One key metric missing (unless I’m blind) is time to acceptance scan. I do a lot of shipping and don’t have time to have the packages scanned in every day so I use the drop box. Last few months it’s taken the post office an extra day or two to even scan them in. So while the packages must only be on average 5-10% late they’re also being delayed a day or more just sitting waiting to be accepted.
It hurt my eBay rating too before I could catch it happening. I’ve resorted to using the scans form just to make sure they’re the ones taking blame, not me.
This feels like a missed opportunity. There's a lot of data Shippo could share about USPS and that would help add more context to what's happening in the news. But it's difficult to understand anything concrete from this. It would be more helpful to see how this fits into long term trends and other context on how abnormal it is.
Cool to have some “observability” on this, but it’s not the most relevant metric. I’ve heard of (and experienced myself), much, much longer delays on envelope/letter mail, but have heard of fewer delays with packages.
Perhaps this is a proxy for measuring personnel/process changes, but totally misses the sorting machine dismantling going on throughout the nation.
Thank you for putting this together. Independent data sources are vital, since we know this administration does everything they can to hide reality.
It’s too bad some of these numbers are reported as averages. It would be nice to see something more percentile based, like p75 or p90, which are more effective ways of measuring quality of service
I'd like to see this data grouped by the population density of the origin zip code.
I think this marketing site would perform better if it had a pretty version of the most interesting graph above the fold.
I think shippers like Amazon pay relatively high rates to USPS, so they are higher on the priority list.
It's the heavily-discounted "slow" shipping like medicine that is having trouble right now.
Does anybody have this data about letters? We need a canary letter mailing service for analytics.
Does USPS publish transit metrics directly? If not, they should.
Most election mail will be sent zone 0 (Local) or 1.
This analysis only covers USPS package delivery, which is different from flats. Ballots are flats. Magazines are flats. Letters are flats. Anything non-bendable isn't.
This is what has been happening to USPS, folks:
1. Because of COVID19, flats volume has collapsed, while package volume has skyrocketed. Flats sorting machines can do absolutely nothing for packages for reasons that are left up to the reader, so USPS has been shutting them down and moving them out of processing facilities in favor of package sorting.
2. Because USPS is losing lots of money overall (higher package volume doesn't make up for the collapse in flats volume)[1], it has been cutting back on overtime, just like any other employer would.
3. People hear about 1 and 2, hear about/experience packages being delivered more slowly, and think that this surely means that "the Trump administration is trying to sabotage the post office to suppress voting!!!!". They do this without thinking about it at all:
3a. As stated, flats volume has collapsed, so there is still a lot of excess capacity.
3b. Even if every single voter were to vote by mail only, this would mean at most two additional flat pieces per voter (one ballot to the voter, and one ballot sent back). Think about how much mail (not packages, mail) you already receive daily on average. Do you really think two additional pieces would collapse the system? Of course not, any more than the USPS collapses every January when the IRS and every single employer, bank, and other financial institution sends out tax-related documents. (The USPS hires seasonal help in December for packages, not for Christmas cards.)
3c. If this really were a sinister Trump administration voter-suppression scheme, it's a pretty weak one that can be defeated by dropping ballots off in person, and/or voting in person.
4. An actual serious issue is states and counties that aren't like Oregon (which has been 100% vote by mail for two decades) trying to convert to vote by mail without preparation. Think of how much mail your home receives for the previous tenant (and the one before that, and the one before that). Think of this all having to be done by early October, to give voters about a month to receive and return ballots. This is what the administration has been pointing out, something rarely heard amidst the nonsense about mail-vote suppression.
[1] Congress mandating the USPS to prepay pensions is a good thing. The postal service is an industry that is, by definition, in secular decline (barring unusual events like COVID19) because of the Internet. Congress recognized this in 2006 and thus required USPS to prepare over 10 years to get its pensions ready, because there's no reason to believe that future revenue (and future employee-count growth) is going to sustain pensions for retirees otherwise.
TL;DR: DeJoy's changes appear to have made USPS slightly slower (0.1-0.5 days) in some places, but much less than the delta caused by a holiday like July 4th (see note buried at the very end).
One thing this doesn't account for is fraudulent delivery reports. In my area a large percentage of my USPS deliveries are reported as delivered around 6pm on the scheduled delivery day, but I can't find the package anywhere. Then it suddenly turns up the next day. People on NextDoor also report seeing this regularly.