Most riders determine "real" speed by using a GPS, though there have been studies which claim this isn't 100% accurate either (see this paper).
When stating speeds, many riders will use "indicated speed" (what the speedometer reads) or "actual speed" (measured by GPS).
There's really no solution for this that anyone has found; most riders just accept it and ride "fast" or calculate approximate speeds in their head on the fly. In reality, most riders don't even bother to look at the speedometer. Just keep up with traffic so you don't get run over and all should be well. Scooterists are rarely ticketed for speeding in traffic. Unless you're flying by surrounding vehicles, riding recklessly, or obviously hauling ass on the open road, most police officers won't even bother paying attention to you on your scooter. Though we don't advise trying this, we've read stories of motorcycle or scooter riders beating tickets by claiming their speedometers are too inaccurate to be trusted and that they were just riding with traffic.
Speedometer accuracy is one of the most discussed and contentious topics on forums for all types of scooters and motorcycles (Google it). While we don't have all the information, we may be able to clear some of the hot air and make some sense of it all.
After exhaustive searches of online sources, here's what we've been able to verify as well as some commonly stated info that's currently unverified. Anyone with additional information is welcome to provide it, but please include some type of citation or reference.
As far as a US federal standard for speedometer accuracy, I've found many mentions of it, but only one actual citation in federal regs that doesn't seem to apply to motorcycles, just commercial vehicles and trucks.
Many sites have references to a quote from a long out of print article ("Auto Tutor", American Automobile Association of California magazine, Oct. 17, 1997) that states: "As of 1997, Federal standards in the United States allowed a maximum 5% error on speedometer readings..." This is completely unverified. We're looking for the source material and will see if we can track this down. Searches of current federal codes and standards have not found any such regulation.
What we have found specific citations for, though, are EU and ECE regulations, whhich are generally accepted around the world:
The Road Vehicles (Construction and Use) Regulations 1986, as amended, allows the use of speedometers that meet the requirements of EC Community Directive 75/443(97/39) or ECE Regulations 39. Both the EC Directive and the ECE Regulation lay down accuracy requirements to be applied at the time of vehicle approval for speedometers. Simply stated, these requirements are that the indicated speed must not less than true speed, nor more than 110 per cent of the true speed plus 4km/h (2.5 mph).
The above testing to EU requirements is done on a new vehicle, under clearly specified conditions (specific speeds, ambient temperature, tire pressure and unladen weight, etc) and typically on a roller dynamometer or a prepared track using a calibrated speed measuring instrument accurate to +/- 0.5%. Thus, trying to replicate the manufacturer's results on your scooter, using a GPS, is hardly likely to get the same results.
This explains why the inaccuracy is permitted, and why many vehicles seem to be as much as 15% optimistic when compared to a GPS. This deviation may be due to some of the variables mentioned above.
The salient part (bold added for emphasis) from the page following the link above:
It's unlikely that this is the source of all speedometer error (again, there are a number of additional factors), but it helps explain why precise readings are unlikely.
Here's a solution MV member jk_single employed for his BV 500:
BV Speedometer makeover/recalibration
Here's one obsessed man's DIY solution:
(It's for a BMW motorcycle, but the same principle may apply to scooters. This method of calibration doesn't account for lean and curvature so is still imprecise.)
Easiest DIY solution for those who are determined to have an accurate reading:
Get your hands on a GPS. Go for a ride and take numerous speedometer vs. GPS readings and different speeds. If you're really anal (instead of just kind of anal), try it at different times of day, in different settings, at different temperatures and on various road surfaces. Create a table with your findings. Average the discrepancies in readings at various speeds.
Create an overlay for your speedometer with average "actual" speeds for each 10 or 5mph marker. This can be anything from a custom die-cut vinyl decal to go inside your headset on top of the stock speedometer to a few pieces of masking tape scrawled on with a Sharpie.
- Speedo GPS (Actual)
And here's a chart to visualize the error:
- Speedo GPS
Here's a chart to visualize the error: