Why do you use ArcGIS for Desktop?

Question

I'm new to GIS, having only dabbled with ArcGIS for Desktop for about 4 weeks in 2007. The business I am at wants to target writing complete solutions that involve GIS as it would benefit their hardware sales. They do not know which market they want to target but want me to pick the GIS software to use, yet I am very green to GIS markets. All i know if target platforms are windows, windows mobile, android and web applications. The idea of mobile applications means data connection is not 100% so internet based services may need to be ruled out, i.e. GoogleMaps.

I was struggling to work out how to phrase this as a question, given the huge range of GIS products available, Esri's ArcGIS for Desktop, CadCorp, MapInfo, GoogleMap/Earth, qGIS, Manifold, openJump, openstreetmap etc.

To avoid the what do I buy question, I instead ask the question why do you use Esri ArcGIS. From my experience its slow to navigate around maps and quite verbose to program in causing quite a lead time in development. Esri is also the most expensive product on the market, and i believe also the most popular. But why do people continue to use it, when other systems like CadCorp have the same functionality if not more, seem to be more responsive, can import Esri data and cost less.

I feel that I must be overlooking something here. I know client demands in some industries like defense, want Esri but why is it still so popular, given I often hear how other products are betters (Is this a Windows vs. Mac example on market share vs. quality)

P.s. This question has been spawned from this one on StackOverflow.

Answer 1

First of all, easy to use. I have been using ArcGIS for almost 7 years now and I can tell you that people love it's simplicity and how to build simple maps and analysis.

I have been tracking and also using QGIS for almost 3 years now and as other have pointed, QGIS is getting closer on funcionality, but still "ESRI" is printed on peoples mind.

ArcGIS has a solid documentation, large use base and support for old products that most of other companies cannot provide. They have been investing in new technologies and spread of new concepts (which may be something or not - check out GeoDesign).

QGIS is a very strong candidate for a replacement of ArcGIS. There is one specific project, for a municipality here in Brazil (a large north capital) where we were able to stick with QGIS and they seem to be very happy with it. Customizing it is another point, but it doesn't seem hard - everyone just need to "get used to it".

One point in favor of ESRIs tool is its native cartographic output which is still the best, and looks like it will be for some years.

EDIT explaining the native cartographic output comment: Well, most of GIS softwares nowdays can export to a different format, such as PDF and SVG, to be edited outside the boundaries of GIS scope.

I've heard that many many people uses FOSS GIS software to generate the basic layouts of their maps, and then change to something like Inkscape and refine the look and feel of that map.

Natively, ArcGIS has the best cartographic control and output of all. It has extensive labelling features, symbol creation, it's has many features of a vector graphics software.

Thats what I meant as native cartographic output :D

Answer 2

Cartography is why we use Arcgis. I can and do use other products when warranted, ogr2ogr is up to 36x faster than arcgis converting shapefiles for example, which has earned it a permanent and dedicated part of my toolset. That doesn't matter though if I can't map the results.

All of our GIS work, even the analytical numbers only stuff, circles around the focal point of "how do we draw this? how do we communicate this visually?". That's where the rubber hits the road and the pudding is proven, to mix clichés horribly. If we can't make a map or some other visual product out of data that conveys meaning, it effectively doesn't exist and the millions of dollars poured into building the data is wasted.

There is no other product we've examined which comes even close to the breadth of options we have available in Arcmap. Back in the Arcinfo Workstation, ArcView3 and ArcPlot days we used Mapinfo for the final stage because of it's cartographic superiority. Adobe Illustrator could be better than Arcmap, there's as much distance between it and Arcmap as there is between Arcmap and __.

I have great hopes for QGIS and SAGA and bretheren -- I want to ply my craft outside of my day job as well, but they're not there yet for the kind of things I do.

update: A year ago I said "If Illustrator could read shapefiles or file geodatabases in situ there's a good chance we'd use that instead." That was before I started delving more seriously into Cartographic Representations. It is a really powerful toolset that, so far as I know, has no analog in any other application, in or out of GIS. In a nutshell, you can keep your decimal accurate geometry for analysis and move the map pushpin over a few hundred metres for legibility on your map while retaining a single datastore.

As we've watched Qgis continue to grow the question of whether it deserves a serious role in our operations is asked more frequently. The advent of Representations in our awareness has pushed those noises far into the background again.

Illustrator (and Inkscape for that matter) are still much, much easier to use than Arcmap for text and graphic handling, once you get to know them. I've used Inkscape for a couple of small map projects but it bogs down really quickly as the number of objects grows, which happens fast in our maps.

A big wish I have is for the usability of Illustrator/Inkscape with the data smarts of Arcgis/Qgis.

Answer 3

While I technically cannot actually answer this, since I don't use ESRI products, I can have a stab at the ". . . but why is it still so popular" question. (This is just my opinion based on discussions over the years with folks much more knowledgeable than me, so I can't really back it up with facts but it seems to make sense to me).

The major reason is market and mind share. "Most people" know ESRI software, and most people who've heard of GIS think of ESRI. That's just general wisdom in the atmosphere, because of huge numbers of companies and government departments and educational institutions that have deployed ESRI software en masse. Market share is one reason to choose something, but (depending on your business model) it can be far more important to assess the technical match of a product to your needs. (That's what I think you should be asking questions about btw).

I think many expect something as 'serious' as GIS to need to be huge, and so the perception is that it's expensive, so it's good, lots of people use it, it must be the best. Since it does cost a lot to invest in it you also won't hear many argue against it's cost-effectiveness no matter what they truly feel, since they are committed to a big investment (this is obviously contentious, and is not meant to antagonize people for whom ESRI really is the right choice).

That is changing with more cost-effective solutions from open source and more affordable products (GDAL and QGIS are very widely known these days, and there are many others), but it's not changing with dyed-in-the-wool ESRI users - partly since it's people who are brand new to GIS that are using the new tools, not the old-school - and also since the current users will be a much harder market to crack - that's just not effective for organizations wanting to invest in software that they want lots of new people to use.

Answer 4

I'm on the other side of the boat, where I started using Qgis in 2006 and switched to ArcGIS in 2009. Although there is "nothing" you can do in ESRI that you can't do in QGIS, the issue is time. If you're a consultant or an employee, with salary of $X per hour. Lean and efficiency dictates that you minimize your operation cost with an acceptable increase in capital cost (license purchase). Producing a model, outputting the results, and making the map look pretty damn sexy are achieved a lot easier using ESRI.

Second, a lot of ESRI users use their software on a lighter base such as mapbooks, printed maps, web maps, dynamic/web interactive maps. Except for the last two, there is no other software out there FOSS or not that can do that as pretty, efficient, and as fast as ESRI

Third, Map and model maintenance, ESRI excels in making maps that are easy to update with current data or pull out archived versions of the map that showcase older data.

Fourth, really it's not that expensive. If you think of TransCAD (transportation GIS) or EMME2/3 (another transportation GIS software) their costs run up to $50,000 per license, and most of the time you still need ArcGIS to finalize the job

Fifth, innovation. I think all FOSS and non FOSS software are playing catchup to ArcGIS, whether we like to admit it or not. Although some might argue that with FOSS you have better R and R-py support, or with GRASS you have better modeling, at the end of the day, ESRI's innovation extends beyond the models, and makes their software a lot easier to use and makes map production from collected data to printed map cycle a lot easier and prettier.

Sixth, I hate ArcGIS's horrible ui, but you get used to it, and I think that is one of their biggest downfalls

Answer 5

I have been in this space 21 years in the Petroleum Exploration Industry and have used all versions of ArcGIS software since the early 1990s both on the desktop and in the enterprise. I've also used most the products you mention in the past. Esri tools are tried, true, tested, and work better than anything else in the marketplace. Are they perfect? No. Is anything, no? Are they responsive to continuous improvement? Yes.

Because of the complexity of solving nuts and bolts mapping problems, I do not see Esri dominance changing. I have used many of the other products you mention, but when you get down to it they do not have the user base, broad functionality, or detailed technical and domain knowledge that you need to get your job done. I'm in Petroleum GIS - an entirely separate world of domain experience and technical know-how is required.

We love some open source solutions, particularly products like postgreSQL. I like that Esri plays well with it - makes us even happier. I mention it so that you can see I am not closed minded to open source solutions if they are good and focused. I go to FOSS conferences. It's great for what it does but they just do not have the financial resources of an Esri for GIS.

Finally, it is our customers' product of choice and we like to be compatible with them and their work flows. It would introduce a lot of hidden and not-so-hidden costs for us to move to anything else even if there were other options.

Answer 6

This is a good discussion topic. Here's my take. I learned to use GIS software in a university setting - a university that has a ELA with ESRI, like many do. This brings many users into GIS on ESRI software. Does ESRI have a large market share? Of course. Is it good software? In my opinion, yes. It is well supported, stable, constantly improved, scalable, inter-operable, and has a HUGE user community to go to for support and ideas. I think that most everyday users (and probably corporations for that matter) are going to go with a product that meets the criteria I described above simply because it's easier for them. The ArcGIS platform is tried, tested, and dependable in the long run.

I work for a medium-size corporation that of course can afford enterprise licensing of a product like ArcGIS. For that matter, ESRI licensing is chump change, as we have other software packages whose ELAs run into the millions of dollars. I also consult on the side, and have a personal license of ArcGIS - ArcView. Why did I spend $1500 on that you might ask? Because my clients all use it and I am in business to provide them solutions using the package that they are also using. Well, that and I honestly like ArcGIS.

Answer 7

From a development point of view, it's the most cost effective option. At $/hour, it's much cheaper to buy an ArcEngine license, with all the smarts that ArcObjects gives you, than start from a lesser base. It's fair to say that in most currencies, an ArcEngine license costs less than a single day of a developers time.

Answer 8

You use ESRI products usually because you have to. Almost every government organization uses them and you have no choice when employed by them. I love the ArcGIS suite of products, I have yet to see another GIS program that is as powerful, easy to use and used by so many. Still, ArcGIS sucks.

• Cost! - The number one reason ESRI sucks: the programs are prohibitively expensive, the extension's are a rip off and it makes the GIS division of any organization look horrible on the budget sheet. Non-GIS users can never understand why it cost's so much. It forces most to look for alternatives. In the end the prices for ArcGIS are to high and unless they lower them, people will bolt soon as QGIS really takes off.
• Functionality - When I first started really delving into the Analysis side of ArcGIS I found a lot of bugs, crashes and slowdowns. In the end I felt like I was sold a product that doesn't preform as well as it way hyped. ArcGIS is great for analysis, but when you spend half your day troubleshooting bugs in the program, restarting your computer and contacting support, that to me is unacceptable.
• Unfixed Errors - ArcGIS 10.0 cannot clip a shapefile that has over 500,000 vertices. It's a fact ESRI rekognizes it.Check out this link here. In the comments an ESRI staffer has something to say. Okay, their is a bug, programs have bugs. But, the fact that they aren't fixing it till 10.1 is like kicking you while your down. They have us a product that doesn't completely work. But, they fixed it, if you can afford to upgrade (or pay the service fee).
• LiDAR Data - If your going to be working with LiDAR, stay away from ArcMap. It can't even open a file without crashing and I have an great computer.

I still say ESRI is the best GIS platform out their but, look at your alternatives. I'd go with something else if I could.

Answer 9

There is currently no alternative, when you consider the ease of use and the range of tools available.

I think this will change in the next few years, with QGIS getting closer and closer, and FOSS. I can't wait until ESRI looses market share and sells a product for a few hundred dollars that is quick and robust.

Answer 10

1. ArcPy Python Site-package

   I accomplish more with the ArcPy site-package than with any other geospatial tools. The ability to combine the functionality of ArcGIS and the flexibility/extendability of Python seals the deal for me.

2. 64-bit background geoprocessing

   Prior to ArcGIS 10.1, I often had to get creative with memory expansive vector operations such as "Integrate". This all changed with 64-bit background geoprocessing.

3. in_memory Workspace

   in_memory workspace has helped me out time and time again by allowing speed and organizational improvement.

4. ModelBuilder

   No GIS has the capacity that ArcGIS's ModelBuilder does. The ability to create and share tools has changed the game.

5. ArcGIS Help

   I can't imagine how many times I have Googled "Con arcpy", "mosaic to new raster arcpy" or any of a number of ArcGIS tools knowing that it will bring me to the ArcGIS Help page on that topic. This is by far the best, most useful and most comprehensive online help I have encountered. Additionally, their pages are predictable such that, for example, I know the tool parameters are located toward the top of the page and the supported environments are toward the bottom.

6. Vector and Raster Functionality

   ArcGIS has pretty amazing vector and raster capabilities compared to a decade ago. Not long ago, digital image processing was solely within the realm of specialized software. Now most GIS techs can accomplish everything they need in this one software package.

7. I have been trained to use ArcGIS

   This is really a brilliant marketing move by ESRI; pretty much give away licenses at the University level to ensure geospatial people rely on these tools after graduation.

Answer 11

It's good for job security also. Been using ESRI for 19 years now since the old command line only days. (back when it worked :o). I spend more hours each week trying to get ArcGIS to work properly, than I do actually using the software.

As people said above, pretty nice/easy for basic map develpment (though has printing issues) and works good for that. otherwise, going downhill fast with each new version and update.

Nowadays, they keep putting in more and more bells and whistles, but break the basic GIS data management tools. I have gotten it down now to where I only have 9.3.1 and 10 installed as none of them have a complete set of "non-bug" tools for data development and scripting. especially python scripting.

will work in one version, not the other. things that then work get broken with the next sp.

I think their "work around" database is probably the largest on the planet as well. Instead of "fixing" stuff, they suggest their "work around".

I think it is like Microsoft. Much better OS's out there, but MS got their foot in the door early, so that is what the "standard" is.

Of course, working with government contractors, don't have much choice. However, if I were starting from scratch, ESRI would be the last one I'd recommend. Well, that's not true, I would NEVER recommend them, last or not.

R_

Professional ESRI beta tester (and not by choice)(and, we, the customers, pay THEM to do it)

Answer 12

In response to the "ESRI is slow ...so I am surprised to see you find it quick to develop on" idea: it's important to separate web-map/services development from "desktop" development from "real" development.

webmaps - It's been a few years since I seriously looked at serving webmaps with an ESRI (ArcIMS) or Open Source (Mapserver) platform, so things might be different now. At that time mapserver was faster/better both in performance and dev time. Reading various conversations here and there on the web indicate things are much improved with ArcGIS Server. The open source side has seen even more activity; from a distance the relative balance looks much the same. In any case web facing maps and services have never been ESRI's strong suite, as much as one can do with arcgis server now, it pales in comparison to the desktop tools. They're pumping boatloads of energy into arcgis server though so this may change eventually.

Desktop development - this is modeller and the command line interface, followed by cleaning up and extending the saved scripts from that (but many never find the need to go that far). With modeller the initial design & protoyping phase is a snap as one drags and drops compponents and defines relationships and dependencies between the processes. The model is not just conceptual, not just a picture. It's a tool that can be saved and re-used on real data at will.

Turn on the CLI and run the canned toolboxes at will. As the tools run the actual commands executed and their parameters are reported. These reports can be copied to a text editor or spreadsheet, tweaked as desired, and pasted back into the CLI for execution. The CLI is also interactive, showing possible commands to choose from as one types, with tooltips for possible parameters, and accepts drag-n-drop from the various arcmap/catalog panels.

Desktop development in ArcGIS is fast, largely intuitive, and well integrated. (I do have a looong list of desired improvements and gripes mind you!)

Real development - this is firing up Visual Studio or [insert-favourite-IDE] and building something direct from ArcObjects in C#, C++, Java, Python, etc. I can't speak to how developing in this environment compares to other GIS platforms as I've not done it. My hunch is that it has the perception of being more difficult because the number of possible objects to choose from is damnably large. I've been told the ESRI COM library is the largest in the world, bigger than anything even Microsoft has built. That's going to take some time to get the gist of.

Answer 13

I rely on ArcGis for years because it is easy to use and it provides the best layout interface to output the map quality I need, without using Adobe Illustrator after. It is a good software also for geoprocessing and for manual editing of geographic objects.

However, the product is costly and the basic version (ArcView) do not implement very basics tools.

For scripting & automating task, I use Manifold System. It can do what I don't find in ArcView and a lot more. Scripting is really easier than in Arc. But, to produce map, it is not a good software.

Together, they make a good and affordable GIS kit.

Answer 14

1. I'm lazy. I've been using ESRI products since 1996. ArcGIS has frustrated the hell out of me but I'm still going back to get punched in the stomach. So many bugs on fundamentally important simple functions. Meanwhile their adding addtional features...

2. Used many products but still find that the cartographic capabilities are superior in ArcGIS. You can get a much better result in Illustrator, but it's a lot easier maintain one map and data within a unified framework.

3. Easy to get training and learn (I don't mean the worthless ESRI sponsored training classes). There is a vibrant user community that will help you when you run into bugs while trying to get that project finished on time. Lot's of books that work with ESRI data formats.

4. I make a living off ESRI software....

Answer 15

Since Esri make applications for desktop, mobile, server and the new-ish ArcGIS.com, it's relatively easy to migrate the same application across all environments.

For example, you can create a map document in ArcGIS Desktop and push it to the cloud-hosted ArcGIS.com with a few clicks (and pay later for online storage and access). You can use that same hosted version in your mobile or desktop applications, webmaps, etc.

Other organisations produce similar "seamless" packages, such as Boundless Geo's excellent OpenGeo Suite or MapBox's suite, and perhaps they'll soon remove the advantages of ArcGIS. But for me, for the meanwhile, ArcGIS provides enough benefits to justify the cost.

Answer 16

As a Ph.D. student doing analyses on economic geography, I can say I wasn't very impressed with ArcMap.

I got a free license to ArcGIS with almost every extension for free, as well as TransCAD.

I found that both continue the MS Excel paradigm of nested GUI items that are 3, 4, even more layers deep for setting options of what, under the hood, is a function call.

If you need to know exactly what is happening, I think it's much easier to actually look at the open source documentation and read what each option of the function call does.

To give you an example, suppose I have 10,000 survey respondents who visited 90,000 places, and I need to route them using shortest path to model transportation behavior.

I also need to couple this analysis with drawing buffers around the respondents' residences as I am modeling their local built environment's effect on their transport behavior.

With drawing buffers and merging data with other layer attributes, ArcGIS and Maptitude/TransCAD can do the job just fine.

But, once you know PostGIS, a simple one liner will draw a buffer and intersect it with all your attributes, and create a new table for you, making it 'reproducible research', something a GUI based system will never be able to do.

I might also add that this buffer process took 2.5 hours in ArcGIS and 20 minutes in PostGIS. I also find the stability of SQL far superior to ArcGIS.

Then we move on to routing. The setup for ArcGIS is laborious, and the documentation and support horrible IMHO. On the other hand, the setup in PostGIS is a simple one liner to osm2po or osm2pgrouting in command line and you have a routable network.

After getting a network set up in ArcGIS and TransCAD as well (which took much more doing), I ran an equivalent function call (i.e. route my survey respondents from place to place, ~90k routes), and I got about a 50% success rate from TransCAD, continual crashing from ArcGIS (which has still never worked) and a 100% success rate from PostGIS with pgRouting, again with a few simple queries following online tutorials.

I don't know quite as much about the mapping side of things; for me, any PostGIS query is readily mapped in QGIS like butter.

And again, once you get past the few days of learning SQL, you realize that any geometry function in PostGIS is always a call of ST_geomfuction(geomA,geomB), and the rest are just options.

As far as deployment of apps and such, it seems to me the remarkably faster SQL functions and ability to create simple functions from SQL statements and report them to QGIS (or R for that matter) would lead to a much more automated workflow after an initial learning curve.

And as far as spatial statistical modeling, there is no substitute for R if you are doing anything beyond the basic stuff.

So I guess I'm a plus one for PostGIS, QGIS and R. I use ArcGIS to open their proprietary database formats and get it into PostGIS.

Answer 17

I have been in the GIS business for 30 years. Used ArcInfo (with Info) on a DEC MicroVax to do overlays! The answer in my mind is to use what is best for the job. Today you have FOSS, Google, Bing, Oracle, SQL Spatial, PostgreSQL. I break down the question into Data Collection, Data Maintenance, Map Production (hardcopy vs online), and Analytics. I would say depending on your domain requirements ESRI may or may not be the best choice. I am in the utility sector specifically in water and wastewater.

Data Collection and Data Maintenance ESRI has a very strong set of tools in this area. Many of my clients start with CAD because that is how they receive their data from contractors prior to moving it into ArcGIS.

Production - Hardcopy ESRI has very good set of tools for hardcopy map production in my domain. We use it often.

Production - Online When it comes to online mapping we look at ArcGIS as well as other solutions such as MapGuide or WMS. They are both open source and free. ArcGIS is very popular but has a big footprint which is complex and slow full of limitations that you end up discovering along the way. It is appropriate for certain applications but too complex for others that need map viewing and query requirements.

Analystics When it comes to analytics we try to do most of it in the Spatial Database if it supports it rather than using ESRI model builder. Some others may prefer ArcGIS depending on the operations they are performing. We like to publish the data out of a Geodatabase into a pure Spatial Database for building applications and integration. Opens our options vs. having to use SDE and Geodatabase using ArcEngine/ArcObjects.

Answer 18

We used to use ArcGIS. As of 2014, they are charging "service credits" to do things like geocoding. Software issues/bugs aside, because we do a large amount of geocoding, we can no longer afford the software. We are still deciding what to use, but may go back to Maptitude, which we used 10 or so years ago.