{"version":1,"tree":{"n":"html","c":[{"n":"head","x":"\n ","l":"\n ","c":[{"n":"title","x":"Desconocido","l":"\n \n "},{"n":"link","l":"\n","a":[["rel","stylesheet"],["type","text/css"],["href","../../stylesheet.css"]]},{"n":"link","l":"\n","a":[["rel","stylesheet"],["type","text/css"],["href","../../page_styles.css"]]}]},{"n":"body","a":[["class","calibre"]],"c":[{"n":"div","x":"| ","a":[["class","calibre_navbar"]],"c":[{"n":"a","x":"Siguiente","l":" | ","a":[["href","javascript:void(0)"],["rel","articlenextlink"],["data-xQvX3JSyyUS4yAvbblzwf5","{\"name\": \"feed_12/article_6/index_u21.html\", \"frag\": \"\"}"]]},{"n":"a","x":"Menú de sección","l":" | ","a":[["href","javascript:void(0)"],["data-xQvX3JSyyUS4yAvbblzwf5","{\"name\": \"feed_12/index_u18.html\", \"frag\": \"article_5\"}"]]},{"n":"a","x":"Menú principal","l":" | ","a":[["href","javascript:void(0)"],["data-xQvX3JSyyUS4yAvbblzwf5","{\"name\": \"index_u63.html\", \"frag\": \"feed_12\"}"]]},{"n":"a","x":"Anterior","l":" | ","a":[["href","javascript:void(0)"],["rel","articleprevlink"],["data-xQvX3JSyyUS4yAvbblzwf5","{\"name\": \"feed_12/article_4/index_u43.html\", \"frag\": \"\"}"]]},{"n":"hr","l":"\n","a":[["class","calibre6"]]}]},{"n":"div","a":[["class","calibre-nuked-tag-article"]],"c":[{"n":"div","x":"Bartleby","a":[["class","calibre8"]]},{"n":"h1","x":"Why so many IT projects go so horribly wrong","a":[["class","calibre9"]]},{"n":"div","x":"And how to do them better","a":[["class","calibre19"]]},{"n":"p","x":"may. 08, 2025 01:50 ","a":[["class","calibre10"]]},{"n":"div","a":[["class","calibre-nuked-tag-article"]],"c":[{"n":"img","a":[["src","images/img1_u53.jpg"],["title","People carrying a laptop with ERROR 404 written on the screen and a football coming out of it. "],["class","calibre3"],["data-calibre-src","feed_12/article_5/images/img1_u53.jpg"]]}]},{"n":"div","a":[["class","calibre11"]]},{"n":"p","a":[["class","calibre12"]],"c":[{"n":"span","x":"L","a":[["data-caps","initial"],["class","calibre13"]]},{"n":"span","x":"ET’S PLAY","l":" a word-association game. Champions (Liverpool). Cyclists (smug). ","a":[["class","calibre14"]]},{"n":"span","x":"IT","l":" project (failure). Ask people what they know about tech projects, and they will probably say that they take longer than expected, cost more than budgeted and deliver less than they are meant to. Is that perception accurate? And if so, what can be done about it? A forthcoming paper, by Bent Flyvbjerg of Oxford University and his co-authors, sheds light on these questions.","a":[["class","calibre14"]]}]},{"n":"p","x":"Mr Flyvbjerg has made a career out of studying the ways in which megaprojects go wrong. These disappoint with amazing regularity (Tottenham Hotspur). Less than 10% of them meet their initial estimates for both cost and time.","a":[["class","calibre12"]]},{"n":"p","x":"Many of the projects he studies are grand infrastructure schemes: dams, railways, nuclear-storage facilities and the like. If you don’t work in one of the relevant industries, your experience of such projects is likely to be limited to complaining about traffic jams and wasted money (Chelsea). But tech projects are ubiquitous. They affect every industry: Gartner, a research firm, estimates that $5.6trn will be spent on ","a":[["class","calibre12"]],"c":[{"n":"span","x":"IT","l":" in 2025. And they can affect every part of an organisation.","a":[["class","calibre14"]]}]},{"n":"p","x":"In their new paper Mr Flyvbjerg and his colleagues claim to have confirmed for the first time that ","a":[["class","calibre12"]],"c":[{"n":"span","x":"IT","l":" does have a particular problem. They compare actual and budgeted costs for ","a":[["class","calibre14"]]},{"n":"span","x":"IT","l":" projects with 22 other categories, and find good news and bad. The good news is that lots of ","a":[["class","calibre14"]]},{"n":"span","x":"IT","l":" projects stick to their budgets (Brighton & Hove Albion): only 40% of them overrun on costs, compared with 100% of Olympic games, 97% of nuclear-power plants and 75% of hydroelectric dams.","a":[["class","calibre14"]]}]},{"n":"p","x":"The bad news is that when bills do start to mount up, they can go completely bananas. Just under a fifth of ","a":[["class","calibre12"]],"c":[{"n":"span","x":"IT","l":" projects have a cost overrun of more than 50% of their budget, which again isn’t too bad. But among these budget-busting projects, the mean cost overrun is above 450%, the worst of all the project types that the researchers looked at. The only thing that comes close is nuclear-waste storage. In the jargon of risk, something about ","a":[["class","calibre14"]]},{"n":"span","x":"IT","l":" projects means they have very fat tails.","a":[["class","calibre14"]]}]},{"n":"p","x":"Theories abound as to what that something is. One explanation is that ","a":[["class","calibre12"]],"c":[{"n":"span","x":"IT","l":" is a relatively young discipline. If people have been building roads for millennia and still regularly screw up new projects, it should not be a surprise when software does worse. Another relates to standards. You wouldn’t be let loose on a new bridge unless you had formal training as an engineer; the barriers for ","a":[["class","calibre14"]]},{"n":"span","x":"IT","l":" are not as high.","a":[["class","calibre14"]]}]},{"n":"p","x":"Intangibility provides a third explanation for why things might spiral out of control. You can set milestones for a high-speed rail line and physically inspect a building site; it’s much harder to visit, or even visualise, a software development. Abstraction exacerbates the risk that managers will start adding in new requests rather than sticking to limited goals (Southampton).","a":[["class","calibre12"]]},{"n":"p","x":"That touches on another point: ","a":[["class","calibre12"]],"c":[{"n":"span","x":"IT","l":" projects reach deep inside organisations, which means tech neophytes often call the shots. It also means these projects bring lots of change. A new road does not require motorists to drive differently, but digitisation usually requires employees to alter how they work, increasing the likelihood of office politics and internal resistance.","a":[["class","calibre14"]]}]},{"n":"p","x":"Mitigating these risks isn’t easy, but Mr Flyvbjerg and his co-authors have some useful ideas. The first, familiar bit of advice is to avoid making software projects too bespoke; the paper lauds Apple’s app store for imposing clear constraints on app designers and developers. Their second (Arsenal) suggestion is to take more time in the planning phase of a project. Big ","a":[["class","calibre12"]],"c":[{"n":"span","x":"IT","l":" initiatives have a mean duration of 3.2 years in the sample used by the researchers, compared with 6.9 years for the other project categories. The researchers speculate that the preparatory work for many ","a":[["class","calibre14"]]},{"n":"span","x":"IT","l":" projects gets rushed, leading to trouble later on.","a":[["class","calibre14"]]}]},{"n":"p","x":"That worry is especially resonant given excitement among businesses over artificial intelligence. Efforts to deploy ","a":[["class","calibre12"]],"c":[{"n":"span","x":"AI","l":" are likely to face all the same problems that beset conventional ","a":[["class","calibre14"]]},{"n":"span","x":"IT","l":" projects, with bells on. ","a":[["class","calibre14"]]},{"n":"span","x":"AI","l":" is novel. It could change the jobs of everyone in the organisation. And belief in its potential is sky-high: bosses who do not understand the technology are demanding that it be rolled out quickly. ","a":[["class","calibre14"]]},{"n":"span","x":"AI","l":" may well be game-changing (Manchester City). But being thoughtful about its deployment will reduce the chances of miserable failure (Manchester United). ","a":[["class","calibre14"]]},{"n":"span","x":"■"}]},{"n":"p","a":[["class","calibre12"]],"c":[{"n":"i","x":"Subscribers to The Economist can sign up to our ","a":[["class","calibre18"]],"c":[{"n":"a","x":"Opinion newsletter","l":", which brings together the best of our leaders, columns, guest essays and reader correspondence.","a":[["href","https://www.economist.com/newsletters/opinion"]]}]}]}]},{"n":"div","x":"\n","a":[["class","calibre_navbar"]],"c":[{"n":"hr","l":"\n","a":[["class","calibre6"]]},{"n":"p","x":"This article was downloaded by ","l":"\n","a":[["class","calibre16"]],"c":[{"n":"strong","x":"calibre","l":" from ","a":[["class","calibre13"]]},{"n":"a","x":"https://www.economist.com/business/2025/05/08/why-so-many-it-projects-go-so-horribly-wrong","a":[["href","https://www.economist.com/business/2025/05/08/why-so-many-it-projects-go-so-horribly-wrong"],["rel","calibre-downloaded-from"]]}]},{"n":"br","a":[["class","calibre-nuked-tag-article"]]},{"n":"br","l":" | ","a":[["class","calibre-nuked-tag-article"]]},{"n":"a","x":"Menú de sección","l":" | ","a":[["href","javascript:void(0)"],["data-xQvX3JSyyUS4yAvbblzwf5","{\"name\": \"feed_12/index_u18.html\", \"frag\": \"article_5\"}"]]},{"n":"a","x":"Menú principal","l":" | ","a":[["href","javascript:void(0)"],["data-xQvX3JSyyUS4yAvbblzwf5","{\"name\": \"index_u63.html\", \"frag\": \"feed_12\"}"]]}]}]}]},"ns_map":["http://www.w3.org/1999/xhtml"]}