Most organisations do not lose money by negotiating a contract badly: they lose it by letting the contract renew itself because nobody looked at the calendar. The expiry date is known months in advance and still manages to arrive as a surprise.
What follows: why expiry dates slip past, what automates well, what still requires judgement, and how to sequence the project.
A contract lives in a shared folder, in the inbox of whoever negotiated it, or in a physical filing cabinet. It rarely lives anywhere you can ask it when it expires.
While the organisation is small, people’s memory is enough. When it stops being enough, the symptom is always the same: automatic renewals nobody decided, on terms nobody renegotiated.
Why expiry dates slip past
The first reason is that no single inventory exists. Procurement knows its own, technology knows its own, and legal sees only those that came through review — so nobody holds the complete list.
The second is that the relevant date is not expiry but notice. A contract requiring ninety days’ notice is decided three months before the date everyone has in mind.
The third is that the owner changes. The person who negotiated the contract left or moved department, and the informal reminder that held the control together left with them.
What automates well
The repository with metadata. Storing the file is not enough: counterparty, subject, value, term, notice period and owner have to be extracted and recorded. Without those fields the repository is a folder with a better name.
Extracting that data. Intelligent document processing identifies dates, amounts and renewal clauses in the text, leaving the reviewer to confirm rather than to key in. That is what makes loading the back catalogue viable.
Tiered alerts. A warning to the owner far enough ahead to decide, another as the notice date approaches, and an escalation if nobody responded. The chain matters more than the first alert.
The renewal route. Renewing should be a decision following the same approval route as the original signature, not an administrative silence. Automating that route turns renewal into an explicit act.
What requires judgement
The decision to renew, renegotiate or terminate is a business one, and it depends on information the contract does not hold: how well the supplier performed, what alternatives exist, how the need has changed.
Automation supplies the moment and the file; the evaluation is done by a person. Confusing the two produces renewals that are timely and badly decided, which is only a partial improvement.
It is also necessary to decide which contracts enter the system. Including everything from the start turns the project into a mass digitisation exercise; starting with the highest-value ones and those with automatic renewal delivers control where it weighs most.
What changes by type of contract
Recurring service contracts benefit most, because their risk is precisely the silent renewal and their annual value is usually high.
Contracts with periodic obligations — deliveries, reports, service levels — also need tracking during the term, not only at the end. There the system has to remember commitments, not just dates.
And those involving the processing of personal data require an additional review, because the obligations of Habeas Data (Ley 1581) in Colombia and of the LFPDPPP in Mexico extend to the relationship with the third party and are worth verifying at each renewal.
How to sequence the project
Start by building the inventory of live contracts with their expiry date and notice period. Done once, that exercise usually turns up forgotten contracts that have already renewed twice.
Load the highest annual value first, along with those carrying an automatic renewal clause. They are few and they concentrate almost all of the financial risk.
Then extend to the rest incrementally, using each renewal to bring the contract into the system rather than attempting a bulk migration.
What has to exist first
An owner per contract — current, not historical — and an agreement about who authorises a renewal at each value. Without the second, the system warns correctly and the decision still has nobody accountable for it.
It is also worth fixing the success measure before starting: the number of automatic renewals nobody decided should trend to zero, and that figure is easy to measure and hard to argue with.
The file the decision needs
Warning in time solves half the problem. The other half is that whoever decides has what they need in front of them, and does not have to reconstruct it in the days remaining.
That file is short: the current contract with its terms, the supplier’s performance history, what was paid over the period, and who inside the organisation uses the service.
Assembling it automatically when the alert fires changes the nature of the warning: it stops being a reminder that generates work and becomes a decision served up ready. That is the difference between a system people attend to and one they learn to ignore.
Where is it best to start?
With the highest annual value contracts and those with automatic renewal. They are few and they concentrate almost all of the financial risk.
Can contract information be extracted automatically?
To a large extent, yes. Intelligent document processing identifies dates, amounts and renewal clauses, leaving the reviewer to confirm rather than key in. That is what makes loading the back catalogue viable.
Which date actually needs watching?
The notice date, not the expiry date. A contract requiring ninety days’ notice is decided three months before the date everyone has in mind.
Which indicator is worth following?
The number of renewals that happened without an explicit decision. It should trend to zero, and it is easy to measure and hard to argue with.
