Operational deadline alerts. It only goes off when there's something to do.
The contract renewed itself because the notice deadline passed three weeks ago. Nobody forgot: nobody knew. This piece checks deadlines every morning and flags only the ones you can still do something about.
It rarely alerts, which is why you read it. Only deadlines that need action come through. The rest stays on the list, where you can look it up, without making a sound.
Every alert has a name on it. An alert sent to everyone belongs to no one, so every deadline has a named recipient and a date by which it has to be dealt with.
Ignored alerts are counted. That's the tracker's health check: if the share goes up, the system is turning into noise and needs recalibrating.
This page covers a single piece of the system. The other pieces, and how we choose which one to start with, are on the services page.
The deadlines that get missed are the ones nobody owns
Nobody forgets a date they're already carrying around in their head. The ones that get missed are buried in a contract signed four years ago by someone who now does a different job, and were never written down anywhere.
The second kind that gets missed is the one with too many owners. A date that concerns three people gets checked by none of them, because each assumes someone else is on it, and that assumption holds right up until the day after the deadline.
A missed deadline doesn't cost you the day it falls due. It costs you the twelve-month renewal that kicks in automatically the day after.
What changes in practice
Auto-renewals stop being a surprise. The notice deadline is flagged with the lead time agreed for that type of contract, to a specific person, and stays open until someone records what they've decided.
The second change is that deadlines stop living in people's heads. Someone going on holiday doesn't take a date away with them, and someone changing roles leaves the tracker behind instead of taking it with them in a notebook.
The third is the most useful, and it shows up after a few months: you find out how many deadlines there were that nobody had ever counted. The first inventory is almost always the biggest surprise.
An alert that always goes off stops being read
That's why this piece rarely alerts, and the evidence comes from a field where alerts matter far more than in an office. A review of 17 studies published in the Journal of the American Medical Informatics Association puts it bluntly: “Drug safety alerts are overridden by clinicians in 49% to 96% of cases”.
These are alerts about patient safety, ignored up to ninety-six times out of a hundred. A reminder about a supply renewal, in a list that sends out fifteen a day, stands far less chance, and the result is a system that's installed and that nobody looks at.
The practical consequence is a design rule, not a preference: the bar for an alert to go off is raised until every alert corresponds to something that needs doing. Everything else stays in a list you can consult, which is a different place from a person's attention.
Alerts that interrupt get acted on less
The form an alert takes changes the outcome, and that's been measured too. A 2019 systematic review in the same journal, which narrowed 1,011 articles down to 39, compared alerts that pop up in front of people as they work with alerts that don't.
Interruptive alerts were accepted in 38.67% of cases, non-interruptive ones in 61.57%, a statistically significant difference. Demanding attention more forcefully gets you less of it, which is the opposite of what almost every alert system assumes.
The authors' conclusion also points to the other lever: “Alert fatigue may be mitigated by redesigning the interactive behavior of CDS and tailoring CDS to clinical roles”. Taken out of the hospital, that means tailoring alerts to the role of the person who receives them: the same alert sent to everyone is worth less than one sent to the person who can act on it.
Hence the choice: the alert lands somewhere people check, with the urgency spelled out and the due date on top. It only interrupts when the window for action is about to close, and in that case the interruption is the message.
What goes into the tracker, and what doesn't
Anything with a fixed date and something you can do about it goes in. A date with no action attached isn't a deadline, it's information, and information belongs in a list, not an alert.
| Type | How it's handled | What you get |
|---|---|---|
| Auto-renewalscontracts and supply agreements | The notice is based on the cancellation deadline written into the contract, not the expiry date, and stays open until a decision is made. |
Renewing becomes a decision again, instead of something that just happens. |
| Documents with an expiry datecertifications, lab analyses | The expiry date is read from the document when it comes in, and the notice period allows for how long that body takes to renew it. |
No expired document discovered by a customer or an inspector. |
| Routine inspectionsequipment and installations | The frequency is set once, and every completed inspection automatically schedules the next one with the same notice period. |
The calendar keeps itself up to date, instead of being rebuilt every year. |
| Tax and legal mattersout of scope | The date is flagged as something to raise with the business's adviser, with no interpretation by the system. |
The reminder arrives, and the judgement stays with the person who signs off on it. |
What the system does, step by step
The check runs every morning at the agreed time and looks at the coming days. Every step leaves a log entry, so at the end of the month you know what was flagged, to whom, and what happened next.
| Step | What happens | What you get |
|---|---|---|
| The inventorydone once | Dates are gathered from contracts, spreadsheets, calendars and incoming documents, and each one is given a type and an owner. |
For the first time there's a complete list, and it's almost always longer than expected. |
| The notice periodset for each type | Each type of deadline has its own lead time, based on how long it actually takes to act. |
The alert arrives while there's still time to do something, not the day before. |
| The filteronly what needs action | Deadlines already dealt with, postponed or with no possible action don't trigger alerts, but you can still look them up. |
Anything that goes off always means something needs doing, so people pay attention to it. |
| The recipientone name, not everyone | Every alert has one person's name on it and stays open until that person records what they've decided. |
No more deadlines that concern three people and get checked by none of them. |
| The reviewevery month | We count how many alerts were closed without any action, and the types of deadline producing them are dropped or recalibrated. |
The tracker stays readable instead of swelling until it turns into noise. |
The number almost nobody tracks
The agreed metric is the share of deadlines flagged with the agreed notice, and it's measured like the others: we look at where it stands today over a real period, set the threshold the work has to beat not to count as a failure, and compare afterwards.
The second number is the one that keeps the system alive, and hardly anyone collects it: how many alerts get closed without anyone doing anything. It's the clinical finding above, translated directly into an office setting.
When that share rises, the answer isn't to send more insistent alerts. You raise the thresholds, take some types of deadline off the list of those that trigger alerts, and check that every remaining alert still has a possible action behind it.
What goes out on its own, and what waits for a person
All of the checking runs on its own: reading the dates, working out the notice periods, filtering, routing to the recipient and flagging anything still open past its due date. They're repetitive jobs that don't commit the business to anyone.
Nothing that amounts to a decision or a communication goes out on its own. A cancellation notice, a renewal, a repair order or any message to a supplier or customer is prepared and waits, because they're decisions. The full reasoning is on our page about the principles we build by.
Same mechanism, different name in every trade
The mechanism is identical; the deadline isn't. It's worth looking at your own case, because that's where you see which dates currently exist only in someone's head.
| Sector | The deadline that gets missed | Where you see it |
|---|---|---|
| Food and agricultureand export | A certificate or lab analysis found to have expired just as a foreign buyer is asking for it to clear customs. |
|
| Restaurantsand bars | Routine equipment inspections and supply contracts that quietly renew on old terms. |
|
| Hospitalityaccommodation and events | Deadlines with booking portals and agencies, and the pre-season deadlines that all land together in the busiest month. |
What this piece doesn't do
It doesn't interpret tax and legal deadlines. It flags them as something to raise with the business's adviser, with the date and the document attached, and stops there, as we do with anything that affects a tax return or a contract.
It doesn't send communications on your behalf. It prepares, flags and keeps things open: the cancellation notice, the renewal and the request to a supplier are still actions a person takes.
It doesn't extract dates by itself from documents it has never seen. Reading incoming documents, with a check before any data goes in, is a separate piece, handled by automatic document entry, and the two work well together. What happened during the month goes into monthly reporting.
Questions and answers
Why not alert on every deadline?
Because an alert that always goes off stops being read, and that's been measured. Van der Sijs's review in the Journal of the American Medical Informatics Association, covering 17 studies, found that doctors override drug safety alerts between 49% and 96% of the time.
Those are alerts about a patient's health: if even those get ignored, a reminder about a contract renewal stands even less chance. So the bar for sending an alert is set high by design, and anything that doesn't need action stays on the list without making a sound.
What kinds of deadline can it track?
Anything with a fixed date and something you can do about it: auto-renewing contracts and supply agreements, notice periods, certifications and lab analyses with an expiry date, routine equipment inspections, price lists and commercial agreements that expire, domain and licence renewals, recurring internal tasks.
Deadlines that depend on a tax or legal judgement are left out. They're flagged as something to raise with your own adviser, and we don't interpret them.
How do you stop alerts turning into noise?
Through three decisions made up front. Every alert has a named recipient, because an alert sent to everyone belongs to no one. Every type of deadline has its own notice period, because a contract with ninety days' notice and a domain renewal don't get flagged at the same time.
And the alert doesn't interrupt: it lands somewhere people check, with the urgency spelled out next to it. A 2019 review in the same journal found that interruptive alerts were accepted 38.67% of the time, against 61.57% for non-interruptive ones.
Do we need new business software?
No. The dates are gathered from wherever they already are: contracts in folders, spreadsheets, shared calendars, deadlines picked up from documents coming into the business. Where an internal system can be connected to from outside, the dates are read directly; where it can't, you start with a shared list using the same fields, which stays compatible with whatever comes next.
The real work is the first inventory of existing deadlines, and it's done once.
How do you measure whether it's working?
By the share of deadlines flagged with the agreed notice, which is the metric, plus a second number almost nobody tracks: how many alerts get closed without anyone doing anything.
If that share rises, the tracker is turning into noise and needs recalibrating, by raising the thresholds or dropping certain types of deadline. An alert system nobody reviews becomes just another list to ignore within a few months.
Notes on sources
- The 49% and 96% figures come from van der Sijs, Aarts, Vulto and Berg, Overriding of Drug Safety Alerts in Computerized Physician Order Entry, Journal of the American Medical Informatics Association 13(2), 2006, pages 138-147, a review of 17 studies, quoted word for word. It looks at doctors responding to clinical alerts, not office staff responding to a contract reminder: we use it for the principle, which is that attention drops when alerts are frequent, not to apply the percentage to your office.
- The 38.67% against 61.57% comes from Hussain, Reynolds and Zheng, Medication safety alert fatigue may be reduced via interaction design and clinical role tailoring, same journal, 2019, a systematic review that narrowed 1,011 articles down to 39. The same limitation on the sample applies, and the authors warn that the studies measure acceptance in ways that can't be compared directly.
- We don't publish the number of minutes it takes to get back into a task after an interruption. The figure quoted everywhere on this doesn't appear in the study it's attributed to: it comes from things the author said later in interviews, and repeating it would dress up an impression as a measurement.
- We don't publish any figure for the average cost of a missed deadline. The estimates available measure large organisations with contracts on a completely different scale, and they vary a lot from one survey to the next, even from the same source. The cost is best worked out on your own case, by looking at the contracts that auto-renewed over the last two years.
- This page doesn't report results achieved for a client, because this piece hasn't yet been delivered to a client. The tests mentioned are functional checks run in a test environment.
The other pieces in this group
Deadlines and numbers nobody checks in timeFifteen minutes, with your case in front of us.
Count the contracts that renewed themselves over the last two years without anyone deciding to renew them. If nobody knows the number, that's already the most useful answer. In fifteen minutes on the phone we'll look at it together and tell you where it makes sense to start, even if we never end up working together.
You'll speak to Mattia Esposito, who then builds the system: there's no salesperson in between. If you'd rather measure things yourself before talking, the Diagnostico (in Italian) is twenty questions and five minutes.