tuning before they’re accurate enough to run a Just-in-time (JIT) delivery system.
Even if you had a good idea of current traffic conditions and disruptions along the route — roadworks, derailments, or inclement weather conditions for example — nothing can predict unexpected incidents like a vehicle crash, emergency services cordoning off a highway, or a systemic failure at a warehouse or sorting facility.
To cope with both the expected and unexpected variables that affect shipment ETAs and the likelihood of making an on-time delivery, you need automated shipment monitoring systems that give you live visibility into how operations are running, where they’re slowing down, and what could spiral into something serious.
You need more data, gathered and processed in real time, to give you the information — and insights — to make critical decisions that affect the outcome before it’s too late to act.
Want to make more accurate ETA predictions?
Start by monitoring and tracking shipments as they happen.
