Read the Reporting page

What every number and chart on Reporting measures, and how to read it, using a real week of data.

Select the bar chart icon in the left rail to open Reporting. The page has four sections, and each answers one question.

Section	The question it answers
1 Volume & Coverage	Are we processing the volume we said we would?
2 Stage Health	Where are invoices stuck right now?
3 Quality & Accuracy	Is the AI doing its job?
4 Efficiency & SLA	Are we faster than the 6-minute manual baseline?

The screenshots show one week of data from before go-live. The examples in the tables use its numbers.

Set the range and entity

Two controls at the top apply to the whole page.

• RANGE: Today, 7d, 30d, or Custom dates.
• ENTITY: All, or one entity: SG, MY, HK, ID, PH or other.

Cards marked vs prior period compare the range you picked with the period of the same length just before it. For example, 7d compares the last seven days with the seven days before. A change for the better shows in green, and one for the worse in red.

Reading p50 and p95. Time metrics are shown as percentiles, not averages. p50, the median, is the middle invoice: half took less time, half took more. p95 shows the slow end: 95 in every 100 invoices took less time. A low p50 with a high p95 means most invoices are quick but a few take much longer.

A week of data: 45 invoices created and 18 posted. 18 against the monthly target of 4,500 is a coverage of 0.40%.

1RANGE

2ENTITY

1 Volume & Coverage

Metric	What it measures	How to read it
Invoices posted	Invoices posted to SAP in the range.	The team’s output. The example shows 18, 28% fewer than the week before.
Invoices created	Invoices that entered Neoflo in the range.	Incoming demand. If it keeps beating Invoices posted, the queue is growing.
Coverage	Invoices posted, divided by the monthly target of 4,500.	How far you are toward the month’s target. A short range always reads low: a week can cover only about a quarter of a month.
Avg SLA (p50)	The median time from Review being selected to the invoice posting in SAP.	The card says Avg, but it shows the median. Lower is better.
Invoices posted, daily trend	Invoices created (dashed line) and posted (solid line) each day. The card’s target is 150 posted a day.	When the solid line stays under the dashed one, invoices are arriving faster than they’re posted.
Posted by entity	Each entity’s share of the invoices posted, with the count.	Where the volume comes from. In the example, MY posted 12 of the 18.
Funnel, count at each stage	Where the invoices from the range are now: Extraction, Faktur Pajak, Matching, ERP Posting or Posted.	Invoices piling up at one stage show where work is waiting. In the example, 18 are posted and 1 is waiting at ERP Posting.
Drop-off by reason	Invoices that left without posting, by the rejection reason chosen. The five reasons are listed in Reject or resend an invoice.	The same rejections as Rejection split, counted instead of shown as shares. The biggest reason is the first to fix.

The same week. The invoice waiting at ERP Posting in the funnel is the one Stage Health shows stuck, in MY.

1Funnel

2Drop-off by reason

3One invoice stuck in ERP posting, in MY

2 Stage Health

Stage Health is marked LIVE. It shows invoices that are stuck right now, counted by entity. It’s the place to start each morning.

Metric	What it measures	How to read it
Stuck in extraction	Invoices in extraction for more than 5 minutes.	Reading an invoice takes a minute or two. Anything here may need the Neoflo team.
Stuck in matching	Invoices in matching for more than 15 minutes. The card says Summary screen, which is the Matching screen.	Usually an invoice waiting on a person, or on a GR.
Stuck in ERP posting	Invoices still in ERP Posting more than 2 minutes after matching was confirmed.	Usually tax codes not set, or a posting error.

To work through stuck invoices, open the Invoice Dashboard and filter by Status.

3 Quality & Accuracy

Metric	What it measures	How to read it
Rejection split	Rejections made by the team, as shares of each reason chosen.	In the example, 13 of the 16 rejections are Others, so the chart can’t point to a cause. Choosing the most specific reason when you reject fixes this.

Rejection split, and the first two Efficiency & SLA cards, for the same week.

1Rejection split

2End-to-end SLA

3Review-to-posted time

4p95

4 Efficiency & SLA

Metric	What it measures	How to read it
End-to-end SLA, percentiles	The median time from Review being selected to posting in SAP, against the manual baseline of 6 minutes an invoice.	Under 6 minutes means faster than by hand. The example’s 2m 29s matches Review-to-posted p50, because both cards measure the same span.
Review-to-posted time, percentiles	The same span, at p50, p75, p90 and p95. The card calls this the contractual user-side window.	In the example, p50 is 2m 29s and p95 is 1,496 minutes. Most invoices were quick, and a few sat for a day or more between Review and posting.
SLA trend (p50 vs. p95)	p50 and p95 over time, on one chart.	p50 tracks the typical invoice. p95 tracks the slow ones. A spike in p95 points to a day when some invoices were left open.
Per-agent throughput	Invoices posted by each person, colored by their median review-to-posted time: green 3 minutes or less, amber 5 minutes or less, red over 5 minutes.	Shows who might need help, and who has capacity. In the example, one agent posted 12 invoices with a median under 3 minutes, and another posted 4 with a median over 5.
Auto-confirm rate by stage	For each stage, the share of invoices where Neoflo’s output was accepted with no changes: Extraction, Faktur Pajak, Matching and ERP Posting.	A low stage is where people correct Neoflo most. In the example, about half the extractions and over a third of the matches needed no change. Memory raises this over time.
Full auto-confirm (STP-eligible)	The share of invoices accepted with no changes at every stage.	These invoices would have posted on their own with straight-through processing on. In the example, 0 of 18. Straight-through processing is off at go-live, so read this as a measure of readiness.

The rest of Efficiency & SLA. Agents' names are hidden in this guide.

1SLA trend

2Per-agent throughput

3Auto-confirm rate by stage

4Full auto-confirm