← Blog Automating a process
How to reduce PQRS response time
To reduce PQRS response time, answering faster isn't enough. You need correct classification, assigned owners, deadline control, and proof that the answer resolves the issue.
Key takeaways
- The goal isn't to lower the average at any cost: if incomplete answers, appeals, or repeat requests increase, the process didn't improve.
- Start with one queue for web, email, WhatsApp, social media, and in-person channels, with a case number, deadline, owner, and visible status.
- Automate only repeatable, low-risk work: receipt confirmations, case-status lookup, missing-information requests, and drafts reviewed by a person.
- Measure time, overdue cases, repeat requests, and quality; those 4 indicators show whether faster PQRS responses are working.
What does reducing PQRS response time mean?
In Colombia, a PQRS isn't closed because someone sent an email. It's closed when the organization gives a clear, complete, consistent answer within the applicable deadline.
Law 1755 of 2015 regulates the right to petition. The general rule is 15 days to respond, 10 days for document and information requests, and 30 days for consultations, unless a special rule applies. It also requires a prompt, complete, substantive decision. This isn't legal advice; it's an operating guide for teams that manage PQRS.
Some industries have their own rules. In communications services, the CRC indicates 15 business days from the day after filing, with a possible extension when evidence or information is required. If the user disagrees, they can file a reconsideration appeal within the period set by the regulation. In public utilities, Superservicios states 15 business days for petitions, complaints, and appeals, and requires a substantive, clear, and consistent answer.
So reducing response time doesn't mean putting pressure on analysts. It means removing useless waits: cases with no owner, duplicate filings, missing attachments, slow internal checks, and answers that come back because they were incomplete.
Where does time get lost in a PQRS?
Before buying a chatbot or moving everything to AI, measure the real workflow. A useful diagnosis follows the PQRS from intake to closure.
- Unify intake. Bring web, email, WhatsApp, social media, and in-person service into one queue. Each case needs a case number, channel, intake date, request type, and owner.
- Classify by deadline and risk. Don't mix a simple complaint with an appeal or a request that requires evidence. Classification defines the deadline, review level, and team that responds.
- Find missing information. Many days are lost because the team notices late that a contract, invoice, customer number, or support document is missing. The system should ask for it at the start, in clear language.
- Assign a named owner. A shared inbox works for intake. To respond, each case needs a responsible person and a visible alert before the deadline.
- Reduce repeat requests. The DIAN recommends checking frequently asked questions before sending a request and not repeating a PQRS that already has a case number. In a company, this means self-service, case-status lookup, and clear status messages.
- Close with evidence. Store who reviewed the case, which documents were used, what response was sent, and how notice was given. The Ministry of Finance PQRSD Manual, version 05/29/2026, uses guidelines for intake, filing, assignment, follow-up, and closure with document traceability and time control.
Which option helps you answer PQRS faster?
The right solution depends on volume, risk, channels, and existing systems. These are the usual options before building custom software.
| Option | When it makes sense | Typical limit |
|---|---|---|
| What your CRM already includes | You have few channels and good usage discipline. | It fails if it doesn't integrate email, web, and case numbers. |
| Automation platform | You need rules, alerts, and handoffs between teams. | It depends on connectors and well-ordered data. |
| Custom development | You have high volume, custom rules, or integration with internal systems. | It needs design, testing, and ongoing operation. |
| AI assistant or chatbot | It helps with case status, repeated questions, and drafts. | It shouldn't decide sensitive cases without review. |
If you need to connect inboxes, CRM, ERP, or document systems, first review integrations and APIs. If the bottleneck is reading, classification, and drafts, AI automation may apply.
How do you measure whether PQRS automation reduced time?
Measurement must compare before and after. If you only measure speed, you can approve bad answers faster.
- Total time by PQRS type: from filing to notified closure, separated by channel and responsible team.
- Overdue cases or cases close to deadline: review volume by owner, cause, and final percentage.
- Repeat request rate: how many users file again because they don't understand the status or response.
- Response quality: a sample of cases reviewed against substance, clarity, consistency, and evidence used.
- Escalation rate: appeals, complaints to an authority, tutela actions, or legal review caused by incomplete answers.
- Classification accuracy: compare the system label against a sample reviewed by people from the area.
- For AI, we use evaluation sets with anonymized real cases. The NIST AI RMF recommends testing AI systems before deployment and during operation, with documented metrics, methods, and test sets. If you want to see how we measure assistants, you can read how we measure whether an AI assistant answers well.
Illustrative example: collections with repeated PQRS
This example is illustrative. It doesn't describe a client. It shows what a real reduction looks like without sacrificing a substantive response.
Intake
PQRS for unrecognized charges
A user writes by email: “you're charging me for an invoice I already paid.” They attach proof of payment but don't include the contract number. The same case also arrives by WhatsApp.
The system detects a possible duplicate using ID, email, and similar text.
Workflow
Classification and missing-data request
The single queue marks the case as a collections claim, calculates the deadline, and asks for the contract number before sending it to the analyst. If a prior case exists, it responds with the case status.
AI helps classify and draft. A person makes the decision.
Response
Draft with evidence
The system checks payments, the invoice, and internal notes. Then it creates a draft with the facts, the verification performed, and the proposed decision. The analyst reviews, adjusts, and sends it.
A fast response only counts if it's complete and traceable.
Operating result
Less rework
The team avoids reviewing the same case twice, asks for data at the start, and prioritizes what expires first. The savings come from removing waits, not from answering without reading.
If automation increases appeals or complaints, it stops and gets fixed.
What we do
Frequently asked questions
How long does an entity have to answer a PQRS in Colombia?
It depends on the type of request and the industry. Under the general right to petition, petitions, information requests, document requests, and consultations can have different deadlines. Public utilities and communications services also have special rules. The team must classify each PQRS at intake because the deadline comes from that classification.
What happens if a PQRS is incomplete or information is missing?
The process should find missing information as early as possible and ask for it clearly. If the team discovers the issue at the end, it loses days and increases the risk of an incomplete response. Automation helps when it checks minimum fields, attachments, and identification data before sending the case to the responsible team.
How do you prevent users from filing the same PQRS several times?
You need case-status lookup, status messages, and duplicate detection. Many users repeat a PQRS because they don't know whether it was received or when they'll get an answer. A system can compare ID, email, phone, subject, and text to suggest that the case already exists, without blocking a person from filing when appropriate.
Which responses can be automated without failing to answer substantively?
You can automate receipt confirmations, status lookup, guidance on requirements, missing-data requests, and drafts for review. Responses that decide on rights, charges, appeals, or denials need human control. Automation should save time in preparation and traceability, not replace analysis when there is risk.
How do you know if a chatbot reduced PQRS response time?
Compare before and after with the same types of cases. Review total time, overdue cases, repeat requests, escalations, and response quality. If the chatbot answers fast but creates more unhappy users, it didn't reduce real time. Measurement should include a sample reviewed by people who know the process and legal criteria.
Written by
Quarl engineering team
We are an AI engineering team in Medellín. We build and run systems that reach production — automations, assistants and agents — and for three years we ran an assistant for two million users across ten countries.
See how we work →Want to apply this in your company?
In fifteen minutes we tell you whether it is worth it, free and with no pitch. The proposal arrives within 48 hours.