Guides
Ways to check the quality of ATM route work
Check ATM route service quality by audit: four measures from your own records, an unannounced customer visit, and the host question that gets a real answer.
What to take away
- Service quality is what the host and the cardholder experienced, not what your procedure says you did.
- Four measures capture almost all of it, and every one of them can be produced from records you already have.
- Audit the machine as a customer, without warning and without your service kit. What you notice is the audit.
- Ask the host the question they will actually answer honestly, which is about the last time rather than about general satisfaction.
- Publish the result to yourself monthly. A measure you do not look at is a measure you do not have.
An operator can follow every step of the daily procedure and still be delivering a poor service, because the procedure measures effort and the host measures outcome. This is an audit rather than a checklist: a way of finding out what your placements are actually experiencing.
The four measures
| Measure | What it counts | Where it comes from | What a bad reading means |
|---|---|---|---|
| Availability | Hours the machine could not complete a withdrawal, for any cause | Fault tickets and visit logs | Report it in hours; a percentage hides a bad weekend |
| Time to know | Hours between the machine going down and you knowing | How you found out: monitoring, a visit, or the host | Anything found by the host is an attention problem |
| Time to restore | Hours between knowing and the machine working | Ticket opened and closed | Track the worst case, because that is what hosts remember |
| Repeat faults | Machines with the same fault twice in a quarter | Fault log | Separates a machine problem from a service problem |
None of these require software you do not have. They require that visit logs and fault tickets are written down at the time.
The unannounced customer visit
Once a quarter, visit a machine as a customer would. No service kit, no keys, no conversation with the manager first.
Unannounced customer visit checks
- Is the machine lit
- Is the screen legible in that light
- Is the fascia clean
- Is receipt paper there
- Is anything stacked against it
- Is fee disclosure clear before committing
- Is the approach obstructed
Stand where a customer stands and look at what they see. Is the machine lit. Is the screen legible in that light. Is the fascia clean, and is the receipt paper there. Is anything stacked against it. Is the fee disclosure clear and readable before the transaction commits. Is the approach obstructed. Would you use this machine at night.
Then complete a withdrawal and keep the receipt. The transaction is the only part of the audit that cannot be argued with.
Write down what you noticed within the hour. What is not written down within the hour becomes an impression, and impressions are what you already had.
The host conversation, asked properly
Do not ask whether the host is happy. Everyone says yes.
Host conversation: real position
Has anyone else approached you about placing a machine here?
, said no -> strong placement
Yes, thinking about it -> act this month
Ask about the last time instead. When was the machine last out, how did you find out, how long was it, who did you call, and what happened?
Those four answers give facts, not sentiment, and you can check them against your own records. A host who has been approached before also weighs the placement agreement terms, not just the machine.
Then ask the question that tells you your real position: has anyone else approached you about placing a machine here. A host who has been approached and said no is a strong placement. A host who has been approached and is thinking about it is telling you something worth acting on this month rather than at the next renewal.
Comparing what the host said with what you recorded
Line up the host's account of the last outage against your fault log. Three outcomes, all useful.
Host account vs fault log
Outcome
- They match
- Records reflect reality
- Host remembers longer
- Communication gap
- No record at all
- Monitoring gap
What it means
- They match
- Host remembers longer
- No record at all
They match, which means your records reflect reality. The host remembers it as longer, which usually means your time to restore was fine but your communication was not. Or you have no record of it at all, which means the machine was out and nobody told you, and that is the finding worth having.
The third outcome is common and it is not a sign of a careless host. It is a sign your monitoring, or your habit of reading it, has a gap.
What to do with the audit
Rank the machines by hours unavailable, then look at the causes rather than the ranking.
Causes and their fixes
Cause
- Cash-outs
- Load or visit interval
- Jams and dispenser faults
- Hardware decision
- Host-reported faults
- Monitoring or attention
- Long restores
- Parts, distance, technician
Fix
- Cash-outs
- Jams and dispenser faults
- Host-reported faults
- Long restores
Cash-outs mean the load or the visit interval is wrong for that site. Jams and dispenser faults concentrated in one machine mean a hardware decision. Faults you learned about from the host mean a monitoring or attention problem. Long restores mean parts, distance, or a technician arrangement that does not fit your route.
Each of those has a different fix, and three of the four are cheaper than they look. What the day's routine should already be catching is set out in the daily procedure, and the rhythm it sits inside is in the operations and workflow guide.
The parts of quality that are not about uptime
Two others matter and neither shows in a fault log.
The first is what a customer's data and your systems are exposed to, and keeping monitoring, reporting and record systems in order is part of the service you sell. The NIST small business quick-start guides give a small operation a place to start.
The second is whether your compliance position is current, because a host inherits your problem if it is not. The register that tracks it is described in the compliance overview.
Your banking relationship reflects the same discipline, since the risk-based review described in the FinCEN statement on independent ATM operators weighs how the operation is run rather than how it is described.
Doing this at scale
At a few machines the audit is an afternoon. Past that, sample rather than survey: audit every machine annually, the worst quartile quarterly, and any machine that has just had a bad month. Sampling also depends on a route schedule that reserves capacity for faults instead of packing every stop.
Sampling at scale
- Audit every machine annually
- Audit worst quartile quarterly
- Audit any machine after a bad month
- Reserve route capacity for faults
The measures stay the same. What changes is that you can no longer hold the answers in your head, which is one of the transitions covered in the material on growing a route.
Common questions
How often should the audit run?
The four measures monthly from records you already keep. The unannounced visit quarterly per site, or on a sample once the route is large.
Should I tell the host I am auditing?
Not before the customer visit, because the point is to see the ordinary state. Tell them afterwards and share what you found, which is a conversation that tends to go well.
What is a good availability figure?
There is no published figure worth trusting for your route. Compare each machine against its own previous quarter, and against the other machines you service. Your own history is the only fair benchmark.
The host says the machine is fine but the numbers are poor. Who is right?
Both, usually. The machine is working and the site is weak. That is a placement question rather than a service question, and it is answered with counts rather than opinions.
