Much is written about bicycle durability, but often little is actually communicated. This note is an attempt to help improve the discussion.
Both sellers and buyers often describe parts as "strong" or "durable". There are several things to consider:
By way of example, I was chatting in 1987 with a friend's roomate who asked my opinion of what pedals he should buy. When I asked about what kind of riding he did, he said something to the effect of "not much -- I've only ridden about 200 miles." At this point I invented and yelled at him the phrase "Don't buy upgrades, ride up grades!" (since mis-attributed to anonymous).
To even casual cyclists, 200 miles in 2 years will seem a small amount. One ten mile ride per month exceeds that.
However, consider that Jobst Brandt has wheels he has ridden over 300,000 miles (2006). Pieces of the wheels -- many parts have been replaced, including the rims, a few spokes and nipples, bearings, races, and the rear axle. However, the hub shells and many of the spokes and nipples have gone the whole way.
Lest this seem like a singularity, Hans-Joachim Zierke reports Sachs Torpedo hubs, made from 1903-1955 had a typical service life of several hundred thousand kilometres. The change from a threaded sprocket reduced service life to 100,000km or less -- and there were customers who noticed.
This raises the question "why do you care?", or perhaps "how much do you care?" For a racer who replaces their bicycle yearly, a bike that fails after 1.25 years is more than good enough. For a rider who wants to go ride and not worry constantly, the more-durable bicycle has clear benefits.
A second consideration is "how does it fail?" A catastrophic failure -- one which often results in rider injury -- is more important to avoid than one which is simply a nuisance. Indeed, a part which fails sooner but fails more kindly may be a very good tradeoff.
A third issue is the rider and type of riding. A part may have good durability with light riders but not heavy ones, or the type and danger of the failure may change. Similarly, riders may be airborne frequently or rarely, may ride in wet or gritty conditions frequently or rarely, and so on. Sometimes, small differences have a big effect. As example, Shimano Octalink crank/spindle joints fail due to joint lash, and failure rates are increased if the rider often stands right-foot-forward, thus "working" the lash.
Another relevant question is "at what cost?" A derailleur chain is narrower than a 1-speed/3-speed chain, and the narrower chain wears out more quickly. However, the wider chain is hard to use with derailleurs, and for many riders derailleur gearing is worth the shorter chain lifetime. Conversely, Jobst's 300,000 mile wheels weigh only tens of grams more than some wheels that have flange failures after only a few miles. Although grams add up, when a modest weight increase brings a dramatic increase in service life, there are riders for whom slightly higher weight but much longer service is the better choice.
In thinking about and discussing these issues, it is also important to consider failure rates. Flat tires are common but rarely dangerous. Snapped or separated steer tubes are much rarer, but despite their rarity probably cause a higher total number of serious injuries.
It is common to speak of what is "normal". Often, people use the word "normal" to mean "average" (either mean or median, rarely stated). In technical use, however, "normal" typically means more than 95% of the samples fall within the range. As example, in casual use people often say "normal riders travel 2,000km per year", whereas in technical use people say "normal riders travel from 200 to 20,000 km per year."