In the digital age, call centres are anachronisms — vestiges of old practices and poor ways to communicate complicated information. Any information transferred through a call centre can be less clear than information transmitted through other methods, as it is literally lost in translation. That said, if you are going to operate call centres, you should do it right. The actual utility (or lack thereof) is measurable and can be used to monitor and improve performance.\nConsider this: an internal Canada Revenue Agency report made public by the CBC this year revealed that the overall accuracy rate at CRA call centres was just 75%. The finding was based on a test conducted in 2014 in which agency employees anonymously phoned in and asked routine tax questions. Previously, CRA had claimed its call centres gave "fully correct" answers 92.5% of the time.\nCRA ordered the survey in response to empirical tests conducted by the Canadian Federation of Independent Business (CFIB) in 2010 and 2012. Those tests found only 61% of callers to CRA call centres ultimately received the correct information; 19% were given incorrect advice.\nPolicy-makers should take a close look at the CFIB reports. They are by definition more objective than the CRA’s survey: they were made public and employed open, repeated, unaltered and understandable methodology. Additionally, the CFIB reports had only four standard questions that any tax practitioner would consider easy, such as, how long must a business owner keep tax records before they can be destroyed?\nCRA’s call centres receive about 17.3 million calls a year. Considering this massive number, CRA call centres are routinely dispensing misleading answers in the millions. And that might only be the calls that get through.\nCRA does not publish accuracy standards for its call centres but perhaps it should. It expects callers to be able to get through 85% of the time, claiming to have exceeded that target in 2013-2014. But the agency’s internal survey says only 56% of test callers got through on the first attempt; the remainder had to call an average of almost three times, some of them more than 14 times. The CFIB’s 2012 survey showed 20% of its test callers could not get through at all, after employing a three-attempt process. What can be done?\nOne problem is connecting. The CFIB conducted 25% of its test calls in French and found a significant disparity. Only 3% of French callers reported an inability to connect, compared with 22% of English callers. This seems to indicate the problem is simply inadequate resources. Perhaps it’s necessary to bring in more people, especially during peak periods. In the meantime, some English-speaking practitioners call the French number first, playing the percentages of getting a bilingual person.\nThe bigger issue is what to do about wrong answers. What if taxpayers act on mistaken advice and then face interest, penalties or audits? In order to avoid any of those outcomes, some callers routinely ask the same question several times and use the most common response, rather than chance one inquiry resulting in problematic advice. This is hardly a best practice.\nPerhaps the CRA should start by putting that ubiquitous call centre refrain into action by recording calls and using them for "quality assurance purposes." It should also invest in developing more consistent, correct and easily searchable public online documents. Most information-savvy enterprises summarize their call centre FAQs (frequently asked questions) online and enable email questions. These steps would reduce the number of calls to CRA call centres and the number of wrong answers. Until taxpayers can more easily navigate the mass of information available to get the answers they need, CRA’s motto might as well be "Don’t call us, we’ll call you."