Skip to content
Book a demo

Proof and case studies

One operator. 10.2% found. Every number has a query.

These figures come from a single-site hospitality operator, read from a copy of production on 23 Oct 2026. We show what’s measured, what’s modelled and what this customer’s data can’t show yet.

MeasuredModelledNot counted
Your booksXero · Sage · QuickBooksApprovedCodedEvidenced
Bills enter a funnel of 5 rings around the SHOW THE ROWS stamp: Read (Read, classified and extracted, however it arrived: email, upload, photo or folder.), Identify (Is this a real, verified supplier? Is it the one we think it is?), Match (Does it match the order, the delivery and the agreed price?), Check (Have we seen it before? Is the VAT rate right? Do the lines and dates add up?), Approve (Approved by the right person, under the rule on the screen.). A bill that fails is held on its ring with a named reason and one fix. Only bills that pass every ring drop into your books: approved, coded, evidenced.
What was read

Everything that came through the door.

9 monthsof documents read, January to October 2026
99.6%of purchase lines read, normalised and put in a category
115categories in use, each mapped to a GL code
1 unitwith invoice capture only: no orders, bank matching or approval rules yet
What the controls did

Held for a reason, paid with evidence.

100%of paid invoices had payment evidence on record
7%of invoices were held for a named reason
45.6%of invoices needed a human fix before they were fit for the books
What the agents are holding open

394 findings, each waiting for a person.

Distinct findings, counted once each, still waiting for the customer to act. The chart shows each class’s share of the money at stake. 14 supplier-behaviour findings carry no £ and aren’t charted.

415 findings already actioned by the customer, including VAT rate mismatches worth 4.0% of spend seen and a duplicate invoice.

Fraud and duplicates13 findings
31.2%
Anomalies (matching, statements)129 findings
23.3%
Supplier risk11 findings
20.2%
VAT and tax63 findings
18.0%
Spend signals3 findings
4.5%
Price variance159 findings
2.7%
Extraction mismatch2 findings
0.2%

Share of £ at stake, by class · 23 Oct 2026

Day zero to today

96% hidden to 99% seen.

At day zero, 96% of this operator's spend was a total on a bill, or a total split only for VAT: no products, no category, nothing anyone could check. Today 99% of the spend OmniPATH reads sits in a line, in one of 115 categories. That's what makes price drift, VAT, duplicates and fraud visible in the first place.

Day zero · sealed 10 Sep 2026
96.3%Hidden
Day-zero spend records, sized by share of spend
  • No lines
  • Split for VAT only
  • Undetermined
  • Full detail
  • 0%of spend records had a category
  • 1.4%of spend had full line detail
  • 62.9%of the year's ledger was a bill total with no lines
Today · read 23 Oct 2026
99.1%Seen
115 categories in use
  • 99.6%of lines read sit in a category
  • 115categories in use, each mapped to a GL code
  • 0.4%of lines still waiting for a category

Your accounting system still holds last year as totals. From go-live, every bill OmniPATH approves goes back to it coded. How the gate works

Different bases, both shown as they are: day zero is the profiled ledger spend over the 12 months to 10 Sep 2026; today is the line spend OmniPATH has read since. Hidden means a bill total, or lines split only by VAT rate. How we measured this

Value found

10.2% of spend seen came back as value found.

Measured against the sealed day-zero baseline. 7.6% comes straight from the records; the rest is time freed, modelled at the declared rates. What-ifs, VAT at stake and spend flagged for investigation are shown separately and never added in.

Overpaid4.1%of spend seen · Price drift and price spikes against what was agreed and paid before
Negotiable3.5%of spend seen · Volume tiers and genuine price swings worth renegotiating
Time freed2.7%of spend seen · Hours not spent keying and fixing invoices, at £24.63 an hour (modelled)
The dimensions carrying the value
Price drift68 products drifting upward; annualised excess cost
3.24%
Bulk pricing181 products ordered often enough to negotiate volume tiers
2.27%
Time savingsHours not spent keying documents, at £24.63 an hour
1.39%
Exception handling45.6% of invoices needed a human touch, 25 minutes each
1.28%
Price negotiation13 products with genuine price swings, pack-size errors removed
1.21%
Late-payment avoidanceInvoices paid late, by 104 days on average; a what-if, not counted
1.00%
Price spikes50 spikes detected, 49 still active
0.83%
Payment timingInvoices paid early, by 99 days on average; a what-if, not counted
0.29%
Found in the records (price drift is annualised, as the module serves it)Modelled at declared ratesWhat-if, not counted

Declared assumptions: £24.63 an hour · 12 minutes per document handled manually · 25 minutes per exception · 12.5% late-payment penalty rate · 4.5% payment-timing rate

Outside value found

At stake, flagged or speculative. Never added in.

Three figures sit beside value found because a person has to act or decide before any of them is worth a pound.

At stakePotential overspend from a wrong VAT rate8.7%

of spend seen is VAT at stake on lines where the rate didn't match what was bought. Keyed as it stood, that's VAT paid and never reclaimed, or reclaimed and owed back to HMRC.

4.0% already actioned · 4.6% still open
Flagged for investigationPayments that look like duplicates or fraud33

payments the agents want a person to investigate: the same amount paid twice, a changed bank detail, an invoice raised out of hours. None of it counts until someone confirms it.

What-if, not countedTiming exposure1.3%

of spend seen, if every late supplier charged statutory interest and paying early cost 4.5% a year.

  • Late-payment interest1.0%Statutory interest a supplier could charge on invoices paid late. Few ever do.
  • Paying too early0.3%The cost of cash on invoices paid before they were due, at 4.5%.
Clean books

What the books looked like before the gate.

Day zero is sealed from the accounting system before OmniPATH changes anything. This is what one year of ledger data looked like going in.

What the gate found at day zeroMeasured
Duplicate supplier records23%

of profiled spend belonged to suppliers keyed under more than one name: one retailer four ways, a food supplier twice, another with a typo.

Dormant supplier contacts9 in 10

supplier contacts in the accounting system carried no spend in the sealed year. Every dormant record is a door left open.

Spend with no lines62.9%

of a year's ledger spend was a total on a bill. The largest supplier, nearly a third of profiled spend, had no lines at all.

VAT at stake8.7%

of spend seen is VAT at stake on rate mismatches. The customer has already actioned 4.0%; 4.6% is still open.

Future-dated bills1.9%

of the year's bills carried a date still in the future when the baseline was sealed. Each one pulls spend into the wrong period.

Off-contract top-ups11%

of profiled spend went to supermarkets, outside any agreed price or catalogue.

From the sealed day-zero epoch of one live single-site hospitality operator (sealed 10 Sep 2026, covering the previous 12 months) and the findings read on 23 Oct 2026. Supplier groupings are our reading of the sealed spend records. How we measured this

With one hand tied

This is invoice capture alone.

Several parts of the platform aren’t switched on for this customer yet. Each one would open up more of the 23 dimensions.

  • Off
    Bank payment matchingBank lines reconciled, and spend with no invoice found
  • Off
    Purchase ordersThree-way match and maverick spend
  • Off
    Approval rulesServed approval chains and straight-through approval
  • Off
    Delivery schedulesThe Deliveries agent, cut-off and delivery reliability
  • Off
    VAT reclaim dimensionActioned VAT findings (4.0% of spend seen) counted as value; the dimension failed on this read
What this doesn't show yet

The parts we won’t claim.

A claim we can’t defend on a live call doesn’t go on this page.

Bank matching is switched off

A year of bank lines is loaded but none are matched yet, so we don't claim reconciliation for this customer.

Credits and deliveries had nothing to work on

One goods receipt and no expected credits are recorded, so those agents have no results to report here.

Forecasts are projections

Spend and savings forecasts are labelled as modelled and kept out of every total. Confidence on this series is too low to quote.

Exposure is not saving

83 actioned supplier-risk findings covered invoices at risk. That's exposure, and it's never added to value.

33 possible duplicate or fraudulent payments are waiting for a person to investigate. They aren't counted anywhere until someone confirms them.

Read on 23 Oct 2026 from a same-day copy of the production database. The organisation is anonymised. Findings are counted once per fingerprint. Every figure has a named query, and the queries are available on request.

Modelled at your size

What the same rates would find for you.

We took the live operator’s value per pound of spend and applied it to your size. It’s a model, not a promise. It leaves out what-ifs, VAT at stake and spend flagged for investigation. It also leaves out multi-unit gains like comparing prices across units, because we haven’t measured them on a live customer yet. A unit is whatever you run: a site, a hotel, an office or a team.

Modelled
£50k£1m per unit
Units
30
Annual spend
£6.00m
Invoices a year
20,928
Value found a year£658.2k
  • OverpaidPrice drift and price spikes against what was paid before
    £244.4k
  • NegotiableVolume tiers and price swings worth renegotiating
    £209.1k
  • Time freedKeying and fixing invoices, at £24.63 an hour
    £204.7k
Published plan cost a year £101.5kFound in the records alone covers it 4.5×

Modelled from one live operator's rates per pound of spend seen (read 23 Oct 2026), scaled by spend and by that operator's invoice rate of about 3.5 invoices per £1,000 of spend. A unit is whatever you run: a site, a hotel, an office or a team. What-ifs (late-payment interest, paying early), VAT at stake and spend flagged for investigation are left out, and so are multi-unit gains such as comparing prices across units, which aren't measured on a live customer yet. Plan cost uses published monthly rates. See the live numbers

Ready when you are

Want these numbers for your own spend?

Connect your accounting system and we'll seal your day-zero baseline in the first session.

No setup fees · No minimum term · UK-based support