Monitor international rates: what to automate, what to keep human, and the data it needs

Software can fetch rates from approved sources, load them into the ledger, flag unusual moves and revalue open foreign balances. People must choose sources, approve exceptions and judge exposure. Automation only holds once rate data has a single owner, a fixed cut-off time and consistent currency codes.

What the work actually involves

On paper this looks like reading a screen and copying numbers. In practice it is a daily chain. Someone pulls exchange and interest rates for every currency the group touches. Those figures then feed accounting, treasury forecasts and intercompany settlement. At period end, a chosen set becomes the official closing and average rates used in consolidation. A wrong figure here travels quietly into every subsidiary's translated results.

The process also watches the balances that rates act on. Foreign currency receivables, payables and intercompany positions all move when a rate moves. Monitoring is only useful if someone connects the rate change to the exposure it affects.

Steps software can take over now

Retrieval is the easiest win. A scheduled feed from a market data provider or a central bank publishes rates at a known time each day. Pulling them by integration removes keying errors and the habit of someone grabbing a figure from whatever website loads first.

Loading into the ledger and treasury system follows naturally. Once the feed lands, a job can write rates to the rate table with the correct date and rate type.

Tolerance checks suit automation well. Set an acceptable movement band per currency pair and let the system compare today's figure with yesterday's and with a second source. Anything outside the band goes to a queue. This mirrors how obligation and expenditure amounts are checked against agreed tolerances elsewhere in finance: the machine confirms the difference is small, and a human only sees the cases where it is not.

Revaluation of open items can also run unattended. Foreign currency receivables and intercompany balances can be restated at the new rate, with gains and losses posted to the right accounts. Ageing reports on those receivables can refresh on the same run.

AI adds value at the edges. It can summarise why a currency moved, draft a commentary for the treasury pack, or spot that a feed has silently stopped updating a thinly traded pair.

Steps that still need a person

Choosing the rate source is a policy decision. Auditors, tax authorities and lenders may each expect a particular publisher. That choice belongs to the finance lead and should be written down.

Approving exceptions needs judgment. A large overnight move may be real, as after a devaluation, or it may be a feed error. Only someone who follows the market and knows the business can tell which.

Sign-off on closing and average rates for consolidation should carry a named approver. Restating a closed period because of a bad rate is painful, so the final set deserves human eyes.

Interpreting exposure is also human work. Deciding whether a movement warrants a hedge, a conversation with a subsidiary, or a change in intercompany settlement timing involves trade-offs no rule captures. The same applies to chasing a foreign debtor whose balance has grown in local terms purely because the currency weakened.

What has to be true about the data first

Every currency needs one code standard across ledger, treasury and banking systems. Mismatched codes for the same currency break automated loads in ways that are hard to see.

Rate types must be defined and labelled: spot, closing, average, budget. If the system holds a figure without saying which kind it is, automation will apply the wrong one.

The cut-off time and time zone have to be agreed. A rate captured in the morning in one region and the afternoon in another produces inconsistent translations across entities.

Direction matters too. Some sources quote units of foreign currency per home unit, others the reverse. The system should store one convention and convert on entry.

The balances being revalued must be clean before any of this helps. Open receivables and intercompany items should reconcile between counterparties, or revaluation will simply restate errors at a new rate.

Questions to ask the people who run it

• Where do the rates really come from on a busy morning, and does that match the documented source?
• Which currencies get entered by hand, and why were they never added to the feed?
• What happens when the feed is late or missing a pair?
• Who decides that an odd-looking rate is acceptable, and is that decision recorded anywhere?
• Are any rates overridden after loading, perhaps at a subsidiary's request?
• How are average rates calculated, and has the method changed without anyone updating the policy?
• Which spreadsheets sit between the rate table and the consolidation, and who maintains them?
• When intercompany balances fail to agree after revaluation, who notices first, and how?

The answers usually reveal manual workarounds that any automation design must either absorb or retire deliberately.

APQC's Process Classification Framework® (PCF) is an open standard developed by APQC, a nonprofit that promotes benchmarking and best practices worldwide. To download the full PCF or to view definitions and measures, please visit www.apqc.org/pcf.