# Early delinquency goes cold in the first 30 days > The 1 to 30 day window is where the most is recovered per contact, and the one that looks least urgent. What happens when volume outgrows the team. Collections ## Early delinquency goes cold in the first 30 days The 1 to 30 day window is where the most is recovered per contact, and the one that looks least urgent. What happens when volume outgrows the team. **Luis Aguilar** September 14, 2026 3 min read IN SHORT - Collections plays out in stages: preventive before the due date, early from 1 to 30 days, mid from 31 to 90 and critical from 91 to 120. - Every day without contact in the early stage moves a current account closer to write-off, and that cost shows up in no report. - With AI agents running the portfolio, contact rates rise 34% and time to first response falls 61%. - In 2025, $2.5M was recovered with AI agents, with $31M in payment promises recorded and validated. CONTENTS - The early stage is the most neglected - What an autonomous collections agent is - What changes in the operation A portfolio is not lost all at once. It is lost in stages, and each stage has its own window. - 01 Preventive before the due date The customer is still current. The contact is a reminder, not a collection. - 02 Early 1 – 30 days The window that decides the most. The customer recognises the debt and can usually pay. - 03 Mid 31 – 90 days The account competes with other priorities. Negotiation begins. - 04 Critical 91 – 120 days Recovery, with different terms and less room. The four stages of a delinquency collections strategy. Each one is a different conversation, and treating them alike is the most common way to lose recoverable receivables. ### The early stage is the most neglected It looks least urgent and decides the most. An account three days past due is still a simple conversation: the customer acknowledges the debt, remembers the date and can usually pay. Forty days later that same account has moved behind other priorities, the customer has stopped answering and the tone has changed. What gets in the way is not the team’s willingness. It is volume. A portfolio of thousands of accounts in the early stage demands thousands of contacts inside a thirty-day window, and that work depends on someone pushing it by hand, one at a time. ### What an autonomous collections agent is A chatbot replies from a script: it recognises an intent and returns text written in advance. An **AI agent** understands the context of the conversation, checks the institution’s systems, decides within the policies it was given and carries out the full task, leaving a trace of every decision. The practical difference lies in what it is allowed to do, and what it is not. That is not programmed: it is declared. **CrediAmigo** Agent · actions live buscarUsuario Database Look up the customer by ID clasificarUsuario Logic Classify the credit level requisitos Data Validate approval requirements aprobacion Logic Conditions to approve credit cuotaMensual Compute Compute the monthly payment Code Interpreter Web Search The panel where you declare what the agent can do, what it cannot, and when it hands the case to a person. Extensions and negotiation limits are the institution's policy, not the vendor's. [See how an agent is built →](https://www.shimliapp.com/en/funciones) Applied to a portfolio, that means the agent reviews each customer’s profile, history and behaviour before writing; decides the message, the channel and the time; negotiates within the authorised extensions; and escalates to a person when the case falls outside policy. ### What changes in the operation | Contact rate customers actually reached | +34% | | --- | --- | | Time to first response from when the case arrives | −61% | Only the two metrics that share a unit. Kept promises multiply by three, but that is a multiple and not a percentage: putting it on the same bar would compare different things. Published Shimli results · 2025 On top of that, **3× kept promises** against those recorded. During 2025, **$2.5M recovered** in portfolio and **$31M in payment promises** recorded and validated. The detail that weighs most is none of those figures on its own, but that the early window stops depending on how many people are available that day. Apply it to your operation - How many accounts enter early delinquency each month, and how many does your team actually reach inside those thirty days? - What extension can a collector offer without asking for approval? If the answer is not written down, it cannot be delegated either. - What happens today to a payment promise nobody follows up: does it get recorded somewhere, or is it lost in the conversation? [See the full path, from portfolio load to validated payment →](https://www.shimliapp.com/en/casos/cobranza) ### Questions on this topic What is early delinquency? The 1 to 30 day arrears stage. It is the window where the customer still acknowledges the debt, remembers the date and can usually pay, so a timely contact recovers more per attempt than at any later point. Why is recoverable debt lost at that stage? Because it is the one that looks least urgent. A portfolio of thousands of accounts in early delinquency demands thousands of contacts inside a thirty-day window, and that work depends on someone pushing it by hand, one at a time. Useful to someone on your team? Share it with whoever decides on this process. [WhatsApp](https://wa.me/?text=Early%20delinquency%20goes%20cold%20in%20the%20first%2030%20days%20https%3A%2F%2Fwww.shimliapp.com%2Fen%2Fblog%2Fearly-delinquency-goes-cold) [LinkedIn](https://www.linkedin.com/sharing/share-offsite/?url=https%3A%2F%2Fwww.shimliapp.com%2Fen%2Fblog%2Fearly-delinquency-goes-cold) [Email](mailto:?subject=Early%20delinquency%20goes%20cold%20in%20the%20first%2030%20days&body=Early%20delinquency%20goes%20cold%20in%20the%20first%2030%20days%20https%3A%2F%2Fwww.shimliapp.com%2Fen%2Fblog%2Fearly-delinquency-goes-cold) Copy link ### Keep reading [See all →](https://www.shimliapp.com/en/blog) - Platform #### FINNOSUMMIT 2026: one in three sessions was about AI Instead of summarising impressions, I counted the sessions: 22 of the 67 had AI in the title. That says a lot about where banking in the region stands. September 25, 2026 - Support #### Your next banking channel is already installed: WhatsApp AI agents are doing to apps what apps did to websites. What that means for an institution running real volume, and what separates a chatbot from an agent. September 25, 2026 - Platform #### The next AI advantage is: trust What the Volcano Innovation Summit 2026 left behind: when capability gets cheap, the advantage becomes trust. And trust has two sides. September 16, 2026 ### Put an agent to work on this process. Create the account and start today. Or we take 30 minutes and define which metric you'll move. [Create account](https://shimli.app/user/auth/create-account) Book a demo [See the full collections case →](https://www.shimliapp.com/en/casos/cobranza)