Transcript 0:00 Your tracker is lying to you, not on purpose. It is lying because you built it to depend on a person remembering to update it. And people are busy, so every day 0:15 the spreadsheet drifts a little further from what really happened until the dashboard tells you one story and everyday life tells you another. I spent a night fixing exactly this in my own business, 0:28 and the fix is not what you think. It is not nagging anyone to be better. Today, I'm going to show you a smarter move. [upbeat music] Welcome to Evolve Out Loud. 0:46 I'm Chris桑澤. This is where I think out loud about AI and education, what's working, what isn't, what I'm building, what I'm seeing in the field, what I'm trying to figure out this week. 0:59 Some of it lands, some of it doesn't. And honestly, that's kind of the whole point of this show. Let's get into it. This is episode ten: Stop Asking People to Log It. 1:13 Just a reminder to everybody listening that Evolve Out Loud presents views and opinions that are my own and do not reflect any organization that I work for or am affiliated with in any way. 1:30 If you are new here, I take the big areas I work in and break each one down with a simple framework. Three questions: What is it? Why does it matter? And how do you implement it? No theory for the sake of theory. 1:47 Just the thing, why it is worth your time, and how to do it Monday morning. Today, we are in my second lane, the tools and automations I build for myself as a practitioner. 2:02 And the idea I want to put in your hands is reconciliation, a system that checks itself against the truth instead of waiting for a human to feed it. Let's start with what it is. 2:14 A reconciliation system is a small automation that reads the place where the truth already lives and updates your records to match. That is the whole idea. You stop typing the same status into a tracker by hand, 2:31 and you point the tracker at the record that is already true. Here is what happened to me. I run an outreach pipeline for my consulting work. Every contact sits in a stage to contact, contacted, follow up. 2:47 And for a while, the only thing that moved a person from one stage to the next was me clicking a button inside my app after I sent an email. The problem is, I do not always send from the app. 2:59 Sometimes I just reply from my email like a normal human. Those emails never counted, so the pipeline showed people sitting in to contact like I had never reached out when I had emailed them three times. 3:12 So I stopped trusting the app's memory. I pointed the whole thing at the one record that never lies, my sent folder. Now a job 3:22 reads what really went out, matches it to the right person, and writes the truth back into the pipeline. Last contact becomes the actual date. 3:33 If someone has been emailed but is parked in the wrong stage, the system moves them on its own. No logging, no discipline required. That is reconciliation. The source of truth already exists. You just 3:47 teach the system to go read it. What it is not is more data entry. It is the opposite. It deletes data entry because the event already got recorded somewhere the moment it happened. 4:00 Okay, why does this matter, especially if you are leading a team? Because when a system is inaccurate, the instinct is almost always the same. 4:10 Ask people to be more disciplined, log more, update the spreadsheet, be more careful, send the email reminding everyone to update the tracker by Friday, and it never works. Not because your people are careless. 4:24 It never works because the problem was never discipline. You build a system that depends on perfect human behavior, and then you acted surprised when humans were human. Think about 4:38 how much of this lives in a school, a compliance tracker that is only right if every case manager updates it on time. An intervention log that depends on a busy teacher entering data between classes. 4:52 A professional learning roster that is accurate only if somebody transfers the sign-in sheet into a spreadsheet. Every one of those is a system waiting to drift because every one of them 5:07 runs on a person remembering to do a second redundant thing after the first thing already happened. And the cost of that drift is not just a messy spreadsheet. You make calls off bad numbers. You chase a deadline 5:25 that was already met. Or worse, you miss one that was not. You spend a staff meeting reconciling by hand what the system should have known. And the people closest to the work quietly learn 5:39 that the tracker is not the truth, so they stop trusting it, which makes it drift even faster. And the event itself already got recorded. The student information system knows who is enrolled. 5:53 The platform you use for plans and timelines knows what is due and when. The sign-in form knows who showed up. The data you are begging people to reenter is already sitting there in a place that is already true. 6:10 So the better move is not a sternly worded reminder. It is to wire your records to reconcile against that source on their own. Find the thing that is already true, and make everything else catch up to it. 6:25 This is the difference between AI that looks great in a demo and AI that holds up in a building on a Tuesday in February. One adds work. It asks your staff to feed the machine. The other removes work. It reads 6:44 what your staff already did and gets out of their way. I'm only interested in the second kind, so let's get practical. How do you build one of these? Four steps. I'm keeping them simple on purpose. Step one: 7:03 find the source that is already true. Pick one tracker you maintain by hand and ask a single question: Where does this event really get recorded the moment it happens? For my outreach, it was my sent folder. 7:18 For enrollment, it is the student information system. For a deadline, it is the platform that owns that deadline. For attendance, it is the sign-in form. If the answer is nowhere except this spreadsheet, 7:33 that is a different problem. But most of the time, the truth already lives somewhere better. Step two: pick the one field that matters and let the source win. Do not try to reconcile everything at once. 7:48 Pick the field that causes the most pain when it is wrong. The last contact date, the compliance status, the did this happen yet flag. Decide on purpose that the source of truth wins and the manual entry stops. 8:07 Step three: wire the read, not more entry. This is where you connect a small automation that reads the source and writes the status back. And you do not need to be a programmer for this. 8:20 You describe the rule in plain language the way you would explain it to a new team member. If a message went to this person, mark them contacted and stamp the date. 8:29 Start it in suggest mode if you can, where it shows you what it would change before it changes anything, so you trust it before you let it run on its own. Step four: run it on a schedule and watch for drift. 8:43 Let it reconcile every night or every hour, whatever fits. Then check it for a week. Does the tracker now match reality without anyone touching it? Watch for the edge cases because there will be a couple. 8:59 In my case, a second email to the same person on the same day was not getting counted, and I had to fix that. Small bug, quick fix. That is normal. You are not aiming for perfect on day one. 9:16 You are aiming for a system that corrects itself instead of one that waits for you. One privacy note because this is education. If the source you are reading holds student data, vet it before you wire anything. 9:32 Where does the data go? Is it FERPA compliant? If you cannot answer that, start with an adult or operational workflow like outreach or professional learning attendance. Improve the pattern there first. 9:48 You can get the whole win without ever touching a student record. So that is the build. Find the source that is true. Pick the one field. Wire the read. Run it and watch for drift. 10:06 Here is why I love this one. It changes the question you ask when something is wrong. The old question is: Who forgot to update this? 10:16 The new question is: What already knows the answer, and why is my system not reading it? The first question creates blame and more meetings. The second one creates a fix, and it gives you back something you cannot buy. 10:30 Trust in your own numbers. When your tracker reflects what really happened, you stop running your team off a dashboard that quietly lies, and you start making decisions on reality. So here's your one move. 10:45 Think of the tracker in your world that is always a little bit wrong, the one you are forever reminding people to update. Do not send that reminder again. 10:55 Ask where the truth already lives and whether your system could just read it. 11:01 And if you want help finding the source of truth for the messy tracker in your world, that is exactly the kind of thing I work on with the people I help. There is a link to book a call in the show notes. All right. 11:16 That's it for this one. [upbeat music] If something here landed, share it with someone who'd get something out of it. If it didn't, hey, it's part of the deal too. 11:27 You can find more of what I'm working on at evalconsulting.com. Book a call if you want to talk through what you're dealing with. Talk to you later. [upbeat music]
