Changing telematics providers is mostly a scheduling problem. Changing an ELD provider is a compliance problem wearing a scheduling problem’s clothes, and the difference shows up in the order you do things rather than in the work itself.
This is that order. It applies whether you are moving to another commercial platform, to something you host yourself, or to a combination — and most of it is about what you do before you tell anybody you are leaving.
Export first. Send notice second.
Under 49 CFR 395.8(k)(1) a motor carrier keeps records of duty status and their supporting documents for six months, plus a backup copy on a separate device. Read that sentence again with the emphasis where it belongs: the obligation is the carrier’s. Not the provider’s.
Your ELD vendor holds those records as a convenience. When the contract ends, the convenience ends, and the obligation does not. Some platforms deactivate an account the day the subscription lapses; the records are still legally yours to produce, and you no longer have a way to reach them.
So the first task, before the sales conversation and long before the cancellation email, is to pull six months of RODS for every driver into somewhere you control. Then open one of the files. A download that turns out to be empty, truncated or unreadable is a normal outcome and you want to discover it while you still have an account.
Motive publishes a public API covering vehicles, users, HOS logs, locations, IFTA, inspections, fault codes and driver performance, so this can be a scripted pull rather than a week of clicking. Two cautions. The documentation does not state how far back the history endpoints reach or what the rate limits are — ask, in writing, before you plan around it. And API access is a commercial arrangement rather than a property right. Motive recently cut the rate at which Highway could reach its customers’ ELD data, and restored it days later once the two settled — a disruption and a resolution inside the same week. Nobody’s data was lost and nobody behaved badly. It is simply a reminder that the pipe between two companies stays open by agreement, which is the reason the export belongs at the start of the project rather than at the end.
What to take, in priority order
RODS and supporting documents, six months. The legal minimum, per driver, and the one thing that has a regulator attached to it.
Unassigned driving events and their annotations. These are the ones an audit asks about, and they carry the explanation somebody typed at the time. A CSV of drive segments without the annotations answers the wrong question.
GPS history. Not required by the ELD rule, and the thing you will miss most in practice. Every route reconstruction, every “where was this truck on the 14th”, every customer dispute about a delivery window lives here. It is also the largest export by far, so start it early.
IFTA mileage by jurisdiction. Rebuilding a quarter’s state-by-state mileage from raw positions after the fact is possible and miserable. Export the computed figures.
Inspections, defects and fault codes. DVIR history is a maintenance record and sometimes a liability record.
The configuration you forgot you built. Geofences, groups, alert rules, working hours, exempt drivers, personal-conveyance settings. None of it is regulated, all of it took somebody a year of small adjustments, and nobody remembers it exists until it is gone.
The three dates that actually bite
The seven-day rolling record. A driver at a roadside inspection must produce the previous seven days. On cutover day those seven days live in the system you just left. Either print each driver a summary for the week before the switch and have them carry it, or key the missing days into the new system. This is the single most common way a clean migration produces a violation, and it is entirely avoidable with a stack of paper.
The 48-hour production standard. During an audit spanning both platforms you produce records from both, within 48 business hours, whether or not you still have an account with the first one. This is the reason the export is a file you hold rather than a login you hope still works.
Sixty days, if you are moving because of a revocation. When the FMCSA removes a device from the registered list, carriers get 60 days to replace it. That is a different project with a different deadline, and it does not leave room for a leisurely pilot.
The hardware is the bill
The Motive Vehicle Gateway is their device, running against their platform. It is not a neutral sensor you repoint at a new server, and the honest planning assumption is that leaving means replacing it in every vehicle.
Which makes one question worth asking every incoming vendor before you sign anything: which devices already in my fleet do you support? Platforms that accept third-party hardware — Teltonika, Queclink, Ruptela and the rest of the common protocols — change the arithmetic completely, because the fleet you are quoting is the fleet you already own. Platforms tied to their own gateway are quoting you a second capital purchase whether or not the quote says so.
Then count the labour honestly. Fitting a device is fifteen minutes; getting the truck to the fitter is the other four hours, and it happens once per vehicle, at a depot, on a day the vehicle was supposed to be earning. Fleet-wide hardware replacement is the most expensive and most disruptive part of any provider migration, and it is the part that slips.
Not every platform is built this way, which is worth knowing when you are choosing what to move to rather than only what to move from. Geotab’s GO devices can report into software other than MyGeotab, so that line can be zero. Samsara’s cannot, and its cameras make the number worse.
Run five trucks first
Pick five. Not one — one vehicle cannot show you a pattern — and not the whole fleet.
Fit them, run both systems in parallel for a few weeks, and compare the numbers that people actually argue about: distance per day, engine hours, fuel where you have a sensor, and the start and end of a couple of specific trips. If the two disagree, find out why now, on five vehicles, rather than in month two with three hundred.
Use the pilot to answer the questions a demo cannot. Does the driver app behave on the phones your drivers actually have. What happens when a device loses signal for an hour in a yard. What does an unassigned drive segment look like, and how much backend cleanup does one create. Whether the automatic drive trigger — five miles per hour, on every ELD, by rule — produces events your dispatchers can live with.
And time the cutover like an operations decision, because that is what it is: not in your peak week, not the week of a scheduled compliance review, not the Friday before a long weekend.
What does not come across, ever
Some things are not portable and it is better to know before you plan around them.
Your driver behaviour scores. Every platform weighs harsh braking, speeding and idling differently, so scores restart. Communicate that to drivers ahead of time or your safest driver will spend a month asking why the new system thinks less of them.
Video, if you have cameras. Footage is tied to the vendor’s storage and retention. Anything you might need for a claim or a dispute has to come out as files before the account closes.
Machine learning that had learned your fleet. Idle baselines, route familiarity, anything that improved over months of your data. It starts over, and the first weeks will produce more false positives than you are used to.
Frequently asked
Can we run both platforms in parallel for a while?
Yes, and you should — that is what the pilot is. The constraint is that a vehicle’s ELD of record must be unambiguous for the drivers using it. Parallel running is per vehicle, not per fleet, and every driver needs to know which system is theirs today.
Do we need to notify the FMCSA?
Not of the switch itself. What matters is that the device you use is on the registered list and that your records are producible on demand across the transition.
What about the notice period?
Read the contract before the export, not after. Notice periods, early termination fees and the window during which data stays accessible after cancellation are all in there, and the last one determines your actual deadline for the export.
Can we do this without replacing hardware?
Only if the incoming platform supports the devices you own — which for a proprietary gateway usually means no. Ask the question explicitly and get the supported-device list in writing. It is the largest single number in the migration.
Where this leaves Routly, honestly
If your fleet is under the US ELD mandate, this article does not end with us. Routly is not a registered ELD and does not claim to be; a mandated fleet needs a certified device from the FMCSA list, and choosing one is a compliance decision rather than a software preference.
What Routly is for is the layer underneath: the telemetry, the trips, the fuel, the geofences and the reports, on hardware you already own — Teltonika, Queclink, Ruptela, Concox and the rest — with the position history in a database you control rather than a vendor’s. Plenty of carriers run a certified ELD for hours of service and something else for everything else, precisely because the two have different reasons to exist and very different data-retention stories.
If that is your situation, the live demo runs on real telemetry and takes no sign-up, and the self-hosting guide is four processes and a compose file. If it is not — if what you need is a certified ELD and nothing else — the export checklist above still applies, and it is the part worth doing right whoever you go to next.
