Skip to the main content.

What We Offer

ICX helps companies grow revenue and profitability by connecting customer experience, sales, pricing, CRM and automation into one commercial system.

Customer Experience

We design memorable, customer-centered experiences that drive loyalty, enhance support, and optimize every stage of the journey. From maturity frameworks and experience maps to loyalty programs, service design, and feedback analysis, we help brands deeply connect with users and grow sustainably.

Marketing & Sales

We drive marketing and sales strategies that combine technology, creativity, and analytics to accelerate growth. From value proposition design and AI-driven automation to inbound, ABM, and sales enablement strategies, we help businesses attract, convert, and retain customers effectively and profitably.

Pricing & Revenue

We optimize pricing and revenue through data-driven strategies and integrated planning. From profitability modeling and margin analysis to demand management and sales forecasting, we help maximize financial performance and business competitiveness.

Digital Transformation

We accelerate digital transformation by aligning strategy, processes and technology. From operating model definition and intelligent automation to CRM implementation, artificial intelligence and digital channels, we help organizations adapt, scale and lead in changing and competitive environments.

 

 

Operational Efficiency  

We enhance operational efficiency through process optimization, intelligent automation, and cost control. From cost reduction strategies and process redesign to RPA and value analysis, we help businesses boost productivity, agility, and sustainable profitability.

Customer Experience

chevron-right-1

Marketing & Sales

chevron-right-1

Pricing & Revenue

chevron-right-1

Digital Transformation

chevron-right-1

Operational Efficiency 

chevron-right-1

10 min read

Product Profitability: How to Uncover Misleading Margins

10 min read

Product Profitability: How to Uncover Misleading Margins

Product Profitability: How to Uncover Misleading Margins
18:48

The product everyone celebrates

There is a product in its portfolio that the commercial management talks about with a certain pride. In the profitability report it looks good, customers ask for it and in each price committee someone defends it. Now comes the uncomfortable question: how much of that margin really exists, and how much is being financed, without anyone having decided, by another product that almost no one talks about?

The indirect costs of a medium-sized company (supervision, maintenance, warehouse, purchasing, quality, invoicing, collections, dispatches) weigh enough that the way they are distributed changes the picture. And very often they are distributed with a rule that someone defined years ago and that no one looked at again: a percentage of sales, on labor hours or on the cost of materials. The rule is convenient and easy. It fits with the income statement, fits on a single sheet and is updated in ten minutes.

Its flaw is that it assumes that all products consume the structure of the company in proportion to what they bill, or what they cost in materials. And this is rarely true.

Think of the product that is run once a week, with a three-hour line change, three purchase orders for special supplies, one inspection per batch and a customer who calls twice per order. Next to it, the product that leaves the line continues without much drama. The second ends up paying for a good part of the complexity of the first, because the delivery rule does not see the change of line or the calls. It only sees colones, dollars or pesos sold.

And with that cost, everything else is decided: what price is set, what discount is authorized, which seller is rewarded, which reference is promoted and which is discontinued. A rounding error compensates for itself. A systematic bias, which always pushes in the same direction, accumulates over years and ends up looking like a strategy.

This article is about how to correct it with what a medium-sized company already has at home. The technique is called activity-based costing, or ABC, and you don't need a world-class software license to begin with (although I'm not going to deny that having one makes things much easier). Nor is it enough with the spreadsheet that you inherited from its predecessor and that only one person understands. Between those two extremes there is a lot of ground, and almost everything that follows happens there.

How Digital Automation Reduces Costs and Improves Margins

How to make a margin that doesn't exist

I'll show you an invented example, with exaggerated figures on purpose so that the mechanism can be seen without the need for a magnifying glass. A company manufactures two products. A comes from a continuous line: 100,000 units per month, price of 1.00 and direct cost (materials and labor) of 0.60. B is produced in short runs: 10,000 units per month, price of 2.00 and direct cost of 1.00. The indirect costs of the month add up to 40,000.

With the usual rule, distributed according to sales, A absorbs 33,333 and B absorbs 6,667. In the report, B seems to be the best deal in the house, with 0.33 margin per unit against 0.07 for A. It's easy to imagine the meeting in which someone proposes to push him.

Now ask yourself the question that the rule does not do: what do you do with those 40,000 and how many times do you do it for each product? The answers are these.

Activity

Monthly cost

Driver

Consumption of A

B Consumption

Line Changes and Cleaning

12 000

Line Changes

4

36

Quality Control

8 000

Inspected Lots

10

40

Purchasing and planning of special supplies

6 000

Purchase Orders

5

25

Line Operation

14 000

Machine hours

700

100

 

Each activity is divided by the total of its driver to obtain a unit cost (300 per line change, 160 per batch inspected, 200 per purchase order, 17.50 per machine hour), and each product pays for what it actually consumes. The result, per unit:

Product

Price

Direct cost

Indirect with the Sales Rule

Margin with Ruler

Indirect with ABC

Margin with ABC

A

1,00

0,60

0,33

0,07

0,16

0,24

B

2,00

1,00

0,67

0,33

2,40

−1.40

 

The company's profit is 10,000 a month in the two calculations. What moved is the attribution: A generates about 23,950 and B loses about 13,950. If sales had been granting discounts to B's customers because the report said that there was plenty of margin, each discount was magnifying a loss that no one saw.

This is how you arrive at those figures, month by month and with each traceable number:

Concept (per month)

Product A

Product B

Total

Sales

100 000

20 000

120 000

Direct cost

60 000

10 000

70 000

Margin over direct cost

40 000

10 000

50 000

Indirect Shares Distributed by Sales

33 333

6 667

40 000

Profit with the Sales Rule

6 667

3 333

10 000

Line changes (A: 4 × 300; B: 36 × 300)

1 200

10 800

12 000

Quality Control (A: 10 × 160; B: 40 × 160)

1 600

6 400

8 000

Purchasing and planning (A: 5 × 200; B: 25 × 200)

1 000

5 000

6 000

Line operation (A: 700 × 17.50; B: 100 × 17.50)

12 250

1 750

14 000

Total indirect with ABC

16 050

23 950

40 000

Utility with ABC

23 950

−13,950

10 000

 

Real figures are rarely so cruel. The sense of distortion, on the other hand, is often repeated: the rules of distribution by volume or by sales favor complex and slow-moving products, and punish simple and massive ones. It is the pattern that gave rise to ABC costing, and it is still valid because traditional accounting measures how much is spent in total, not how many times each thing is done.

Neither Excel on the loose nor a world-class platform

Before someone in the room says "we do this in Excel": yes, you can, and in the example above a sheet was enough. Calculation has never been the problem. What usually fails is everything that surrounds calculation.

A cost model that lives in a personal file has only one owner, and when that person changes jobs or companies, the model becomes a black box that no one dares to touch. It has versions that circulate in the mail with names like "final_v3_ok." It depends on data that someone copies and pastes every month from three systems, until the unfortunate day it is pasted wrong. And when the general manager asks why the cost of a product changed from month to month, no one can trace the result to the cause. They are failures of governance of the model, and they are solved with design and discipline, even if the tool is still a spreadsheet.

On the other hand, there is the temptation to solve everything by buying. There are very good costing platforms, and at ICX we incorporate them into projects when the case requires it (My ABCM and Cost Perform are two examples). But buying before you have the model in mind is an expensive way to automate an arbitrary rule. The difficult part of ABC is deciding what activities exist, which inducers explain them, where each piece of data comes from, and who is going to use the result. No software answers that for you. What it does do is run the calculation with less friction, and that's worth it when the volume justifies it.

Between the two extremes there is a model that an SME can sustain: a spreadsheet with separate sheets for resources, activities, drivers and products; an assumption sheet where each parameter has a source and date; a reconciliation that forces the assigned cost to match the income statement; and a person named as responsible, with an alternate. None of this requires a license. It requires someone to want to do it in an orderly manner.

There is also a variant that should be known, activity-based costing by time (TDABC). It asks for two estimates per activity, how much a minute of capacity costs and how many minutes each type of work consumes, and avoids endless staff surveys. It has a very valuable side effect for a CFO: it leaves in view the capacity that is paid for and not used, something that traditional costing buries within the cost of the product.

Common Mistakes When Allocating Costs in ABC Models and How to Avoid Them

 

Ten activities yield more than eighty

A model of eighty activities sounds more serious than one of ten, and therein lies the catch. To get it off the ground, you have to interview half the world, time processes, discuss whether "dealing with complaints" is an activity or part of "customer service," and in the third month someone asks when you will see results. When it is finally ready, no one in management can explain where the cost of a product comes from, and a number that cannot be explained does not change any decision.

A model of ten activities fits on a blackboard. It can be defended against the owner in a one-hour meeting. It can be updated without calling a consultant (although I prefer you to keep calling me), and that depends on whether it is still alive in two years.

What about accuracy? It's the logical objection of any CFO, and my answer is that it matters to the point where it stops changing the decision. If a product costs 2.40 or 2.10 and in both cases loses money at the current price, arguing with that 0.30 is a waste of time. The useful question is how much error each decision tolerates: setting a minimum selling price requires far more finesse than deciding whether an entire line is worth following.

To choose the "ten activities" it is convenient to start from what already exists, the cost centers of accounting and the departments of the organizational chart. They are subdivided only when two conditions are met at the same time: the cost is large and consumption varies greatly between products. Drivers come out of what is already counted (batches, purchase orders, machine hours, dispatches, invoices issued, technical visits). If a new system has to be set up to measure a driver, it is probably not yet necessary.

One question remains completely unresolved: when to add activity number "eleven". The most honest criterion I can think of is this: when its absence can change a price, product mix or customer decision. Before that, it is a detail that you pay twice, when building it and when maintaining it.

What we've seen in four organizations

I show you four Costa Rican organizations, with four activity-based costing models, most of them developed quite some time ago, and a common trait: none was born out of accounting curiosity. They were all born from a decision that was waiting for a number.

CEK Corporation is a group with several businesses, and the ABC model was built to decide which direction to take with each one. Such a model has to accomplish something difficult: that each business responds with its own figures and not with a portion of its neighbor's cost. In a group, the way of distributing the shared costs can decide on its own which business seems profitable and which seems a burden.

At Magneto Diesel (Madisa) the question was more concrete: what strategy to follow with the diesel engine workshop. A workshop brings together what keeps any manager awake at night, specialized technicians, expensive equipment and work that does not arrive evenly. The possible solutions in such a case (maintaining it, reducing it, rethinking it) cannot be defended without knowing how much an hour of work loaded with everything the workshop consumes really costs.

The case of El Ángel S.A. changes the type of reader. A tariff model was developed for its hydroelectric plant, with the purpose of having a technical basis to justify its costs to the Public Services Regulatory Authority. In front of a regulator, "that's how we have been calculating it" is useless; you have to show the path that goes from expense to tariff, step by step, so that another person can walk it and get to the same place. Whoever supports a model before that reader quickly learns that clarity weighs more than sophistication.

The National Registry is perhaps the most illustrative for this article. What was developed was a high-level ABC model, whose purpose was to set the guidelines that the institution must comply with to build a cost model on a national scale that allows it to establish fair rates for the services it provides. That is, a thick model to decide how to do the fine model well. Starting with little detail can be the right way to start, even when the end goal is ambitious.

Four cases do not prove a law, and it is worth saying so. In all four, the model was at the service of a specific decision (what to do with a business, with a workshop, with a tariff, with a methodology) and not of just another report for the archive. A model that is not tied to a decision soon becomes a formality. And the procedures are no longer updated.

When the owner already knows which products are profitable

In almost every SME there is someone who has been in the business for years and "knows" what sells well and what does not. That intuition is worth more than analysts usually admit, because it was formed with real customers, through errors and not with spreadsheets. He also has a blind spot: he does not perceive the complexity that occurs far from his desk. The three-hour line change, the special purchase order, the customer who calls twice a request happen on the floor or in an apartment where the owner almost never peeks.

The cost model enters that conversation as an intruder, and the typical reaction is "that number can't be," and I've heard it many times. There's a way to lower the tension. Before showing anything, ask the owner and his sales team to order the products from most to least profitable according to their criteria, and keep that list. Then show the result of the model and compare them. Where they coincide, intuition is validated and the model gains credibility. Where they don't, the discussion stops being one number against another and becomes what part of the resource consumption was not in sight, something that can be explained product by product: so many line changes, so many orders, so many machine hours.

A detail that should be said frankly: a model that never surprises is useless. If it coincides in everything with what was already believed, either the intuition is exceptional or the model is reproducing the previous rule under another name. It is worth finding out which of the two is the one that is happening.

And also for the following: sometimes the owner holds a product that loses money on paper because it opens the door to a large customer, or because without it an entire line falls. It can be a perfectly valid reason, as long as it is a conscious decision and with the cost in sight. What is not valid is subsidizing without knowing it.

What to do when the number is already on the table

Let's go back to product B in the example, the one that lost 1.40 per unit. The immediate reaction is to discontinue it, and it's almost always the worst. The model says something more useful: where the cost is. Of the 23,950 assigned to B, 10,800 come from line changes. That opens up options that discontinuing doesn't even contemplate: a minimum run size, a small order charge, production schedules that group orders, a line change redesigned to last less.

Each of these levers belongs to a different area. The price is moved by sales, the lot size is moved by operations, the discount policy is moved by finance. That is why a costing presented alone by the CFO yields little: the result needs to seat all three areas at the same table.

The same goes for customers. A profitable product can be sold to a customer who makes it expensive, because he orders little and often, demands deliveries to remote points or pays in ninety days. Paying per customer as well as per product is the natural step he takes, and it is usually more surprising than the first.

It is the logic with which ICX organizes its Pricing, Costing & Revenue Management practice, three linked questions: how much does it really cost to produce, sell and serve; what price must be charged to capture value; and how to optimize revenues and demand on that basis. Answering the first without the second leaves the model in a drawer. After the diagnosis, so that the models do not age, there are periodic support services such as Costing IQ for costs and Pricing IQ for prices.

And when is a specialized tool justified? When the model needs to be updated every month with large volumes of data, when there are several plants or thousands of references, when several users need to consult it at the same time or when each update cycle consumes more hours than the team can dedicate to it. To get to that point, you have to have built the conceptual model beforehand, and a well-ordered spreadsheet is a very reasonable laboratory to do so.

Turn Your Order-to-Cash Process into a Profit Driver with Appian

The decision that is up to you

If you want to try this without compromising half a company, the bet is small: one product line, one month of data, ten activities. The goal of that first attempt is modest. To know if the cost you use today is similar to the cost revealed by the model.

The hard part comes later, and it's not technical. Suppose the number tells you that the house's favorite product loses money at its current price. Who in your organization has the authority, and the desire, to change that price? And what if that person is precisely the one who believes in the product the most?

A real cost that no one is willing to use is a pretty spreadsheet. Before building the model, it's worth finding out if your company has that question answered.

 

 

Content added to ICX Folder
Default Save Save Article Quit Article

Save for later

Print-Icon Default Print-Icon Hover

Print

Subscribe-Icon Default Subscribe-Icon Hover

Subscribe

Start-Icon Default Start-Icon Hover

Start here

Suggested Insights For You

What's next?

ARE YOU READY?

ICX SUBSCRIPTION
Stay ahead with exclusive insights from our senior consultants. Subscribe for the latest thinking on CX, growth, and digital transformation.