StraightLine Insight Note
This month’s Insight Note draws from four StraightLine proof pieces: When the Report Stops Explaining the Number, The Headline Looked Manageable, From Historic Delivery Performance to Operational Risk.
July looked at the point where apparently sound financial and operational figures can still mean different things depending on definition, timing and purpose.
August moved the question one step further:
What happens when the headline looks reasonable, but the detail underneath is trying to tell you something else?
That was the thread running through this month’s work.
· A small percentage difference.
· An average payment time that looks acceptable.
· A delivery KPI that explains last month.
· None of those figures is necessarily wrong.
· But none of them is enough on its own.
· The work begins where confidence stops.
What I noticed this month
A report does not have to contain an obvious error to deserve another question.
Sometimes the first signal is much quieter.
A revenue variance of less than 0.3%.
An average settlement time inside the agreed payment terms.
Five late orders in a relatively small sample.
Individually, each can look manageable.
The temptation is to accept the headline and move on.
But averages, percentages and totals compress detail. That is their job.
The problem begins when the detail being compressed is exactly where the business risk sits.
August’s work kept returning to the same distinction:
Is the report describing the position, or helping somebody understand where they need to act?
Those are not always the same thing.
The August proof theme
August’s work examined that issue through three practical financial-operational questions.
1. A small percentage difference can still be worth explaining
The first case study compared reported revenue and cost with values independently reconstructed from quantity and unit values.
At headline level, the difference looked small:
0.27% on revenue
0.29% on cost
Those percentages could easily be dismissed.
But against the size of the dataset, they represented approximately:
£12.6 million of revenue
£7.9 million of cost
that could not be fully explained from the fields available.
That changed the question.
It was no longer:
“Is the percentage difference material?”
It became:
“Can we explain how the reported number was produced?”
The analysis then moved below the total.
Some product groups reconciled closely.
Others did not.
Cooking Gear showed a more noticeable difference, so the investigation narrowed further to one product, TrailChef Kettle.
That product showed differences across both revenue and cost, with the pattern changing by year and sales channel.
The available data was not enough to prove why.
And that matters.
A responsible analysis should know where evidence stops.
The appropriate next step was not to redesign the dashboard or declare the existing figure wrong.
It was to trace a small number of transactions through the underlying process:
Order → Delivery → Return / Cancellation → Invoice / Credit → Payment
Only then could the business establish whether the issue sat in the report, the operational process or simply in the meaning of the fields.
The lesson was simple:
The report may contain the right number. Trust depends on being able to explain how it got there.
2. An acceptable average can hide repeated behaviour
The receivables case study started with another headline that looked reasonably comfortable.
Invoices took an average of 26.44 days to settle, broadly consistent with 30-day terms.
At first glance, that does not sound like a serious payment problem.
The detail gave a different picture.
35.56% of invoices were paid late, representing almost £54,000 of the £147,700 invoiced.
When invoices were late, they were settled an average of 9.68 days beyond their due date.
The analysis then went below the average and looked at two things:
invoice value;
repeated customer behaviour.
Most late invoices were individually worth less than £100.
That is important because small invoices are easy to deprioritise.
One £60 invoice being a few days late may not deserve much management attention.
But repeated small late payments across the same customer base create a different issue.
Several customers paid more than 80% of their invoices late, with some above 90%.
The analysis also found that 46.02% of late invoices were disputed.
By contrast, only 0.53% of invoices were more than 30 days late, worth £871.61.
So the central problem was not a large population of severely overdue invoices.
It was repeated customer-specific lateness, with disputes forming a significant part of the pattern.
That distinction changes the management response.
A consistently late but undisputed customer may need a conversation about payment terms, account ownership or credit control.
A customer with repeated disputed invoices may point instead towards pricing, invoice accuracy, delivery evidence or slow dispute resolution.
The report became more useful when it stopped describing lateness and started showing where the behaviour repeated.
A small invoice can be easy to overlook.
A repeated pattern of small invoices is a process signal.
3. A dashboard can explain yesterday without protecting tomorrow
The customer-order case study began with a familiar operational KPI.
On-time delivery had fallen from 100% in May to 75% in June and 80% in July.
Warehouse capacity had also fallen.
Staffing pressure was therefore an obvious explanation.
But when the capacity data was added, the relationship was less straightforward.
Average warehouse capacity fell again in July, yet delivery performance improved.
Capacity may have contributed to the problem, but the data did not support treating it as the whole explanation.
The order-level detail raised a different set of questions.
Some orders were:
overdue;
inactive;
on hold for extended periods;
repeatedly affected by credit holds.
Normal-priority work could also sit untouched long enough to become operationally urgent.
That creates an important reporting distinction.
A status or priority field tells you how the order is classified.
It does not necessarily tell you how much operational risk has accumulated around it.
Ageing and inactivity can change the meaning of the original priority.
An order marked Normal today may not remain normal after two weeks without meaningful activity.
This is where a dashboard needs to move beyond activity.
Counts, totals and historic delivery percentages explain what has happened.
Operational leaders also need visibility of:
ageing;
inactivity;
blocked work;
repeated holds;
emerging risk;
items requiring intervention.
The useful management question was therefore not simply:
“Why were deliveries late?”
It was:
“Does the current reporting show managers where intervention is needed before delivery performance deteriorates?”
The report could explain yesterday.
It was less clear whether it could protect tomorrow.
The operational problem underneath
Across all three case studies, the same reporting problem appeared in different forms.
The headline was not obviously wrong.
It was simply incomplete for the decision being made.
A percentage variance could not explain the transaction trail underneath it.
An average settlement time could not show repeated customer behaviour.
A delivery KPI could not show which open orders were quietly becoming risky.
This is why reporting confidence is not only a question of accuracy.
It also depends on traceability, context and actionability.
Traceability
Can the headline be followed back to the transactions and rules that created it?
Context
Can the reader see whether the pattern is widespread, concentrated, repeated or exceptional?
Actionability
Does the report help somebody identify what needs investigating or changing next?
A dashboard can be technically accurate while still being weak on one or more of these.
And when that happens, the workaround usually appears somewhere else.
Someone exports the data.
Someone reconciles two reports.
Someone maintains a separate list.
Someone asks the experienced colleague who knows which orders to worry about.
Someone decides that the variance is “probably fine” because investigating it feels disproportionate.
These behaviours are not always signs of poor discipline.
Sometimes they are signals that the formal report has stopped short of the decision.
That is why a technically correct number can still create an uncertain decision.
The pattern behind the month
Across August’s work, the same sequence appeared:
A headline measure looks broadly reasonable.
The number passes an initial sense check.
The detail is examined.
A pattern appears beneath the average, total or status.
The original explanation becomes less complete.
The investigation narrows rather than expands.
A clearer management question emerges.
That last point matters.
Good analysis does not always produce a more complicated answer.
Often it produces a smaller, better question.
Not:
“Why is the entire revenue report wrong?”
But:
“Why do these transactions not reconcile?”
Not:
“Why are customers paying late?”
But:
“Why do the same customers repeatedly pay late, and how many of those invoices are disputed?”
Not:
“Why is warehouse performance falling?”
But:
“Which orders are ageing or inactive before they become late?”
The purpose of analysis is not to investigate everything.
It is to find where looking more closely is likely to change a decision.
One practical check
Choose one report that your business uses regularly.
It might be:
receivables;
sales;
order book;
stock;
purchasing;
delivery performance;
margin;
production;
cash;
service performance.
Then ask these six questions.
1. Decision
What decision is this report supposed to support?
2. Owner
Who is responsible for acting on what it shows?
3. Definition
What exactly does the headline measure include?
4. Source
Can the number be traced back to the underlying transactions?
5. Timing
At what point is the information complete enough to use?
6. Exceptions
What does the headline hide that might require action?
Then ask one final question:
What does somebody do immediately after receiving the report?
Do they:
export it to Excel;
reconcile it against another report;
check several transactions manually;
ask another department for its figure;
maintain a separate exception list;
rely on experience to decide which items actually matter?
That next action is useful evidence.
It may show you exactly where reporting confidence stops.
The August proof pieces showed the same pattern in different forms:
a small revenue or cost variance became more useful when the analysis narrowed to the product and transaction level;
a manageable average payment time became less reassuring when repeated late-payment behaviour was visible by customer;
a delivery KPI became more useful when ageing, inactivity and blocked work were added to the picture.
The practical lesson is not to investigate everything.
It is to find the point where the headline stops being enough.
What SMEs can do next
The first step does not need to be a new ERP system, a dashboard rebuild or a large data project.
A focused review of one important report may be enough.
That review should ask:
What decision is this report meant to support?
Which headline number matters most?
Can that number be traced back to the transactions behind it?
What exceptions are hidden inside the total or average?
Does the report show repeated behaviour, or only the overall position?
Are ageing, inactivity, disputes or blocked items visible?
Where do users still export, reconcile or check manually?
Which of those checks could change a commercial or operational decision?
The aim is not to produce more reporting.
It is to make the existing reporting easier to explain, easier to trust and easier to act on.
A useful review may end with a surprisingly small next step:
clarify one definition;
trace a handful of transactions;
add one exception view;
separate disputed from undisputed debt;
monitor ageing alongside priority;
make one recurring manual check visible.
Small changes are often enough to show whether the issue sits in the data, the report, the process or the way the business is using the information.
Question for the month
Which report in your business looks reasonable at headline level, but still sends people somewhere else before they are willing to act?
And what are they looking for when they go there?
The StraightLine view
Totals, averages, percentages and status fields are useful.
But they compress detail.
That means they can describe the overall position while hiding the few exceptions, repeated behaviours or process gaps that matter most.
A small variance may deserve attention because of where it is concentrated.
An acceptable average may hide customers who repeatedly pay late.
A delivery percentage may describe performance without showing which orders are quietly becoming risky.
A trustworthy report does not need to show everything.
It does need to make it clear when the headline is no longer enough.
Good reporting does not remove judgement.
It gives judgement somewhere firmer to stand.
How StraightLine Data & Training can help
StraightLine Data & Training helps SMEs look underneath ERP, finance and operational reporting to understand where confidence begins to weaken.
That might involve:
reviewing a report that users still check manually;
tracing headline values back to the underlying transactions;
identifying where reported and recalculated figures diverge;
separating averages from repeated customer or process patterns;
making ageing, inactivity, disputes and exceptions easier to see;
comparing finance and operational reporting logic;
clarifying definitions, timing and ownership;
improving an existing report before deciding whether a wider rebuild is needed.
The starting point can be one report, one business question and one place where somebody still feels the need to check.
If the number looks acceptable but the decision still feels uncertain, that gap is worth understanding.
The work begins where confidence stops.
Comments