{"course":{"slug":"ai-for-small-business","title":"AI for a Small Business","url":"https://addaly.com/learn/ai-for-small-business","level":"beginner","track":"work"},"lesson":{"slug":"six-months-later","title":"Going back to the numbers you wrote down","url":"https://addaly.com/learn/ai-for-small-business/six-months-later","position":88,"of":88,"minutes":9,"answer":"The trial measured a fortnight under attention, so the only honest verdict comes from re-measuring six months later against the same baseline.","keyIdea":"The trial measured a fortnight under attention, so the only honest verdict comes from re-measuring six months later against the same baseline, when the novelty has gone and the tool is being used the way it will actually be used."},"body":{"format":"markdown","text":"## Why the fortnight was not the answer\n\nThe two-week trial gave you something real: a baseline, a comparison, a kill line. It also measured the best possible fortnight. One person, interested, careful, watched.\n\nSix months on, three things have changed. The novelty has gone, so the tool is used the way it will be used for years. Other people are using it, usually less carefully. And the world has moved: your prices, your products, the model underneath, possibly the vendor.\n\nSo the six-month look is not a formality. It regularly reverses a trial's verdict in both directions: tools that seemed transformative turn out to be used twice a month, and tools that seemed marginal turn out to have quietly absorbed a job nobody misses.\n\n## The measurement, which is the same one\n\nGo back to the five-day tally from module one. Run it again, for the same task, for five ordinary days.\n\nThen four questions:\n\n**Is the task taking less time than the original baseline?** Not less than it felt like. Than the number you wrote down.\n\n**Is anybody still doing it this way?** Check honestly. A common six-month finding is that the person quietly went back to the old method in week seven and did not say so, because saying so felt like a criticism of a decision you had made.\n\n**What did the freed hours become?** From module one: hours saved and not reassigned are hours that evaporate. If nothing changed (no new work taken on, no shorter days, no task that was not getting done) then the saving was real and had no effect, which is worth knowing before you do it again.\n\n**What has gone wrong, and how did you find out?** The errors, the near-misses, the customer who noticed. If the answer is nothing at all, ask how you would have known, and put a deliberate wrong item through the approval queue.\n\n## The sunset list\n\nWhile you are there, the quarterly audit from earlier in this module. Anything nobody has opened in six weeks goes.\n\nAdd one category the audit misses: automations nobody has looked at. An automation that has been running for six months without a single human glance at its output is not working well: it is running, which is different, and the log lesson exists for this moment.\n\n## Honouring the kill line\n\nYou wrote down, in advance, the result that would mean stopping. This is the point at which you either honour it or discover you were never going to.\n\nThe pressure against it is real. You chose the tool. You told people it would help. There is an annual subscription. Sunk cost is a cost you have already paid, and it cannot be recovered by paying more of it, which is easy to write and hard to do when the decision was yours in front of people you work with.\n\nA negative result honoured is worth more than a positive one hoped for. It is also the thing that makes the next trial believable to everybody watching.\n\n## Following the field without chasing it\n\nThe last practical question: how much attention should you give to what changes next?\n\nNot much, and on a schedule. An hour a quarter, spent on two questions. Has anything become possible that was not: the honest answer most quarters is no, and occasionally it is yes and worth a fortnight. And has anything I depend on changed underneath me: prices, terms, the model, the vendor.\n\nIgnore the rest. The pace of announcement is far faster than the pace of anything that changes what a small business should do, and a year of headlines usually reduces to two or three things that mattered.\n\n## What good looks like after a year\n\nTwo or three tasks faster, measured against numbers you still have. A written record of what the tools get wrong. More than one person able to do the work. Your customer data somewhere you can account for. And a couple of tools cancelled along the way without regret.\n\nThat is an unexciting list and it is what success actually looks like here. The businesses that do badly with this are not the ones that moved slowly. They are the ones that never wrote down the number they started from, and so never found out."},"check":{"question":"At six months a firm finds the drafting task takes 40 minutes a week against a 90-minute baseline, but the two staff who do it have quietly reverted to writing by hand. What is the most likely explanation for both findings?","options":[{"text":"The staff are under-reporting their time because they are reluctant to admit they abandoned a tool the owner introduced, so the 40 minutes is unreliable.","correct":false,"tempting":"A plausible social dynamic, and it would produce a low figure without the task genuinely being faster."},{"text":"The task itself changed (fewer or shorter items) so the time fell for reasons unconnected to the tool.","correct":true},{"text":"The tool improved the templates and phrasing the staff now reuse by hand, so the saving is real and persists even though the tool is no longer opened.","correct":false,"tempting":"A genuine spillover that happens, and it would be visible as reused text rather than as a halving of the time."},{"text":"The baseline was measured during an unusually busy period, so 90 minutes overstated the ordinary week and the apparent saving is an artefact of when the tally was taken.","correct":false,"tempting":"A real risk with any baseline, and the five-day tally was specifically taken over ordinary days to avoid it."}]},"prev":{"slug":"the-two-week-trial","title":"The Cheap Tool, and an Honest Two Weeks","url":"https://addaly.com/learn/ai-for-small-business/the-two-week-trial"},"next":null,"authorship":"Founder and series editor: Dr Mustafa Mun. Written by AI models to a written standard. Exam keys checked blind by a second model: 3,665 of 3,666 agreed. Academic review invited.","review":{"status":"Draft, open for academic review"},"cite":"Addaly (2026) 'Going back to the numbers you wrote down', in AI for a Small Business. Edited by M. Mun. Addaly. Available at: https://addaly.com/learn/ai-for-small-business/six-months-later (Accessed: 13 October 2026).","licence":{"name":"CC BY-NC-SA 4.0","url":"https://creativecommons.org/licenses/by-nc-sa/4.0/"},"reader":{"schema":1,"title":"Going back to the numbers you wrote down","courseTitle":"AI for a Small Business","module":{"index":9,"slug":"cost-risk-and-the-long-run","title":"What it costs, what it risks, and what happens in a year"},"answer":null,"keyIdea":"The trial measured a fortnight under attention, so the only honest verdict comes from re-measuring six months later against the same baseline, when the novelty has gone and the tool is being used the way it will actually be used.","hero":{"kind":"plate","image":null,"alt":null,"caption":null},"blocks":[{"t":"h2","id":"why-the-fortnight-was-not-the-answer","runs":[{"text":"Why the fortnight was not the answer"}]},{"t":"p","runs":[{"text":"The two-week trial gave you something real: a baseline, a comparison, a kill line. It also measured the best possible fortnight. One person, interested, careful, watched."}]},{"t":"p","runs":[{"text":"Six months on, three things have changed. The novelty has gone, so the tool is used the way it will be used for years. Other people are using it, usually less carefully. And the world has moved: your prices, your products, the model underneath, possibly the vendor."}]},{"t":"p","runs":[{"text":"So the six-month look is not a formality. It regularly reverses a trial’s verdict in both directions: tools that seemed transformative turn out to be used twice a month, and tools that seemed marginal turn out to have quietly absorbed a job nobody misses."}]},{"t":"h2","id":"the-measurement-which-is-the-same-one","runs":[{"text":"The measurement, which is the same one"}]},{"t":"p","runs":[{"text":"Go back to the five-day tally from module one. Run it again, for the same task, for five ordinary days."}]},{"t":"p","runs":[{"text":"Then four questions:"}]},{"t":"p","runs":[{"text":"Is the task taking less time than the original baseline?","b":true},{"text":" Not less than it felt like. Than the number you wrote down."}]},{"t":"p","runs":[{"text":"Is anybody still doing it this way?","b":true},{"text":" Check honestly. A common six-month finding is that the person quietly went back to the old method in week seven and did not say so, because saying so felt like a criticism of a decision you had made."}]},{"t":"p","runs":[{"text":"What did the freed hours become?","b":true},{"text":" From module one: hours saved and not reassigned are hours that evaporate. If nothing changed (no new work taken on, no shorter days, no task that was not getting done) then the saving was real and had no effect, which is worth knowing before you do it again."}]},{"t":"p","runs":[{"text":"What has gone wrong, and how did you find out?","b":true},{"text":" The errors, the near-misses, the customer who noticed. If the answer is nothing at all, ask how you would have known, and put a deliberate wrong item through the approval queue."}]},{"t":"h2","id":"the-sunset-list","runs":[{"text":"The sunset list"}]},{"t":"p","runs":[{"text":"While you are there, the quarterly audit from earlier in this module. Anything nobody has opened in six weeks goes."}]},{"t":"p","runs":[{"text":"Add one category the audit misses: automations nobody has looked at. An automation that has been running for six months without a single human glance at its output is not working well: it is running, which is different, and the log lesson exists for this moment."}]},{"t":"h2","id":"honouring-the-kill-line","runs":[{"text":"Honouring the kill line"}]},{"t":"p","runs":[{"text":"You wrote down, in advance, the result that would mean stopping. This is the point at which you either honour it or discover you were never going to."}]},{"t":"p","runs":[{"text":"The pressure against it is real. You chose the tool. You told people it would help. There is an annual subscription. Sunk cost is a cost you have already paid, and it cannot be recovered by paying more of it, which is easy to write and hard to do when the decision was yours in front of people you work with."}]},{"t":"p","runs":[{"text":"A negative result honoured is worth more than a positive one hoped for. It is also the thing that makes the next trial believable to everybody watching."}]},{"t":"h2","id":"following-the-field-without-chasing-it","runs":[{"text":"Following the field without chasing it"}]},{"t":"p","runs":[{"text":"The last practical question: how much attention should you give to what changes next?"}]},{"t":"p","runs":[{"text":"Not much, and on a schedule. An hour a quarter, spent on two questions. Has anything become possible that was not: the honest answer most quarters is no, and occasionally it is yes and worth a fortnight. And has anything I depend on changed underneath me: prices, terms, the model, the vendor."}]},{"t":"p","runs":[{"text":"Ignore the rest. The pace of announcement is far faster than the pace of anything that changes what a small business should do, and a year of headlines usually reduces to two or three things that mattered."}]},{"t":"h2","id":"what-good-looks-like-after-a-year","runs":[{"text":"What good looks like after a year"}]},{"t":"p","runs":[{"text":"Two or three tasks faster, measured against numbers you still have. A written record of what the tools get wrong. More than one person able to do the work. Your customer data somewhere you can account for. And a couple of tools cancelled along the way without regret."}]},{"t":"p","runs":[{"text":"That is an unexciting list and it is what success actually looks like here. The businesses that do badly with this are not the ones that moved slowly. They are the ones that never wrote down the number they started from, and so never found out."}]}],"check":{"question":[{"text":"At six months a firm finds the drafting task takes 40 minutes a week against a 90-minute baseline, but the two staff who do it have quietly reverted to writing by hand. What is the most likely explanation for both findings?"}],"options":[{"text":[{"text":"The staff are under-reporting their time because they are reluctant to admit they abandoned a tool the owner introduced, so the 40 minutes is unreliable."}],"correct":false,"tempting":[{"text":"A plausible social dynamic, and it would produce a low figure without the task genuinely being faster."}]},{"text":[{"text":"The task itself changed (fewer or shorter items) so the time fell for reasons unconnected to the tool."}],"correct":true,"tempting":null},{"text":[{"text":"The tool improved the templates and phrasing the staff now reuse by hand, so the saving is real and persists even though the tool is no longer opened."}],"correct":false,"tempting":[{"text":"A genuine spillover that happens, and it would be visible as reused text rather than as a halving of the time."}]},{"text":[{"text":"The baseline was measured during an unusually busy period, so 90 minutes overstated the ordinary week and the apparent saving is an artefact of when the tally was taken."}],"correct":false,"tempting":[{"text":"A real risk with any baseline, and the five-day tally was specifically taken over ordinary days to avoid it."}]}],"order":[2,3,0,1],"why":null},"prev":{"slug":"the-two-week-trial","title":"The Cheap Tool, and an Honest Two Weeks","minutes":9},"next":null}}