# What estimates mean *Which number are you asking for?* --- Your team says the project will take two months, and they're being honest: two months is the number they'd give you if everything they can foresee goes the way they expect. Could it land in one month? Almost impossible, because there's a floor on how fast work can go. Could it take five? Easily - a dependency slips, a requirement changes, a key person gets pulled onto something urgent. There's no ceiling on delay. That asymmetry means [the average outcome sits above the honest middle](https://erikbern.com/2019/04/15/why-software-projects-take-longer-than-you-think-a-statistical-model.html), and the gap is bigger than people expect: the early finishes save you weeks, the late ones cost you months. None of that is visible in the number itself. A single estimate misleads the way [[Reading the numbers|a monthly pack]] does, only more quietly, because nobody argues with it and so nobody checks it. As a rule of thumb, plan against the team's number times 1.2 for routine work, and times 1.5 where the work is genuinely new. Treat those as floors rather than forecasts. Two separate bodies of evidence land on an average overrun of about 1.8 times: one on time, from software task records, and one on cost, from more than five thousand IT projects across sixty-six countries. Both find a heavy tail: most projects land near their estimate, and the average is made almost entirely by the few that don't. The deeper problem is the shape in our heads. Almost every statistic we're taught assumes the bell curve, where outcomes land symmetrically around the middle and every kind of average is the same number, sat dead centre, as likely to be beaten as missed: <svg viewBox="0 0 345 152" role="img" aria-label="symmetric bell curve, all averages coincide" style="width:100%;max-width:430px;display:block;margin:0.4em 0 1.1em;"> <polygon points="10.0,121.7 13.6,121.6 17.3,121.5 20.9,121.4 24.5,121.2 28.2,121.0 31.8,120.8 35.4,120.5 39.1,120.1 42.7,119.7 46.3,119.2 50.0,118.6 53.6,117.8 57.2,117.0 60.9,116.0 64.5,114.9 68.1,113.6 71.8,112.1 75.4,110.4 79.0,108.4 82.7,106.3 86.3,103.9 89.9,101.3 93.6,98.5 97.2,95.4 100.8,92.1 104.5,88.6 108.1,84.8 111.7,80.9 115.4,76.9 119.0,72.7 122.6,68.4 126.3,64.1 129.9,59.8 133.5,55.5 137.2,51.4 140.8,47.5 144.4,43.7 148.1,40.3 151.7,37.2 155.3,34.5 159.0,32.2 162.6,30.4 166.2,29.1 169.9,28.3 173.5,28.0 177.1,28.3 180.8,29.1 184.4,30.4 188.0,32.2 191.7,34.5 195.3,37.2 198.9,40.3 202.6,43.7 206.2,47.5 209.8,51.4 213.5,55.5 217.1,59.8 220.7,64.1 224.4,68.4 228.0,72.7 231.6,76.9 235.3,80.9 238.9,84.8 242.5,88.6 246.2,92.1 249.8,95.4 253.4,98.5 257.1,101.3 260.7,103.9 264.3,106.3 268.0,108.4 271.6,110.4 275.2,112.1 278.9,113.6 282.5,114.9 286.1,116.0 289.8,117.0 293.4,117.8 297.0,118.6 300.7,119.2 304.3,119.7 307.9,120.1 311.6,120.5 315.2,120.8 318.8,121.0 322.5,121.2 326.1,121.4 329.7,121.5 333.4,121.6 337.0,121.7 337.0,122 10.0,122" fill="currentColor" opacity="0.06"/> <polyline points="10.0,121.7 13.6,121.6 17.3,121.5 20.9,121.4 24.5,121.2 28.2,121.0 31.8,120.8 35.4,120.5 39.1,120.1 42.7,119.7 46.3,119.2 50.0,118.6 53.6,117.8 57.2,117.0 60.9,116.0 64.5,114.9 68.1,113.6 71.8,112.1 75.4,110.4 79.0,108.4 82.7,106.3 86.3,103.9 89.9,101.3 93.6,98.5 97.2,95.4 100.8,92.1 104.5,88.6 108.1,84.8 111.7,80.9 115.4,76.9 119.0,72.7 122.6,68.4 126.3,64.1 129.9,59.8 133.5,55.5 137.2,51.4 140.8,47.5 144.4,43.7 148.1,40.3 151.7,37.2 155.3,34.5 159.0,32.2 162.6,30.4 166.2,29.1 169.9,28.3 173.5,28.0 177.1,28.3 180.8,29.1 184.4,30.4 188.0,32.2 191.7,34.5 195.3,37.2 198.9,40.3 202.6,43.7 206.2,47.5 209.8,51.4 213.5,55.5 217.1,59.8 220.7,64.1 224.4,68.4 228.0,72.7 231.6,76.9 235.3,80.9 238.9,84.8 242.5,88.6 246.2,92.1 249.8,95.4 253.4,98.5 257.1,101.3 260.7,103.9 264.3,106.3 268.0,108.4 271.6,110.4 275.2,112.1 278.9,113.6 282.5,114.9 286.1,116.0 289.8,117.0 293.4,117.8 297.0,118.6 300.7,119.2 304.3,119.7 307.9,120.1 311.6,120.5 315.2,120.8 318.8,121.0 322.5,121.2 326.1,121.4 329.7,121.5 333.4,121.6 337.0,121.7" fill="none" stroke="currentColor" stroke-width="1.4" opacity="0.85"/> <line x1="173.5" y1="28.0" x2="173.5" y2="122" stroke="#c8503a" stroke-width="1.2"/> <text x="173.5" y="20" font-size="10.5" fill="#c8503a" text-anchor="middle">most likely = even odds = the mean</text> <line x1="10" y1="122" x2="337" y2="122" stroke="currentColor" stroke-width="0.8" opacity="0.5"/> <text x="10" y="138" font-size="10.5" fill="currentColor" opacity="0.65">sooner</text> <text x="337" y="138" font-size="10.5" fill="currentColor" opacity="0.65" text-anchor="end">later</text> </svg> Projects don't get that shape. Early is bounded, late can blow up, and the tail drags the averages apart: <svg viewBox="0 0 345 152" role="img" aria-label="right-skewed project duration curve, averages diverge" style="width:100%;max-width:430px;display:block;margin:0.4em 0 1.1em;"> <polygon points="11.5,122.0 14.0,122.0 16.6,121.4 19.1,119.8 21.6,116.5 24.1,111.7 26.6,105.6 29.2,98.6 31.7,91.1 34.2,83.4 36.7,75.9 39.3,68.6 41.8,61.9 44.3,55.8 46.8,50.3 49.4,45.5 51.9,41.3 54.4,37.8 56.9,34.9 59.4,32.5 62.0,30.8 64.5,29.4 67.0,28.6 69.5,28.1 72.1,28.0 74.6,28.2 77.1,28.7 79.6,29.4 82.2,30.4 84.7,31.5 87.2,32.7 89.7,34.1 92.3,35.7 94.8,37.3 97.3,38.9 99.8,40.7 102.3,42.4 104.9,44.2 107.4,46.1 109.9,47.9 112.4,49.8 115.0,51.6 117.5,53.4 120.0,55.3 122.5,57.1 125.1,58.8 127.6,60.6 130.1,62.3 132.6,64.0 135.1,65.7 137.7,67.3 140.2,68.9 142.7,70.4 145.2,71.9 147.8,73.4 150.3,74.9 152.8,76.3 155.3,77.6 157.9,79.0 160.4,80.3 162.9,81.5 165.4,82.7 167.9,83.9 170.5,85.1 173.0,86.2 175.5,87.3 178.0,88.3 180.6,89.3 183.1,90.3 185.6,91.3 188.1,92.2 190.7,93.1 193.2,94.0 195.7,94.8 198.2,95.6 200.7,96.4 203.3,97.2 205.8,97.9 208.3,98.7 210.8,99.4 213.4,100.0 215.9,100.7 218.4,101.3 220.9,101.9 223.5,102.5 226.0,103.1 228.5,103.7 231.0,104.2 233.5,104.7 236.1,105.2 238.6,105.7 241.1,106.2 243.6,106.6 246.2,107.1 248.7,107.5 251.2,107.9 253.7,108.3 256.3,108.7 258.8,109.1 261.3,109.5 263.8,109.8 266.4,110.2 268.9,110.5 271.4,110.8 273.9,111.1 276.4,111.4 279.0,111.7 281.5,112.0 284.0,112.3 286.5,112.6 289.1,112.8 291.6,113.1 294.1,113.3 296.6,113.6 299.2,113.8 301.7,114.0 304.2,114.2 306.7,114.4 309.2,114.6 311.8,114.8 314.3,115.0 316.8,115.2 319.3,115.4 321.9,115.6 324.4,115.7 326.9,115.9 329.4,116.1 332.0,116.2 334.5,116.4 337.0,116.5 337.0,122 11.5,122" fill="currentColor" opacity="0.06"/> <polyline points="11.5,122.0 14.0,122.0 16.6,121.4 19.1,119.8 21.6,116.5 24.1,111.7 26.6,105.6 29.2,98.6 31.7,91.1 34.2,83.4 36.7,75.9 39.3,68.6 41.8,61.9 44.3,55.8 46.8,50.3 49.4,45.5 51.9,41.3 54.4,37.8 56.9,34.9 59.4,32.5 62.0,30.8 64.5,29.4 67.0,28.6 69.5,28.1 72.1,28.0 74.6,28.2 77.1,28.7 79.6,29.4 82.2,30.4 84.7,31.5 87.2,32.7 89.7,34.1 92.3,35.7 94.8,37.3 97.3,38.9 99.8,40.7 102.3,42.4 104.9,44.2 107.4,46.1 109.9,47.9 112.4,49.8 115.0,51.6 117.5,53.4 120.0,55.3 122.5,57.1 125.1,58.8 127.6,60.6 130.1,62.3 132.6,64.0 135.1,65.7 137.7,67.3 140.2,68.9 142.7,70.4 145.2,71.9 147.8,73.4 150.3,74.9 152.8,76.3 155.3,77.6 157.9,79.0 160.4,80.3 162.9,81.5 165.4,82.7 167.9,83.9 170.5,85.1 173.0,86.2 175.5,87.3 178.0,88.3 180.6,89.3 183.1,90.3 185.6,91.3 188.1,92.2 190.7,93.1 193.2,94.0 195.7,94.8 198.2,95.6 200.7,96.4 203.3,97.2 205.8,97.9 208.3,98.7 210.8,99.4 213.4,100.0 215.9,100.7 218.4,101.3 220.9,101.9 223.5,102.5 226.0,103.1 228.5,103.7 231.0,104.2 233.5,104.7 236.1,105.2 238.6,105.7 241.1,106.2 243.6,106.6 246.2,107.1 248.7,107.5 251.2,107.9 253.7,108.3 256.3,108.7 258.8,109.1 261.3,109.5 263.8,109.8 266.4,110.2 268.9,110.5 271.4,110.8 273.9,111.1 276.4,111.4 279.0,111.7 281.5,112.0 284.0,112.3 286.5,112.6 289.1,112.8 291.6,113.1 294.1,113.3 296.6,113.6 299.2,113.8 301.7,114.0 304.2,114.2 306.7,114.4 309.2,114.6 311.8,114.8 314.3,115.0 316.8,115.2 319.3,115.4 321.9,115.6 324.4,115.7 326.9,115.9 329.4,116.1 332.0,116.2 334.5,116.4 337.0,116.5" fill="none" stroke="currentColor" stroke-width="1.4" opacity="0.85"/> <line x1="72.1" y1="28.0" x2="72.1" y2="122" stroke="currentColor" stroke-width="1" opacity="0.55" stroke-dasharray="3 3"/> <text x="72.1" y="20" font-size="10.5" fill="currentColor" opacity="0.7" text-anchor="middle">most likely</text> <line x1="106.0" y1="45.2" x2="106.0" y2="122" stroke="currentColor" stroke-width="1" opacity="0.55" stroke-dasharray="3 3"/> <text x="110" y="41" font-size="10.5" fill="currentColor" opacity="0.7">even odds</text> <line x1="150.3" y1="74.9" x2="150.3" y2="122" stroke="#c8503a" stroke-width="1.2"/> <text x="153" y="70" font-size="10.5" fill="#c8503a">the mean</text> <line x1="10" y1="122" x2="337" y2="122" stroke="currentColor" stroke-width="0.8" opacity="0.5"/> <text x="10" y="138" font-size="10.5" fill="currentColor" opacity="0.65">sooner</text> <text x="337" y="138" font-size="10.5" fill="currentColor" opacity="0.65" text-anchor="end">later</text> </svg> The most likely outcome, the even-odds outcome and the mean are now three different numbers, in that order, and each has a natural customer. Ask a team cold and you usually get the most likely, because an estimate is made by [imagining the plan going to plan](https://en.wikipedia.org/wiki/Planning_fallacy), which is why it is the one number nobody should plan against. A deadline people can coordinate around wants even odds, and a decent project manager pads the raw estimate towards it, whether they'd use those words or not. Whoever owns the budget needs the mean. Anyone quoting you one of these is not wrong; they may just be answering a different question from the one you asked. I've learnt this one the hard way, having been too afraid to look unknowledgeable, or to look like I was overly questioning someone who was notionally the expert. Where the bets matter, if someone can't explain a timeline clearly to you as a non-expert, that should be an amber flag. Business cases and delivery plans land on my desk the same way: they always show things delivered on time and strong returns. So my baseline is that things get delayed and plans under-deliver, and I nudge from there on the credibility of the paper. The better thought out it is, the smaller the adjustment. > [!note]- The working > Task durations tend to follow a [lognormal distribution](https://en.wikipedia.org/wiki/Log-normal_distribution), which is the skewed curve above: bounded on the left, unbounded on the right, so the mean sits above the median and the single most likely outcome sits below both. The gap widens with uncertainty, which is why the multiplier is bigger for work nobody has done before. The 1.2 and 1.5 are calibrated rules of thumb rather than anything the distribution hands you. The practical move is to hold two numbers at once, doing two different jobs. The team plans against the honest middle, because that's the number everyone coordinates around, and if you gave them the padded one they'd plan to it and take it. The board gets the mean, because across enough projects that's the number that makes the budget land roughly on plan. Most of the trouble estimates cause comes from one number being asked to do both jobs. ## Inside a business case A business case is where all of this lands at once. Six months of development at £500k, fifty sales at £30k, a 3x return. Every figure in it is a single point standing in for a range, and the ranges are lopsided in the same direction as the estimates above: costs have a floor and no ceiling, sales ramp late and start discounted. Rerun that case with nine months of development and thirty early deals at £25k and the 3x has evaporated to breakeven. The second set of assumptions is no more true than the first. The headline had that much room in it, and the case never said so. So the useful version of the slide says "between 1x and 3x, depending mainly on how long development actually takes", and the conversation changes from approve-or-reject to a constructive debate about what needs to be true. The front page of the last three-year plan I signed off carried a short list of exactly that - no unplanned exits, delivery on time, no adverse change in policy - and the sign-off conversation was about the list, not the spreadsheet. ---