I frequently draft contracts for clients. I rarely recommend arbitration.
Pros and Cons
Arbitration is faster than going to court. Arbitration is not much cheaper (if at all) than going to court, although the spending is more compressed. On average, arbitration decisions themselves aren't meaningfully more or less accurate than having a case decided by a judge. But, unlike a court decision, an arbitration award generally isn't subject to appeal, so if the arbitrator gets it wrong, you have no recourse.
There are rare cases where an arbitration decision gets overturned by a court -- sometimes the decision itself doesn't provide an answer to the parties have because it isn't clearly drafted, sometimes it is possible to prove corruption on the part of the arbitrators, and sometimes the arbitrators are so grossly wrong that a court intervenes. (The rare linked case where a court intervened was one where a court first vacated an arbitration award for manifest disregard of employment law, and then ignored the court's order to consider the law on remand to the arbitrator).
Arbitration is generally not a good option when part of the relief a party needs is to direct a public official to take some prompt action (e.g. evict someone, enforce a non-competition agreement with injunctive relief, or impose a preliminary injunction against a governmental rule). It is also not a good option where important evidence is likely to be held by third parties who aren't party to the agreement.
If the stakes are high relative to your net worth, you don't want to have to risk the uncertainty that comes with arbitration, which is even greater than the uncertainty that comes with court resolutions where appeals and review of previous public proceedings can provide insight.
But, arbitration awards can be, and routinely are, affirmed by courts even when it is demonstrably clear that the arbitrators made clear mistakes of fact not supported by the evidence that they heard, or definitely get the law wrong. To be overturned by a court (as in the case linked above), they need to be told by a party what the correct law is, and still snub their nose at the law in a situation where the law is unambiguously clear (in the case above, no attorneys fees at all were awarded to an employee who was awarded substantial compensatory and punitive damages in an employment discrimination case as a result of the employer's proven age discrimination, until a federal court remanded the case for reconsideration).
When Arbitration Makes Sense
For arbitration to be desirable for a client, one of the following things must be true:
1. No individual matter arbitrated must be too important for the client to lose big because the arbitrator reaches the wrong result, and the client must have a large number of matters to arbitrate, OR
2. The client would be significantly harmed by allowing the public to view the process, OR
3. Swift resolution of claims has a value greater than an accurate resolution of claims, OR
4. The skill set or ideological disposition required for a just resolution is completely beyond the competence of judges and juries.
Most cases where arbitration makes sense fall into my first category above.
For any big business that handles lots of transactions with consumers, a substantial percentage of which produce disputes, arbitration is great. The stakes in any individual case are small, the cases are small enough for cases wrongly decided in a customer's favor to be balanced by cases wrongly decided in the company's favor, cases get off the balance sheet quickly, and class actions which can threaten a company's existence are largely prevented from arising. The risk of paying large court costs at the end of a process selected by one of the parties whom the arbitration arrangement has an institutional interest in keeping happy also discourage disgruntled consumers from complaining in the first place.
The laws of probability themselves favor plaintiffs in class actions based on weak cases (often weak because of a legal determination that a judge has to stretch to reach, rather than a factual determination), for example, cases with only a 20%-30% chance of success. On average, most of the plaintiffs will lose one by one, if cases are decided independently. But, they can have a real chance of all winning if they get lucky and a court (and single jury hearing the case) resolve the case in their favor, and this gives class action cases great settlement value, because many businesses could not survive losing a class action case.
Arbitration may be tolerable in these cases for consumers in small transactions too, as arbitration may be somewhat more friendly to pro se parties in small cases. But arbitration is a nightmare for employees or others in transactions which are a big deal for the small player, but no big deal for the big business. The asymetry in this class of cases is what makes arbitration controversial. Probably the most important cases of abuse in this class of casees involve arbitration agreements between securities firms and the employers and investors they do business with, where the track record of abuse is abysmal (not surprisingly, the arbitration award overturned in the above link involved this system).
Another manifestation of my first category involves resolving routine small disputes between repeat players who are "big boys" in a way the prevents all out war from breaking out. Labor arbitration happens because both unions and management see the resolution of individual employee disputes as expendible and small, relative to the need to continue dealing with each other. Similarly, arbitration is a good way to resolve disputes between Realtors over who is entitled to a commission (and how much) when both were involved in a transaction, or between multiple insurance companies over who has to cover a claim when the damages are clearly less than the coverage limits but who is on the risk is unclear. Arbitration is also often a good mechanism for resolving disputes between a general contractor and a subcontractor who frequently work together on construction jobs.
A good example of a case in the second category is a dispute between a patent lawyer and a client over fees. Preserving the confidentiality of the secret patent information and the litigation strategy involved may supercede other interests. Another example of a case in the second category is an arbitration agreement in place between units within a political party over financial disputes between subunits of the party; where the harm litigation could do to the common purpose of all party members outweighs other considerations.
An example of the third class of claims are disputes in sports matches (some of which are high stakes business propositions). Major League Baseball can't afford to let the "pine tar incident" go unresolved (and with it, for the outcome of the season's pennant race) to go unresolved, for the several years that a court process accompanied by appeals can take.
Examples of the fourth class of claims include situations like property disputes involving a gay couple in a region that is hostile to gays, disputes involving the right to hold clerical office, resolutions of disputes over technical scientific questions, and business disputes in countries with poorly developed legal systems in the third world.
In contrast, mediation, which simply requires the parties to discuss resolution of a case on a facilitated basis, is appropriate in almost every dispute, unless (1) it is too small a matter to be worth the extra effort to be worthwhile to hire both a third party mediator and a third party judge, (2) it requires too quick a resolution to allow for that delay (e.g. election cases), (3) the is no middle ground for compromise in the dispute, or (4) the parties lack the authority to resolve the matter by mutual agreement.
Courts require, or strongly encourage, mediation in most classes of civil cases and even some criminal matters, anyway.
Mediation's problem, shared with the court system generally, is that the party with the most staying power and best understanding of the likely resolution of the case on the merits is at an advantage. Mediation can legitimatize exploitation of a financially pressured or unrepresented party.