The P&L, line by line

The P&L covers a period, usually a year or a month. It starts with what customers bought, takes away costs in layers, and ends with what was left for the owners. Each layer has a name, and each subtotal divided by revenue gives a margin. Here is Mangrove Market's year to 31 December 2025, in millions of dollars.

Read the four margins from the top down, because each one answers a different question.

• Gross margin, 23.0%. Revenue minus cost of goods sold, over revenue. It tells you how much of each dollar is left after paying for the groceries themselves. Buying terms, prices, waste and the product mix move it. A grocer lives on a thin gross margin.
• EBITDA margin, 4.2%. EBITDA is what is left after the running costs of the business: staff, rent, delivery, technology, marketing and admin. It is the closest P&L line to the cash the operations produce, so lenders and buyers of companies like it. Note that it is not a line that IFRS or US GAAP require; each company defines it in its own reports.
• Operating margin, 2.0%. Take off depreciation and amortisation (D&A), the yearly share of what was spent on stores, fittings, tills and software in earlier years, and you get operating profit, also called EBIT. It answers: does the business earn back its investments? From 2027, the new IFRS 18 standard requires companies that report under IFRS to show an operating profit subtotal in a set way.
• Net margin, 1.2%. After interest on Mangrove's $120 million of borrowings and 25% tax, $21 million is left for the owners. That is about 1.2 cents of profit on every dollar a customer spends.

Two habits make this table useful. First, always look at a line as a percentage of revenue, because that is how you compare years and companies of different size. Second, notice how thin the bottom is. Mangrove's total running costs are $331 million, and the whole technology line is 7% of that. A small change in any cost line moves net profit a lot in percentage terms, which you will use at the end of this lesson.

Accruals: counting things when they happen

The P&L does not follow the bank account. It follows accrual accounting: revenue counts when it is earned, and costs count when they are incurred, whatever day the money moves. For a grocer, revenue is earned when the customer gets the goods. IFRS 15, the revenue standard, puts it as when control of the goods passes to the customer.

Three Mangrove situations show the difference.

• A partner shop pays late. Mangrove delivers a $250 wholesale order to a corner shop on 20 December. The shop pays on 12 January, which fits the average of 23 days that partner shops take (the terms say 14). The $250 is December revenue. On 31 December the balance sheet shows it as a receivable, and the cash arrives in next year's cash flow.
• The cloud bill comes late. The December cloud invoice, about $540,000, arrives in January. Finance still books it as a December cost, with an accrual (a liability) on the balance sheet until it is paid.
• A licence is paid up front. A $120,000 yearly software licence paid in January is a prepayment. It goes on the balance sheet as an asset and becomes $10,000 of cost each month.

Accruals exist so that each month's costs line up with that month's sales. This is why finance asks you at month end what work vendors have done that they have not invoiced yet. If you forget, one month looks too cheap and the next too expensive, and nobody can trust the trend.

The balance sheet: one photo on one day

The balance sheet is a snapshot taken at midnight on one date. It always obeys one equation: assets = liabilities + equity. Everything the company owns was paid for either with someone else's money (liabilities) or with the owners' money, including profits kept in the business (equity). Mangrove's balance sheet on 31 December 2025, in $M, leaves out leases to keep things simple. Under IFRS 16 most store leases would appear as assets and liabilities, and much of the rent line would move to depreciation and interest.

The most interesting number is not on the page. Take the money tied up in trading: inventory 93 plus receivables 6, minus payables 130. That is −$31 million of working capital. Suppliers are funding Mangrove's shelves.

The mechanism is timing. Divide each balance by a day of cost of goods ($1,355M ÷ 365 ≈ $3.7M a day). Stock sits on the shelf for about 25 days (93 ÷ 3.7). Store customers pay at the till the same day, so receivables are tiny. Mangrove pays suppliers after about 35 days (130 ÷ 3.7). Goods are sold and paid for about ten days before the supplier is paid. That pattern is common in grocery.

Two consequences matter for tech teams. When sales grow and the days stay the same, the gap grows, so growth releases cash: 10% more sales makes working capital about −$34M, freeing roughly $3M. And anything that changes the days changes cash. Paying suppliers 10 days faster would cost about $37M of cash once, most of the $45M in the bank. In the other direction, if partner shops paid on their 14-day terms instead of 23 days, about $1.3M would come in (9 days × $52M ÷ 365). A Partner app feature that makes paying invoices easier is a cash project, as well as a customer one.

Cash flow: why profit is not cash

Mangrove made $21M of profit in 2025, but its cash rose by only $2M, from $43M to $45M. The cash flow statement explains the gap. Under IAS 7 (and ASC 230 in US GAAP) it has three sections.

Operating activities                       $M
  Net profit                               21
  + Depreciation and amortisation          38   cost in the P&L, no cash left this year
  + Working capital                         4   payables grew faster than stock
  = Cash from operations                   63

Investing activities
  New stores and DC fit-outs              -40
  Tech hardware (tills, scanners, laptops) -2.5
  Capitalised software                    -1.5
  Other                                    -8
  = Cash used in investing                -52

Financing activities
  Debt repaid                              -6
  Dividends to owners                      -3
  = Cash used in financing                 -9

Net change in cash   63 - 52 - 9 = +2      (cash 43 at the start, 45 at the end)

Read the operating section as a correction of the P&L. It starts from net profit and adds back D&A, because that $38M is a cost in the P&L for cash that left when the assets were bought, mostly in earlier years. Then it adjusts for working capital: in 2025 payables grew a little faster than stock, which brought in $4M. This way of starting from profit is called the indirect method, and most companies use it. From 2027, IFRS 18 makes operating profit the starting line, but the logic is the same.

The investing section is where capex shows up in full, in the year the cash goes out. The P&L will only see it slowly, as depreciation over the coming years. The financing section shows money to and from lenders and owners. Interest paid is inside net profit here, so it sits in operating cash flow, which IAS 7 allows today. IFRS 18 moves interest paid to financing for most companies from 2027.

The three statements are one system. Net profit flows into equity (180 includes this year's 21, minus the 3 paid as dividends) and starts the cash flow. Capex leaves through investing, sits on the balance sheet as an asset, and comes back to the P&L as depreciation. The net change in cash is the change in the balance sheet's cash line. If any of these links fail to add up, a number is wrong.

Where technology shows up

The P&L line called Technology is $24M. Mangrove's real spending on technology in 2025 was about $28M, about 1.6% of revenue, and it is split across all three statements.

• Opex in the P&L, $24M. Opex: tech staff $11.0M, cloud $6.5M, SaaS and software licences $4.0M, agencies and contractors $1.5M (Lagoon Digital about $0.39M of it), network and telecoms $1.0M. The team is about 210 people, and an engineer costs about $5,000 a month fully loaded.
• Capitalised software, $1.5M. Tech people cost about $12.5M in total. About $1.5M of that was time spent building new software that met the rules for capitalisation, so it went to the balance sheet instead of the Technology line. It is amortised over 3 years: about $0.5M a year of cost in the D&A line.
• Tech hardware capex, about $2.5M. Tills, scanners, laptops and network gear go to property and equipment and are depreciated over their useful lives, also inside D&A.
• Cash. The $24M of opex leaves through operating cash flow. The $4M of capex ($2.5M hardware plus $1.5M software) leaves through investing.

So the Technology line understates technology. Part of the cost of every scanner and every capitalised feature sits lower down, in D&A, below EBITDA. That placement has a side effect people notice: capitalising work makes EBITDA look better in the year it is done. Which costs may be capitalised is set by accounting standards (IAS 38 under IFRS, ASC 350-40 under US GAAP) and by the company's own policy, and Budgets, capex and opex goes into the rules. For a real company, the answer always comes from its finance team and auditors.

Now the number that changes how finance hears you. A $100,000 yearly saving in cloud cost adds $100,000 to profit before tax, and $75,000 to net profit after 25% tax. To earn the same $75,000 of net profit from extra sales at a 1.2% net margin, Mangrove would need about $6.3M of extra revenue (75,000 ÷ 1.19%). You get the same answer from the pre-tax side: 100,000 ÷ 1.6% pre-tax margin. The comparison is rough. Extra sales usually earn more than the average margin, because rent and head office do not grow with every order, and unit economics is the right tool to measure that. The direction is still right: at thin margins, a dollar of cost saved is worth many dollars of sales.