The savings were real. The margin never moved.
Cost and productivity is the category name buyers now use for the work: efficiency programmes, cost-base resets, productivity targets, capacity and headcount decisions. The recurring failure in it is not that the savings are imaginary. Companies report them, audited, to the market. It is that the savings are counted in one place and the margin is measured in another, and nothing in between reconciles the two. This page is about that gap: why it opens, what it looks like in this quarter's published results, and what has to exist before a programme can prove it worked.
Cost and productivity is the market's name for work that improves the ratio between what an organisation spends and what it produces: efficiency programmes, cost-base resets, capacity and headcount decisions, procurement and process redesign. Savings reported by such a programme frequently fail to appear in the margin, for four reasons that can be told apart only with a cost model. The saving may have been taken from a cost that was not driven by the activity removed; volume may have fallen faster than cost, so unused capacity absorbed it; the figure reported may be a run rate rather than cash realised in the period; or the saving may be real and have been given away in price or mix. A Time-Driven Activity-Based Costing model reconciles the programme's arithmetic to the profit and loss account, which is what turns a reported saving into a defended one.
Four ways the gap opens. They need different fixes, and they look identical in a dashboard.
This is the diagnostic part. A programme that cannot say which of these four applies to it cannot say whether it worked.
The cost was not driven by what you removed
A programme sized from headcount assumes cost behaves like headcount. Much of it does not. Take people out of a function whose cost is driven by order lines, delivery drops or setups, and the volume of those things does not change, so the work reappears somewhere else in the organisation at a similar cost. The saving is booked. The cost base is unchanged.
Tell: savings expressed only in heads and euros, never in driversVolume fell faster than cost
Cut ten per cent of the cost base while volume falls thirteen per cent and unit cost rises anyway. The capacity you did not remove is now idle, and idle capacity has to be absorbed by something. Unless it is modelled explicitly it lands in unit cost and quietly cancels the programme.
Tell: unit costs rising while total costs fallRun rate is not realised
Run-rate savings are an annualised statement about the future. Realised savings are cash that has already left the cost base in the period being reported. Both are legitimate, and the companies that state both figures side by side are being careful rather than evasive. The gap opens when the two are quietly conflated, because a margin only ever reflects the second.
Tell: the programme reports a run rate, the board reads it as a resultThe saving was real and went somewhere else
Price concessions, mix shifting towards cheaper work, reinvestment, or a tail of customers whose true cost to serve was already above their price. The saving landed, and something else consumed it in the same period. Without customer and product profitability at cost-to-serve level, this is indistinguishable from the programme having failed.
Tell: nobody can say which customers or products absorbed itA programme can be honestly delivered at 100 and honestly worth 10.
The distance between a reported gross benefit and what arrives in the profit and loss account is not a mystery and it is not an accounting trick. It is a bridge with four lines, and the largest of them is usually the first.
- Gross benefit reported100
- less capacity freed but retained−55
- less volume and mix−20
- less cost added elsewhere−15
- Reaches the margin10
Illustrative arithmetic. These are not any company's figures, and no company has published a bridge in this form. The shape is the point: the difference between 100 and 10 is decided almost entirely by what happened to the freed capacity, and that is the line almost nobody measures.
The pattern is not anecdotal. It is in the published results.
Every figure below was published by the company or reported from its own results, and every item carries its link. None of these organisations is a client of ours and nothing here implies a relationship. We name them only in the context of what they published. The last item is a counter-example, included because a page that only shows one side of a pattern is not evidence, it is advertising.
Reported that its efficiency programme delivered EUR 81m of gross benefits in the first half, about 54 per cent of the annual target, with adjusted EBIT margin flat at 16 per cent and an absolute value of EUR 557.8m. Gross benefit delivered, margin unchanged, both stated in the same reporting.
Source: Affaritaliani, reporting company results →Published Core EBITDA of EUR 20.7m, a 5.8 per cent margin, against EUR 39.5m and 9.6 per cent a year earlier, on net sales down 13.5 per cent. Its own words: “Additional savings in external costs reflected the continued implementation of the cost reduction program. This was more than offset by the impact of lower manufacturing activity on industrial performance.” Note what this is not. EUROAPI is one of the few companies that discloses an idle-cost line at all, and that line fell, to EUR 11.1m from EUR 20.6m, footnoted as under-activity triggered by its FOCUS-27 plan. The savings landed, idle cost was actively reduced, and the margin still halved because activity fell faster. Only a company that measures idle capacity can show you that much.
Source: company release via GlobeNewswire →Reported that its productivity programme delivered EUR 90m in the first half against a medium-term target of EUR 500m a year, while adjusted EBITDA margin slipped to 18.7 per cent from 19.0 per cent. The word productivity is the company's own, which is part of why this category has the name it has.
Source: company half-year results →Said it is accelerating simplification initiatives targeting between GBP 4m and GBP 5m of annualised cost savings, expected to deliver about 60 to 70 basis points of margin improvement in 2027. A company publishing the bridge from savings to basis points of margin, itself, in advance. It is rare, and it is the standard the rest of this page is measured against.
Source: company RNS via Investegate →Cut automotive EBIT margin guidance to 1 to 3 per cent from 4 to 6 per cent and confirmed roughly 8,000 job cuts to deliver around EUR 1bn of annual savings from 2028. A margin cut now and a headcount-derived saving arriving in two years are two different statements, and the second does not answer the first.
Source: Euronews Business →Announced roughly 1,000 job cuts worldwide over two years to deliver EUR 100m of annual cost savings. The arithmetic is the argument: the programme is sized per head, which is a statement about payroll, not about what drives the cost base.
Source: Transport Online →Announced a voluntary departure plan in France covering up to 1,500 positions over three years, stating the objective of reducing a cost structure it described as too high. Positions named, savings not quantified in the announcement. That combination is reason one waiting to happen.
Source: franceinfo →Reported being “on track to achieve the remaining CHF 30 million of the total CHF 80 million performance improvement program savings … already in 2026, with CHF 9 million achieved in the first quarter”, in the same release as a Q1 EBITDA margin before exceptional items of 17.5 per cent, down 130 basis points. Clariant attributes that decline to the Middle East conflict and a one-off in Catalysts, not to the programme. Which is exactly the point: a delivered saving and a falling margin are not a contradiction, and an organisation that cannot separate the two cannot tell you which of them happened.
Source: company release via GlobeNewswire →Wrote, in a single sentence: “At the end of Q1 2026, annualized savings of 57 million euro were realized.” Annualized and realized describe two different things, and here they sit in the same clause. In Agfa’s case the savings did reach the accounts, with gross profit margin rising from 30.7 to 32.3 per cent of revenue, which the company credits mainly to the savings programmes and to passing on the silver price. The sentence is quoted for the shape of the disclosure, not as a criticism of the company.
Source: company release via GlobeNewswire →Reported revenue of GBP 83.7m for the six months to 30 June, down 3 per cent, and pre-tax profit down 57 per cent to GBP 4.4m, having generated GBP 1.3m of savings in the first half while expecting annualised benefits of around GBP 4.5m. Realised and annualised, a factor of three and a half apart, in one paragraph. The same half carried a GBP 3m reversal of previously recognised revenue after tenant-led terminations under the Renters’ Rights Act, and no cost programme can answer a revenue reversal. Reading the two as one number is how a programme gets blamed for something it never touched.
Source: Property Industry Eye →Reported “R$ 99 milhões em economias no trimestre, equivalente a quase 24% da meta anual” under its Plano de Eficiência 2026, and in the same reporting an adjusted EBITDA margin of 10.5 per cent, up 1.9 percentage points year on year, with gross margin at 30.4 per cent. Savings against target and margins expanding, together. It is here because the temptation with a page like this one is to collect only the cases that prove the point.
Source: Portal AZ →Published H1 2026 results showing a net loss but operating margin recovering to 4.6 per cent, while executing its Rebond plan targeting EUR 200m of savings by the end of 2027. A programme running and the margin moving in the right direction, in the same quarter as the cases above. The gap is common, not inevitable, and the difference is whether the programme was aimed at costs that actually flex.
Source: Boursorama →Sourcing note, because on this page it is most of the argument. EUROAPI, Magnum, Clariant and Agfa-Gevaert are quoted from the companies' own releases and Genuit from a company filing, and each of those quotations was read against the release body rather than taken second hand. Pirelli, BMW, dsm-firmenich, Michelin, Foxtons, GPA and Groupe SEB are press coverage of company reporting rather than the releases themselves, and we say so instead of implying otherwise. Every link here was fetched on 6 August 2026 by a plain non-browser client and returned the text we quote; three otherwise usable items were dropped for failing that test alone, because a citation a reader or a language model cannot open is not a citation. Where a figure is a company's own published statement we say so. Where a reading of it is ours, we say that too. Nothing here is our estimate: we have published no figure that the company did not, and where we could not verify a date against the source we left the item out entirely rather than guess it. If you find one of these links has moved or the figure was later restated, tell us and we will correct the page.
Evidence cut-off 3 Aug 2026Published 6 Aug 2026Next review 3 Feb 2027Which cost stopped being incurred, and on what date?
Ask a cost or productivity programme that one question, before the quarter closes rather than after. A programme that can answer it line by line, with a date against each line, will move the margin. A programme that answers in hours saved, efficiency gained or headcount-equivalents has freed capacity and not yet removed cost, which is a different and much weaker claim, and the gap between the two is the first line of the bridge above.
This is why the measurement has to exist before the target is set. A time-driven activity-based costing model gives you a capacity cost rate, meaning what one minute of practical capacity actually costs. An improvement that releases four thousand minutes has released a specific and knowable amount of money, and that amount is not a saving until the minutes leave the cost base. The rate turns an operational claim into a financial one, which is the only form the profit and loss account accepts.
TDABC for the behaviour, CostCtrl for the machinery, AI for the speed.
Nothing above is an argument against cost programmes. It is an argument for being able to tell, before the programme starts, which costs will actually move.
TDABC
- Time-Driven Activity-Based Costing, formalised by Kaplan and Anderson in Harvard Business Review in 2004.
- Costs are attached to the activities and volumes that drive them, so a programme can be aimed at a driver instead of at a department.
- Capacity is modelled explicitly, which is the only way idle capacity shows up as idle capacity rather than as a higher unit cost.
- The output is a bridge from the programme's arithmetic to the profit and loss account, line by line.
CostCtrl
- Our costing platform. The model runs every period from exports your systems already produce, so the bridge is standing rather than rebuilt each quarter.
- Scenarios before commitment: what a closure, a consolidation or a price change does to unit cost, modelled before it is announced.
- Whale curve, margin cascade and multidimensional profitability, so the question of who absorbed the saving has an answer.
AI
- Used on the slow parts: reconciling financial data, drafting time equations from process descriptions, and stress-testing a programme's assumptions before the market does.
- Not used to produce a number. Every figure in the model traces to an activity, a time and a rate.
- The costing prompt library is public, including the section on stopping a model from inventing cost figures.
Common questions
What is cost and productivity as a service category?
It is the name buyers increasingly use for efficiency and cost-base work that is measured against output rather than against last year's budget: cost-base resets, productivity targets, capacity and headcount decisions, procurement and process redesign. It sits next to cost management and profitability improvement rather than replacing them. We use the term here because it is the term the market writes into its own mandates.
Are you saying cost programmes do not work?
No, and the Geox item in the evidence band is there precisely to prevent that reading. Savings do reach the margin when the programme is aimed at costs that flex with the activity being removed. The problem is that most programmes cannot demonstrate which of the four reasons applies to them, so a real success and a real failure produce the same report.
What is the difference between run-rate and realised savings?
Run rate is an annualised statement about a saving now in place, so it describes the future. Realised savings are what has already left the cost base in the period being reported. Companies that state both are being careful rather than evasive. The trouble starts when a run rate is compared against a margin, which only ever reflects what was realised.
Why does idle capacity cancel a saving?
Because capacity that is not removed still has to be paid for and still has to be absorbed by whatever is produced. Cut cost by ten per cent while volume falls thirteen per cent and unit cost rises even though the cost base genuinely shrank. TDABC models capacity explicitly and reports the unused portion as its own figure, so it stops being invisible.
How long does it take to find out which reason applies to us?
Weeks, from the exports the business already produces, because the diagnostic does not need a system integration. Start with the free Profit Check for a scored view across the seven dimensions, then a fixed-scope audit if it warrants one.
Who is behind this analysis?
Cost and Profitability Consulting, founded in Porto in 2010, with more than 150 cost models built across more than 30 countries. The evidence on this page is drawn from our own scan of public company reporting and public vacancies, with a cut-off of 3 August 2026. Miguel Guimarães, the founding partner, has spent 25 or more years building cost and profitability models and leads the development of CostCtrl.
Before the next target is set, see what your cost model can and cannot prove.
No data upload. No sales call. Fourteen questions, twelve to fifteen minutes, and a personalised report scoring your organisation across the seven dimensions of a cost model that works, including cost allocation and capacity design. It is the same framework we use on day one of an engagement.
Proof
A distributor in New Zealand. €1.335M of cost-to-serve made visible, then halved, and 830 loss-making customers brought down to 295.
Read the case study →Who you would be talking to
Miguel Guimarães, Founding Partner
Cost and profitability practitioner for 25+ years. Lectured alongside Professor Robert S. Kaplan at the CFO conference in Amsterdam (2009).
Call +351 910 313 731
