I came to write exactly this comment, specifically I wanted to check what the pricing for forward geocoding was, as I've got a very significant monthly Google Maps bill for geocoding and if Radar gives results which are 95% as good as Google's at a cheaper price then I'd be jumping
ESRI is as evil as Google (if not worse). If pricing is the only consideration, maybe ESRI is the way to go, but I would take other aspects into account, too (e.g. walled-garden signals, tricking customers, software-lock-in).
I believe the point is that Op doesn't _want_ to chat. I believe the originally point is you should be up front and _clear_ with your pricing rather than trying to force a conversation.
Yep, I get it. Working on a self-serve pricing calculator. Reality is that enterprise convos do make sense at sufficient scale. If it does here, our inbox is open!
Often, if a company has "Contact Sales" on their pricing page it's because they only want customers who have a budget big enough to warrant contacting sales.
At an early stage, it's often easier and more lucrative to build for and support a few large customers than many small ones.
As someone who used to work at an ad tech company, I can tell you that we're very much not an ad tech company. In fact, we set out to build the opposite. More here: https://radar.com/blog/our-commitment-to-privacy
Short, easy-to-spell name that suggests location. It's worked out pretty well for us, and we got the dot com!
What's the terms on your forward geocoding? Many providers have terms that require live lookup of address->latlong, no caching permitted. My client's use case has long-lived addresses that may need to be mapped many times, and currently we're paying for one of the only geocoding services that permits us to persist the coordinates involved.
> Licensee shall not … store any address or point of interest data from Radar's geocoding, autocomplete, or place search APIs for more than thirty (30) days;
So if an e-commerce site user starts typing their address, it’s autocompleted via Radar, and the user confirms it, then the site must delete the user’s address after 30 days?
Not sure if I missed it, but how do you calculate address completion request? I mean, Google let me pass a random key for every Word I type and charge me once. Do you have something similar? Or each letter counts as 1 request?
As an OpenStreetMap contributor I really appreciate the proper attribution you have in the bottom right, rather than hiding it behind a button like Mapbox and some others do - thanks!
There’s no attribution that I can see on the second embedded map on the page (on mobile). A honking great Radar logo, but no OSM attribution.
This trend of “our branding takes priority over the required attribution for the free map data we’re using, but hey, they’re a little nonprofit so can’t afford to sue us” really ticks me off. Mapbox started it as a calculated move, and since then others have followed claiming it’s “the standard”.
Their geofencing API seems to advertise tracking users. It seems to provide some kind of metadata storage alongside a geofencing API. My guess is that some websites are tracking their users though this API.
I haven't dug into it much further, but if their API is used for tracking, it should stay on the list.
Edit: looks like they provide a dashboard to follow the location history of any user using their API. Yeah, no, I'm adding this to my personal blacklist just in case.
The example map of New York on the linked page has rather odd labeled places and street names. At the default zoom, it’s marked up as if New York, Hoboken, Weehawken, etc are literal points marked by tiny circles. This is particularly silly in the New York area, which is a continuous gridded metropolitan area.
Then, when zooming in, you have to zoom very very far in before street names show up. This makes a map zoomed slightly less far in mostly useless, because you can’t tell where you are. I realize that people mostly don’t navigate by map any more, but if you are literally selling map tiles, presumably mostly not tied to the end user’s GPS, then whoever is looking at the map would like to know where they’re looking. For example, in Manhattan, the actual useful coordinates are the cross-streets, and the map is not very helpful without the street names.
Google is the gold standard, but our coverage is quite good in the US. We incorporate a mix of open and commercial datasets both for POI and address data.
More details in the post, but Google starts at $7 per 1K map loads and $5 per 1K geocoding API requests, whereas Radar starts at $0.50 per 1K for both. We've roughly been helping our enterprise customers cut their maps bill in half.
I was actually looking into options for maps, geocoding and routing for a freelance thing that I'm doing and this actually seems great! I might go with Mapbox for the already started project, but will probably use Radar for at least something in the future.
The other options I considered were Mapbox and MapTiler, since it seems that actually self-hosting everything yourself is a bit demanding on hardware resources. I mean, I love that you can get the OSM dataset into PostGIS pretty easily, but getting the vector data or raster tiles to the client can take a bit of CPU power and bunches of storage! I wonder whether I missed any other good cloud services, or a self-hosted stack that would make things easier.
That said, I wonder why so many seem to be moving away from raster tiles to vector data. I guess you can cache both and vector data is actually less storage intensive, but at the same time pretty much every vector map implementation lags badly on my phone and sometimes even a bit on my netbook (Radar still lags for me, although less than Mapbox, probably because Mapbox does the whole fancy 3D globe thing with fog and stuff).
Either way, major props for making this exist and pricing it so competitively! Would be cool to also fix the attribution text that's on the lower right on desktop but doesn't show up on mobile, at least for me.
For a self-hosted vector tile stack you can have a look into https://github.com/onthegomap/planetiler I found it very easy to get started and when you know the other stacks it is also very fast to create these vector tiles even for planet-scale.
(note, that I'm not affiliated with them, but they use some source code from us for the efficient import and also contributed to GraphHopper, but this did not influence my experience ;) )
> I wonder why so many seem to be moving away from raster tiles to vector data.
The flexibility of styling. And you can easily serve customers that need different default languages. This makes maps also more accessible for countries without Latin alphabet. Also when you rotate the map (like for GPS navigation) the labels can rotate too if necessary.
This looks cool, but the pricing isn't exactly clear. Website says $0.50/1K API calls and then the HN title says $0.50/1K map loads. Is a map load === 1 API call?
A map load is what it sounds like, an API call is a geocoding/search/routing API call. Pricing starts at $0.50 per 1K for both. Pricing page could be clearer, we're working on it!
I used to work for a company that made an app for drafting evacuation signs, and we had a usecase for using a map as an underlay to show the surrounding buildings/terrain for the siteplan. I recall that quite a few map providers had a TOS that prevented that usecase (as the evac signs would be saved to pdf for distribution or printed), so I'm interested what the take is here.
A static maps api that produces an image of the relevant vector tiles is on the near term roadmap (next 1-2 months). Can you say more about the geojson drawing part?
I'm interested in the address autocompletion in particular, but you guys don't want to tell me how much it's actually going to cost apparently.
I also have no idea what a "monthly tracked user" is or how it's relevant.
LMK when you drop the "Contact Sales" aspect.