FlowsBook a callSign inCreate free accountSign up
Workflows by industry

How credit teams watch the merchants in a portfolio for closures

A shop that quietly shuts still owes its loan or keeps its card terminal. How lenders and payment firms spot closed merchants in a portfolio within days.

Part of the guide Workflow automation

By the xMap team · · 5 min read

Credit and risk teams watch merchants for closures by looking up every business in the portfolio on a regular timetable, comparing what they find with what they saw last time, and raising a case only when something has changed: the listing now says closed, the phone number or website is different, or the name and address have moved. A closure seen in the first week leaves time to call the owner, pause a credit line or recover equipment. A closure found three months later, when payments stop, usually does not.

This article is for credit risk analysts, collections and portfolio managers at banks, small-business lenders, merchant acquirers and buy-now-pay-later providers. It covers the early signals of a merchant in trouble, how to run the check across thousands of names, a made-up example, the pitfalls, and how an AI employee does the work each week.

Why closures slip through

Most portfolio systems learn about a merchant from its own behaviour: missed repayments, falling card volumes, an unanswered statement. Those signals arrive late and are noisy. A seasonal shop has a quiet month every year; a café that closed for good looks the same as one on holiday for the first few weeks. Meanwhile the outside world often knows sooner. When a business closes, its public listing on online maps is usually updated by the owner, by customers or by the map itself, sometimes within days.

Watching those public listings gives the team a second, independent signal. It does not replace payment data, but it explains it: a drop in card volume with a listing that now says closed is a very different case from a drop with a listing still open and a rating that just jumped.

What to watch for

  • Closed or closed for now. The strongest single signal. Closed for now often precedes closed for good, so treat it as an early warning.
  • A new phone number or website. Can mean new ownership, a rebrand or a move. Worth a call before the next review.
  • A new name at the same address. Often means the business was sold or replaced, and your borrower may no longer be the operator.
  • A rating falling fast. Not a closure, but a run of poor reviews can come before trouble in a consumer business.

How to run the check across a portfolio

The practical method is the same whether the portfolio holds two hundred merchants or twenty thousand.

  1. Keep the merchant list in one place: a table you upload, or a query on your own database, with each merchant's trading name, address, city and, where you have it, position.
  2. Look each merchant up on online maps by name near its address, so each one comes back with its current status, phone, website, rating and address.
  3. Compare with the last check and keep only the merchants whose details changed.
  4. Send the changes to the team that acts on them, with the old and new values side by side.
  5. Write the outcome back to your own system so the next review starts from the latest picture.

A worked example (made up)

Say a made-up small-business lender has 3,400 active merchants in Mexico City and Lima, mostly restaurants, salons and corner shops. Each Monday its flow looks up every merchant. In a typical week, the flow might report something like this:

Change foundMerchantsWhat the team does
Listing now says closed6Call the owner the same day, flag the account
Closed for now9Call within the week, note the reason
New name at the same address3Check who now operates the business
New phone or website11Update contact details, no action on credit
Rating fell by 0.5 or more4Add to the watchlist

A made-up weekly result for an imaginary lender. The counts are illustrative, not measured.

Thirty-three changes out of 3,400 merchants is a list one analyst can work through on a Monday morning. The six closures go to collections first; the rest are spread through the week.

Mistakes to avoid

  • Acting on a single closed flag. Listings can be wrong. Treat closed as a reason to call, not as proof.
  • Checking too rarely. A quarterly review loses most of the head start. Weekly is a good rhythm for consumer-facing merchants.
  • Mixing new merchants into the comparison. The first time a merchant is looked up there is nothing to compare with. Let the first run set the baseline.
  • Sending every change to everyone. Collections needs closures; the data team needs new phone numbers. Route each change to the people who use it.
  • Losing the history. Keep the previous values. A merchant that changed name twice in six months tells its own story.

How to do it with an AI employee

The template Merchants at risk of leaving, every week is built for payment teams, and Clients that closed or changed, for my insolvency team suits recovery work. Built from the blocks, a weekly closure watch looks like this:

  1. Schedule: every Monday at 06:00.
  2. Database query on your own database, or My data with an uploaded table: the active merchants.
  3. Map business scan, around each incoming place, looking for each place itself: current status, phone, website, rating and address.
  4. What changed: Closed, Phone or website, Name or address, and Rating moved by 0.5.
  5. xMap LLM: one line per merchant on what changed and what to do first.
  6. Email to the credit team, and Save to my database to write the changes back.

What lands in the inbox on Monday is a short table of only the merchants that changed, each with a change line such as "Closed; New phone", the previous values, and a suggested first step. A quiet week sends nothing. Teams that want a person to look before any record is updated can add an Approval step before the database write.

Questions

Is a closed listing proof that a merchant has stopped trading?

No. It is a strong public signal, but listings can be out of date or wrong. Use it to prioritise a call, and confirm with the merchant or your own payment data before acting.

How many merchants can be checked in one run?

Map business scan looks up to 100 places itself per run, so a large portfolio is checked in batches, for example by city or by day of the week, with each batch running on its own timetable.

Do we need positions for every merchant?

A name and address are the starting point. A position from your records makes the lookup more reliable, especially for common names.

Can the results go straight into our core system?

Yes. Save to my database writes into a table in your own PostgreSQL, BigQuery or Snowflake database, which your systems can read.

Templates that do this

Read next

More on workflows by industry · Every article · Templates

Put it to work

An AI employee in xMap Flows does this on live data, on its own timetable. Start from a template or describe it in plain words.