[Talk-GB] New Post Office Data and Comparison Tool

Robert Whittaker (OSM lists) robert.whittaker+osm at gmail.com
Mon Feb 19 13:29:28 UTC 2018

Some of you may have already seen that a few weeks ago I eventually
got a positive response from Post Office Ltd. (POL) to a request I
made for a re-usable list of their branches, and permission to use it
to help improve OpenStreetMap:

The raw branch list data can be found at
http://robert.mathmos.net/osm/postoffice/data/ and it licensed under
the Open Government Licence v3. It includes ID numbers, branch names,
addresses, locations, and opening hours.

I've adapted one of my previous comparison tools to compare the
dataset to what is currently in OSM, and the results can be seen at
http://robert.mathmos.net/osm/postoffice/progress/ . I was hoping we'd
be able to use some form of automated matching and import, but I don't
think the data is quite good enough for that. In particular, the
locations aren't always accurate and sometimes appear not to have been
updated following a branch move. (Based on a small sample that I've
matched so far, it looks like 10-15% may be out by more than 100m.)
Also the address data isn't easy to automatically parse into the OSM

For those who are interested in using the tool to help improve Post
Office coverage, I've put together some mapping notes at
http://robert.mathmos.net/osm/postoffice/mapping-notes.html . You'll
find various JOSM Remote Control links on the individual branch pages
(e.g. http://robert.mathmos.net/osm/postoffice/branch/19669 ) but
please use these with care. If you want a source tag to add to
individual objects or the changeset, then I've been using

To help match the dataset to OSM, I'd suggest adding the branch ID
number to the appropriate OSM objects. What has been supplied is the
primary key used in the Post Office Ltd database, which isn't the same
as the FAD code for the branch (which POL want to try to keep
confidential for some reason). Following other similar third-party
keys, my suggestion would be to use ref:pol_id=* for this number.
That's currently what the tool looks for, but if there's a consensus
to use a different key name, it would be easy enough to change it.

There are two other tagging issues it might be useful to discuss:

1/ When you have a post office that is a counter in a larger shop,
should you add amenity=post_office to the shop object, or create an
additional node specifically for the Post Office? I would usually do
the latter for counters in larger shops, but may to the former for
smaller village convenience stores where the Post Office part is more
central to the operation. (The tool will pick up either, and will cope
with opening_hours:post_office=* if the hours of the PO counter differ
from the main shop.)

2/ How to tag things that are functionally like post offices (e.g.
they provide mail sending / receiving services), but aren't Post
Office Ltd branches. This includes Royal Mail / Parcelforce public
offices, and private courier forms with high-street offices (e.g.
Doddle, Cnuk, Mail Boxes Etc). If these types of businesses are tagged
as amenity=post_office (which doesn't seem inappropriate given the
wiki definition) then it would be useful to have some way in the
tagging to be able to automatically distinguish them from Post Office
Ltd branches.


Robert Whittaker

More information about the Talk-GB mailing list