There are only two reasonable answers as to why a company or an organization should consider implementing an Enterprise Content Management program:

1. There is an expected positive Return On Investment (it is good for the bottom line), or
2. There is an external force demanding some function or feature that an ECM program provides (we are being forced to)

I'll start with number 2 because it is easier and then I'll show you at least 5 different ways that ECM can help the bottom line.

There are a very few special cases of "content" that if not managed correctly could see some sort of charges laid against people and it is quite reasonable that an ECM program be considered to ensure the proper handling of such content. The only real example I can think of is the Sarbanes-Oxley act (SOX) and the provision that the CFO and CEO are personally responsible for the truth of financial statements thus a program around those statements seems valid even if the program itself is more hype than reality (and yes I am talking about you, SOX). But this is very rare, almost all laws and regulations only have a potential financial impact on the corporate entity (fines, penalties, lost opportunities etc) and so are just a special case of number 1.

It's always about the money

Money, making more of it or (much more often) spending less of it is the only reason that most businesses or organizations would choose to implement an ECM program. And the ways that ECM can help save money are many and varied:

1. Productivity increases, there is a potential from a variety of sources:
   • automation of business processes (saving time), this is usually referred to as Business Process Management [BPM] though many other ECM tools provide a workflow to automate at least part of a process... there is an assumption built in that the time saved will be applied to other business needs (but usually there is little ability to verify that)
   • decreased number of errors (saving time and legal costs) ... an ECM program should be able to find errors early, often preventing them from being introduced at all and saving the time it takes to redo and/or undo work
     
     • as well as the costs of managing any potential fallout from the error (eg: legal suits and/or remediation efforts that do not happen)
     • but remember that an automated process can also introduce errors and at a rate that cannot be matched by manual processes; only a computer could mail out 10,000 inaccurate rebate checks or electrical invoices for billions of dollars.
   • speed between process steps (saving time) ... though somewhat difficult to quantify, many manual processes have built in time governors like the time it takes to physically mail/courier a letter, or to physically pull paper from a filing cabinet, or deliver paper to the next recipient. ECM can remove many of the time "wasting" portions of a process even if the process itself is not automated
     • it would likely be a mistake to believe that all of the time saved from these menial tasks would be directed to more productive activities. While some benefit should be expected and likely found, much of the "wasted" time is in fact useful and required distractive activity for people. We need to interact, move about, and distract ourselves in order that we be able to focus on our tasks.
2. Better business decisions -- saving time may lead to more productivity but making better business decisions always leads to an improvement to the bottom line
   
   • "Knowledge is power" is a commonly held belief but what we don't always think about is the reverse is true as well ... a lack of knowledge strips power
     • Worse is when you don't know you are missing knowledge, then you think you are powerful and may act accordingly but to the possible detriment of the business decisions at hand -- in other words it is possible that a poorly implemented or malfunctioning ECM program can be worse then no ECM program at all; it may lead to a false sense of empowerment.
   • Breaking down information silos -- An ECM program has inherently a wider scope than the manual and / or departmental programs they replace and thus a more comprehensive view is possible. This means that it is possible to see the big picture even while working on a piece of it, potentially allowing for better integration with the bigger picture.
     
     • But if the ECM program does not offer a way of understanding the relevance of "enterprise" content the extra information may be difficult to use appropriately. Just like too many cooks can spoil a broth, too much content can confuse the business.
3. Lowered Infrastructure Cost ... There is no doubt but that an ECM Application comes at a cost, but with content there usually is the ability to gain efficiencies with scale. 
   • Shared storage means less duplication
   • Archive Server means compression and Single Instance Archiving which mean even less storage
   • Less storage means fewer and faster backup requirements
   • BUT larger capacity and required throughput comes with an increase in costs of both hardware and associated software, ECM can help at reducing Tier 1 storage but many implementations only increase Tier 1 storage requirements
4. Business Continuity -- it is the content that matters
   • Most Disaster Recovery plans focus on the bits and bytes, the hardware and software required to restore the ability to conduct business, but it is CONTENT not applications that drive business. A good ECM program will identify critical business content and make sure it is available on a priority basis for recovery should the need ever rise
5. Reduced external costs
   • The clearest example is the costs associated with a Legal Discovery Order; everything that is provided due to such an order must be reviewed by legal counsel of both the plaintiff and defendant, these costs can range from $50->$500 / hour ... an ECM program can assure the minimum amount of data is provided (by removing content in an appropriately mandated fashion) and by filtering existing data to assure only the wheat (and not the chaff) is delivered
   • If critical content is not identified then often all content is treated as critical. An ECM program can find the critical data and make sure the is treated differently (not stored off site, not encrypted, not on Tier 1 storage, etc)

With so many ways to save money using ECM, why don't most companies implementing ECM speak to the ROI they receive? The answer is a bit complicated but boils down to the fact that these savings are very difficult to identify and measure but the costs associated with ECM are easy to calculate. Thus an accurate ROI is very difficult to account for but with the costs so easily calculated most people selling ECM solutions would rather not speak about ROI except in the very general sense.

Not performing a ROI study is a mistake for those implementing ECM

The problem with not attempting to define, measure, and calculate a ROI is that it becomes almost impossible to know whether or not the ECM program was a success. You can measure the success, as it is often done, by an adoption rate, a customer support ticket rate, or a number of other external indications of success, but in business there is only one measure that really matters ... did it have a positive ROI. Adoption rates may tell you that the application/program is accepted, even liked, but so too is facebook; "happy people" may be "productive people" but if that IS the case then it should be possible to measure the productivity.

I admit it is very difficult to even begin to account for most of the supposed benefits to ECM ... mostly because businesses are generally woefully unaware of how they actually work to begin with. While there are always going to be exceptions to this generality, it is fair to say that most business processes that are not automated are not well understood,  how long each business process currently take, what each step of the process is, how much duplication they currently have, how much time is wasted searching, how much knowledge is lost, how much or even what sort of business-critical data they have. That is one of the major benefits of taking on ECM, at least some of this knowledge is gained as a side-effect of the program.

The reason most ECM programs work is because these data are not known but once a new ECM program is designed and in place, the inefficiencies of the old system become known. Until you go looking for duplicate files, you don't know what sort of benefit you will find. Until you have the ability to search through all enterprise data, you don't know how that will affect things and if you don't know how long people search for content today then you will not be able to know that a post-ECM deployment was of benefit to them (and thus you cannot apply the savings to a productivity increase).

A ROI study is worth the expense

The work of defining the metrics associated with a ROI calculation is absolutely worth the expense, even though that expense itself will lower the ROI. Only real knowledge is real power...remember the saying, "[they] know just enough to be dangerous". Like any major application, ECM application costs can be quite high, as much as $50/user/month (plus bandwidth and client workstations), though that certainly describes a high-end solution. It is not easy to find that kind of long-term cost savings within an organization but often it can be done. 

The time it takes to complete various steps within a process can be estimated to a workable degree of accuracy by the people and/or applications performing them before the ECM application is implemented. The required knowledge may not be readily available but it must exist or the processes simply wouldn't get done.

Old process may still be needed

In a perfect world, once a business process is moved to an ECM system the original process will never be used again. We do not live in a perfect world. While most processes moved to an ECM system do not necessarily require a backup, there are some business critical processes that cannot rely 100% on any single application or solution. Most often IT addresses this possibility with redundancy and Business Continuity Plans to recover the (ECM) application but some processes are so critical that they require a manual backup.

As a consumer we often see where such a thing is needed, say when there is a communication issue and credit/debit cards cannot be approved ... without a manual process the sale is lost. A similar thing could be the case for any business process that is automated and fully documenting an old business process might even be considered as business insurance.

Don't expect much support from your sales representative

Sales hates the thought of a ROI study, even if they won't admit it. The problem is really that the costs side of the study are far easier to calculate then the potential savings, and the results of the study can easily put their future relationship at risk even as it is mostly out of their control. Also, with the software costs only being about 15% of the total costs of an ECM deployment, a good ROI study will only tend to highlight how expensive an application actually is. Worse, for the sales rep, is that most of the other costs involved cannot be controlled by the customer, the only squeeze possible is the software license.

You really can't blame them for not wanting much to do with ROI, it takes a very inefficient process to make the ROI obvious and worthy ... pure-play archiving often solve such problems and sales reps of such tools do like and use ROI studies but you'll see that as you move away from archiving towards the management techniques, the costs rise and the measurable benefits are harder to find. It is an excellent business tool but often not the sales representative's friend.

It is only an incremental cost

I won't claim that a ROI study will be free or even necessarily cheap, but it shouldn't be thought of as massively expensive either. Much of the information required for a ROI study will (or should) be gathered as a part of the application design, the problem space must be understood to some degree or the application will be sure to fail (or, less likely, be incredibly wasteful with resources).