Microsoft’s profits grew by 32% this quarter. Or 18%, depending on which measure you use. Unilever reported healthy growth while sales in its developed markets went backwards. Neither is a contradiction, but both are rather more interesting than the headline—and this week offered plenty of reasons to read the small print.

So, let's jump in.

The week at a glance

Microsoft: the growth rate has a backstory

Wednesday’s results put Microsoft’s quarterly net profit growth at 32%. Strip out the impact of its OpenAI investments, as Microsoft’s adjusted measure does, and growth was 18%. Last year’s comparison included a $3.1 billion OpenAI-related loss; this year’s included a small gain. Part of the acceleration therefore came from no longer having last year’s accounting headwind.

Eighteen per cent is hardly a distress signal. But the gap illustrates something useful about results: growth measures the distance from a starting point, and an unusually low starting point can make the journey look spectacular. “Adjusted” figures deserve scrutiny, but dismissing them automatically can be as misleading as accepting the largest number. Here, the adjustment makes the story less flattering—and more informative.

Unilever: the average customer doesn’t exist

Unilever’s 5.2% underlying sales growth on Wednesday concealed a much sharper split: emerging markets grew 9.5%, while developed markets declined 0.9%. Underlying volumes rose 6% in emerging markets but fell 1.7% in developed ones. Those are meaningfully different commercial environments inside the same results announcement.

So a verdict that “the consumer is resilient” needs an address. The geographical mix can support the group while concealing pressure in markets more familiar to British readers. Diversification is doing useful work here, but it also makes the headline a poor description of any individual customer’s circumstances. The average looks comfortable; unfortunately, it doesn’t do the shopping.

Shell: the danger of getting used to exceptional

Shell reported $10.8 billion of third-quarter adjusted earnings on Thursday, alongside a planned $5 billion share buyback. Its release cited record refining margins—the profitability of converting crude into fuels—as well as elevated commodity prices and operational performance.

“Record” is attractive in a results headline and awkward in a forecast. If unusually favourable conditions are doing some of the work, projecting that quarter indefinitely would build the exception into the baseline. The buyback returns capital by purchasing shares, but its size alone cannot tell us whether the price paid is sensible. A company buying its own shares still has to think like a buyer.

The Budget: “temporary” belongs in the headline

Wednesday’s Budget proposed increasing the bank corporation tax surcharge from 3% to 6%, running from January 2027 to March 2028. That doubles the surcharge, not the entire corporation tax bill, and the announced increase lasts 15 months.

The duration changes the economics. A temporary charge reduces profits over a defined period; a permanent one also changes assumptions about the years beyond it. Neither makes the extra bill welcome. But treating them as equivalent would be like valuing a business on the assumption that this year’s office refurbishment happens annually. There are expensive tastes, and then there is that.

One story worth understanding: the baseline is part of the story

The thread connecting these announcements is how much depends on the comparison. Microsoft’s growth looks different once last year’s investment loss is removed. Unilever’s group average combines expansion and contraction. Shell’s record margins raise the question of what a more ordinary quarter might earn.

This is where business reporting becomes more useful than a scoreboard. Before deciding whether a number is impressive, we need to know what it is being compared with—and whether that comparison is fair.

My strongest impression this week is that the qualifications are often where the story starts. They reveal why two people can read the same results and reach different conclusions without either being innumerate. One is extrapolating the headline; the other is questioning what repeats.

What we’re watching next week

Tuesday 3 November: AMD reports after the US market closes. The useful distinction will be between demand already turning into sales and expectations about future AI spending. Both matter; they carry different amounts of uncertainty.

Thursday 5 November, noon GMT: The Bank of England publishes its rate decision and Monetary Policy Report. I’ll be looking for what would change its next decision, not just the explanation of this one.

Related reading: “A good company, but an expensive share”—our explainer on why business performance and shareholder returns can part company.