Why the weekly report eats a day
Every Monday, someone at head office opens the payment provider, the till exports, the rota system and whatever app the sites use for opening checks. They copy last week's numbers into a master spreadsheet, fix the site names that never quite match, work out royalties, then cut the sheet into one tab per franchisee and email them out. By the time a franchisee reads it, the week it describes is eight days old.
That is the pattern IVP.ai saw at a UK fast food franchise chain. Collating and sending the weekly financial reports took two members of staff a full day each, every week, and the manual handling added errors and delay. The published case study describes what changed: the reports now build themselves, franchisees have dashboards of their own commercial performance, and head office has clearer sight of brand safety and compliance risk across the network.
This page sets out the structure of that kind of report: a network roll-up for head office and a one-page report per site for each franchisee. The worked example and the asset use invented figures; the case study publishes no site numbers, and none are used here.
Two layers: the site page and the network roll-up
A franchise performance report has two audiences with different questions.
- The franchisee wants to know how their site did against its own recent weeks, and whether it looks normal next to the rest of the network. They should not see another franchisee's takings.
- Head office wants the total, the royalty and levy due, and a short list of sites that need a call this week. It needs every site on one screen, ranked by what is wrong rather than alphabetically.
Both are built from the same weekly table, one row per site. The site page is a filter of that table plus network medians. The roll-up is the whole table with totals and flags. If the two are produced separately, they drift apart, and a franchisee will eventually ring up to ask why their figure differs from the one the area manager quoted.
The data model
Sources
| Source | What it contributes | Grain |
|---|---|---|
| Card payments (Stripe in the case study) | Card takings, refunds, fees | Per transaction |
| Till or POS export | Cash takings, transaction count, the till's own card total | Per site per day |
| Rota or payroll | Hours and labour cost | Per shift or per site per week |
| Checklist app or forms | Scheduled and completed opening, closing and food safety checks | Per check |
| Site master (usually a spreadsheet) | Site ID, town, franchisee company, region, royalty and levy rates, open date | Per site |
The site master is the least glamorous file in the stack and the one everything joins through. Each source system has its own name for a site, so the master holds a column for each: the payment provider's site key, the till's store number, the rota location name and the checklist venue. When a new site opens, adding one row there is what makes it appear in the report.
Getting card sales by site out of Stripe
Stripe's balance transaction object records each movement in the account with an amount (gross), a fee, a net figure (amount minus fee), a reporting_category and a source pointing at the object that caused it, such as a charge or a refund. It does not say which site took the payment.
For that you need metadata: key-value pairs you attach to objects yourself. Stripe's documentation notes that metadata is not copied to related objects automatically, with some exceptions; one is that a PaymentIntent's metadata is copied to the Charge it creates, as a one-off snapshot. In practice, if each site's checkout sets a site_id on the payment, you can follow balance transaction to charge and read the site from there. If sites share one Stripe account and nobody set that key, card sales cannot be split by site from Stripe alone, and the till's card total becomes the primary figure instead.
The joins
| From | To | Join key | Notes |
|---|---|---|---|
| Stripe balance transaction | Charge | source = charge ID |
Gives access to the charge's metadata |
| Charge | Site master | metadata.site_id = site ID |
Only works if the checkout sets it |
| Till export | Site master | store number | One mapping column per system |
| Rota | Site master | location name | Names drift; keep the mapping, not a lookup on text |
| Checklist results | Site master | venue ID | Count scheduled and completed separately |
| This week | Prior four weeks | site ID + week ending | Same site, same weekday mix |
The formulas
| Measure | Formula |
|---|---|
| Card net sales | Stripe card charges − refunds (before processing fees) |
| Net sales | card net sales + till cash sales |
| Prior 4-week average | sum of last four weeks' net sales ÷ 4 |
| Change on 4-week average | (net sales − prior 4-week average) ÷ prior 4-week average |
| Average transaction value (ATV) | net sales ÷ transactions |
| Labour % | labour cost ÷ net sales |
| Checks completed % | checks completed ÷ checks scheduled |
| Till vs Stripe gap | (till card total − Stripe card net sales) ÷ Stripe card net sales |
| Royalty due | net sales × royalty rate in the franchise agreement |
| Marketing levy | net sales × levy rate |
Two choices in that table matter more than they look. First, comparing against the site's own four-week average rather than the same week last year keeps new sites in the report from their fifth week, and avoids comparing a week with a bank holiday to one without. Second, Stripe fees are left out of net sales because many franchise agreements set the royalty base as sales to customers, not what the franchisee receives after card costs. Check your own agreement's definition of the royalty base before copying either rule.
Leading and lagging measures
The case study mentions the chain identifying leading and lagging indicators of its key performance metrics. In a weekly report the distinction is practical. Net sales and labour % are lagging: by the time they move, the week is over. Missed checks, a rising refund rate or a growing gap between the till and the payment provider tend to show up first, and they are the things an area manager can act on in a Monday phone call. Put them next to the sales figure on every site page, not in a separate compliance report.
Week ending 1 November 2026
Network roll-up · Pellowby Grill (fictional)
£156,961 across 8 sites,
4 need a conversation.
Net sales from Stripe and till cash, labour from the rota, checks from the daily checklist app. Each franchisee gets their own site page; head office gets this view.
- Network net sales
- £156,961
- +1.2% on 4-week average
- Royalty due
- £9,418
- Plus £3,139 marketing levy
- Labour, network
- 25.8%
- Median site 25.3%
- Checks completed
- 97.3%
- 654 of 672
Sites, week on 4-week average
| Site | Net sales | vs 4-wk | ATV | Labour | Checks | Flags |
|---|---|---|---|---|---|---|
| PB-01 SwindonKelbrook Foods Ltd | £15,844 | +4.7% | £8.70 | 25.6% | 97.6% | None |
| PB-02 LincolnArlow & Daughters Ltd | £12,535 | -12.0% | £9.68 | 31.0% | 91.7% | Sales below 4-week averageLabour above 28%Checks missed |
| PB-03 IpswichKelbrook Foods Ltd | £23,656 | +2.6% | £8.78 | 24.3% | 97.6% | None |
| PB-04 WrexhamTeague Hospitality Ltd | £17,464 | 0.0% | £9.42 | 24.3% | 100.0% | None |
| PB-05 ExeterSandry Eats Ltd | £20,673 | +5.2% | £8.83 | 23.8% | 100.0% | None |
| PB-06 CarlisleTeague Hospitality Ltd | £24,546 | +2.0% | £9.74 | 28.5% | 98.8% | Labour above 28% |
| PB-07 DundeeMorrow Street Food Ltd | £24,298 | +1.0% | £8.29 | 25.0% | 92.9% | Checks missed |
| PB-08 PeterboroughArlow & Daughters Ltd | £17,945 | +3.0% | £9.40 | 25.5% | 100.0% | Till and Stripe disagree |
Watch thresholds: below -8% on the 4-week average, labour above 28%, checks below 95%, till and Stripe card totals more than 1% apart.
Site report · sent to Arlow & Daughters Ltd
PB-02 Lincoln: net sales, last five weeks
Dashed line: prior 4-week average, £14,244.
| Measure | This site | Network |
|---|---|---|
| Net sales | £12,535 | £156,961 total |
| vs prior 4-week average | -12.0% | +1.2% total |
| Transactions | 1,295 | 17,365 total |
| Average transaction value | £9.68 | £9.10 |
| Labour % of net sales | 31.0% | 25.3% |
| Checks completed | 77 of 84 (91.7%) | 97.3% total |
| Refunds % of card sales | 0.7% | 0.9% |
| Till vs Stripe card gap | 0.4% | 1 site over 1% |
| Royalty + levy due | £1,003 | £12,557 total |
Network column: total for money, median for ratios except where marked.
Weekly build · run log
- Mon 05:30Load Stripe balance transactions, week to Sun 1 Nov8 sites by charge metadataSynced
- Mon 05:31Load till exports, rota hours and checklist results3 sources · 8 sitesSynced
- Mon 05:32Reconcile till card totals to Stripe1 site over 1%For review
- Mon 05:33Build network roll-up and site reports£156,961 · 4 sites flaggedDone
- Mon 05:35Send each franchisee their own site report5 franchisees · 8 reportsDelivered
- Mon 05:35Send network roll-up to head officeOps and financeDelivered
- Mon 05:36Raise royalty invoicesOff in this set-up: finance decidesSkipped
Worked example (illustrative, fictional)
Pellowby Grill is an invented eight-site franchise. Its site PB-02 in Lincoln, run by the fictional Arlow & Daughters Ltd, had a poor week ending Sunday 1 November 2026. The example rules are a 6% royalty, a 2% marketing levy, and watch thresholds of labour above 28%, checks below 95% and sales more than 8% below the four-week average.
| Input | Value |
|---|---|
| Stripe card charges (gross) | £11,372 |
| Stripe refunds | £85 |
| Till cash sales | £1,248 |
| Till card total | £11,332 |
| Transactions | 1,295 |
| Labour cost (rota) | £3,886 |
| Checks scheduled / completed | 84 / 77 |
| Prior four weeks' net sales | £14,187, £14,804, £13,902, £14,083 |
Step by step:
- Card net sales: £11,372 − £85 = £11,287.
- Net sales: £11,287 + £1,248 = £12,535.
- Prior 4-week average: (14,187 + 14,804 + 13,902 + 14,083) ÷ 4 = 56,976 ÷ 4 = £14,244.
- Change on the average: (12,535 − 14,244) ÷ 14,244 = −1,709 ÷ 14,244 = −12.0%. Flagged.
- ATV: 12,535 ÷ 1,295 = £9.68, against a network median of £9.10.
- Labour %: 3,886 ÷ 12,535 = 31.0%, against a network median of 25.3%. Flagged. At 28% labour would have cost £3,509.80, so the site spent £376.20 more than the threshold allows.
- Checks completed: 77 ÷ 84 = 91.7%. Flagged.
- Till vs Stripe gap: (11,332 − 11,287) ÷ 11,287 = 45 ÷ 11,287 = 0.4%. Within the 1% tolerance.
- Refunds: 85 ÷ 11,372 = 0.7% of card sales.
- Royalty: 12,535 × 6% = £752.10, rounded to £752. Levy: 12,535 × 2% = £250.70, rounded to £251.
Lincoln contributed 8.0% of the network's £156,961 net sales that week. The headline for the franchisee is not "sales are down"; it is that sales fell 12% while labour stayed at its usual hours and seven checks were missed. That points to a staffing or management issue rather than a market one, and it is a different conversation from the one about Peterborough, where the only flag is a 2.1% gap between till and Stripe card totals (£336) that finance should look at before the royalty invoice goes out.
Beyond fast food
The structure carries over to other franchise models; only some of the measures change.
| Franchise type | Swap in | Keep |
|---|---|---|
| Retail | Units per transaction, stock shrinkage | Net sales, ATV, labour %, royalty |
| Personal or home services | Bookings, cancellations, revenue per technician | Net sales, labour %, royalty, checks |
| Field service franchises | Jobs completed, utilisation, planned maintenance compliance | Net sales, royalty, site master |
Field service franchises often run a job management system such as Uptick or Simpro in each territory. The per-site page then borrows measures from the engineer utilisation template and the PPM compliance report. More report structures are on the field service reporting hub.
How to automate it with IVP.ai
The IVP.ai Data Layer pulls payments, till exports, rota and checklist data into one governed model each night, holds the site master as the mapping between systems, and calculates the table above. Anything with a REST API, a SQL database or an Excel export can be a source, which matters in franchising because sites rarely all use the same tills.
Deskless by IVP.ai then builds and sends the reports on Monday morning: each franchisee gets their own site page and network medians, head office gets the roll-up with flagged sites first, and any till and Stripe gap over tolerance goes to finance for review. In this set-up Deskless does not raise royalty invoices. It can be configured to, but here finance checks the figures and raises them.
To see the report on your own sites, start with the free 48-hour pilot. We connect up to three sources, such as your payment provider, till export and site master, clean the data and set one report or agent running within 48 hours of read-only access. You then have 14 days of access to choose a plan or walk away.