Your new system already works and nobody uses it: the eight steps of change, with Mexican cases

In May 2022, hundreds of people were lining up from early morning outside the offices of the SAT, Mexico's tax authority, on Paseo de la Reforma in Mexico City. The line ran along the sidewalk and turned the corner. They were there for their tax status certificate (constancia de situación fiscal) [9].

Some human resources departments were asking for it. Many believed that, without that piece of paper, they would not be able to collect their salary [9].

What had happened? The SAT had launched version 4.0 of the electronic invoice [7]. On paper it was a technical change. In practice it forced companies and taxpayers who issue invoices to review something harder than a system: how they captured the data of their customers and their employees.

One clarification before we start: Hábil did not take part in any of these cases, nor did it work with the companies mentioned here. We tell them using what the SAT, the companies themselves and the press reported publicly, with the date of each fact.

In 1 minute

• The risk: that your new system launches on time and your company keeps working as before. You paid for the system and you do not get the control or the numbers you were after.
• The case: with e-invoice 4.0, the tax deadline arrived before clean data did. The SAT moved deadlines and opened a new channel so companies could obtain their workers' data [10][12][14][15].
• What inaction costs: as long as the old system stays switched on, you pay for two systems, your people capture everything twice and the board has two versions of the truth. The benefit that justified the investment does not arrive.
• What you take away: thresholds for launching and for switching off the old system, the costliest mistakes and a one-page decision sheet for your committee.
• What you have to decide on Monday: who on the business side, named and with authority in writing, owns the change. Not the system: the change.

How to read this article: facts come with their source [n]; interpretations appear as Our reading; recommendations sit under "What your implementation must have"; examples without a source appear as Example scenario.

In this article

1. What happened with e-invoice 4.0?
2. Where do the eight steps come from?
3. What does each step mean for your implementation?
4. What numbers should you ask for before launch and before switching off the old system?
5. Which mistakes cost the most?
6. What should you take to Monday's meeting?
7. Key takeaways
8. What does Hábil do?
9. Frequently asked questions

What happened with e-invoice 4.0?

The tax deadline arrived before clean data did, and the SAT had to move deadlines and remove obstacles.

On January 1, 2022, version 4.0 of the electronic invoice came into force. It coexisted with the previous version, 3.3, until March 31, 2023 [7].

Version 4.0 required mandatory data about whoever receives the invoice: name or company name, tax regime and the postal code of their tax address [7]. The SAT's current guide requires the name to match the RFC (the taxpayer ID) and the postal code to be associated with that RFC [28]. And the validation is done by the certification provider when it stamps the invoice [28].

That was where the real change lay. The press reported cases in which employers had their employees registered with the company's address [9]. In business terms: a badly captured piece of data can turn into an invoice that does not get stamped, that is, one that is not recorded as valid with the SAT. The SAT's error matrix has a code for exactly that: the recipient's name does not match the one on their RFC [29]. And a payment collection gets stuck.

What followed, by date:

According to the SAT itself, two months after the January channel opened, more than 100,000 companies had used it. As a result, more than 7 million salaried workers did not have to go to its offices [14]. Both figures come from the SAT, with no second source.

The mandatory dates changed several times; for payroll, the exception lasted until July 2023 [10][12][15]. Our reading: the deadline can arrive before the data, and before the people who capture it.

Where do the eight steps come from?

It is a map from 1995 for planning an implementation, not a scientific law.

In two lines: John P. Kotter, now professor emeritus at Harvard Business School, published in Harvard Business Review in 1995 the eight errors that make a change effort in a company fail [1]. His central lesson: change goes through phases, takes time, and "skipping steps creates only the illusion of speed and never produces a satisfying result" [1].

The eight, in one line: urgency · a coalition with power · vision · communicating it · removing obstacles · early wins · not declaring victory too soon · anchoring the change in the culture [1].

For those who want the backing:

• Where it comes from. A decade observing more than a hundred companies that were trying to change; most ended up halfway, closer to the bad side [1].
• How it evolved. A reprint in 2007 [2], a book in 1996 (Leading Change) [3] and today, under other names, at his firm [4]. We use the 1995 version.
• The "70% fail" figure is unsupported. Kotter does not give that figure [1], and a researcher at the University of Brighton reviewed five publications that repeated it without finding valid evidence to support it [5].
• How much weight the evidence carries. A 2012 review supports most of the steps, but found no studies of the complete model; it recommends using it for planning, alongside other tools [6].

What does each step mean for your implementation?

Each step brings a condition your project has to meet, before or after launch.

Does your leadership team believe that staying the same is worse than changing? (step 1)

If the only reason to change is a date, people work for the date, not for the change.

What Kotter says. It all starts when someone puts the uncomfortable numbers on the table. According to what he observed, well over half of companies got stuck right here [1]. His rule: there is enough urgency when about three out of four executives are convinced that carrying on as before is unacceptable [1].

In the e-invoice 4.0 case. The urgency was not born inside the companies: the SAT set it with a date, and the date moved [10][15]. Our reading: borrowed urgency lasts as long as the deadline does. The sources do not measure how many companies waited because of that.

What your implementation must have. A business case in pesos, not in adjectives:

• How much double data entry costs today.
• How many days the month-end close takes.
• How many invoices come back with errors.

If the only reason is "the old system is no longer supported," you bought an upgrade, not a change.

Who on the business side is pushing, by first and last name? (step 2)

If the project "belongs to IT," every area will treat it as a favor it does for IT.

What Kotter says. No major change moves forward with one or two people. It takes a group with real power: through position, through information and expertise, through reputation and through relationships [1].

And he warns against leaving it in a support area, such as human resources, quality or planning, instead of with an executive from operations: without that leadership, the group never gathers the power it needs [1].

Case. We do not use one: the public sources do not say who led each project.

What your implementation must have.

• One person from the business with authority in writing to say "this is how it will be done" when two areas cannot agree, from day one.
• Alongside that person, those who know the operation: the warehouse manager, the woman who runs collections, the salesperson who bills the most.

Can you explain the change in five minutes? (step 3)

If nobody can say what each role will do differently, the project does not yet know what it is changing.

What Kotter says. A vision goes beyond the figures of a five-year plan: it makes clear where you are heading [1]. His test: if you cannot explain it in five minutes and get the other person to understand it and care, it is not ready yet [1].

In the e-invoice 4.0 case. What the SAT required was that the invoice carry the same information as the recipient's certificate [14][28]. Compare two ways of announcing it inside a company:

Example scenario:

What your implementation must have. One page that says what each role will do differently on launch day: the cashier, the data-entry clerk, the salesperson, the accountant. If nobody can write it, there is no vision yet.

Have your people already seen the system working? (step 4)

An email and a kickoff meeting are not communication; seeing the director use the system is.

What Kotter says. Companies tend to communicate the vision about ten times less than they need to: that is the name he gives that error [1]. A meeting or an email is a drop in everything people hear in a year [1].

And actions weigh more than words: nothing slows a change down as much as a senior executive who says one thing and does another [1].

In the e-invoice 4.0 case. The requirement was for the data to match [12]. In some companies it reached workers as "bring your certificate," and some believed that without it they would not get paid [9]. The SAT had to clarify that it was enough to know the data and that payment of salary does not depend on the payroll invoice [10][11].

Our reading: the case does not measure how many times the message was communicated. It shows what happens when an instruction arrives distorted: people act on what they understood.

What your implementation must have.

• Regular demonstrations with what has been built, not presentations with screenshots.
• Executives who use it first. If the director keeps asking for the report in Excel, the system is optional.

What is getting in your people's way, and who is removing it? (step 5)

Removing one obstacle pays off more than repeating the deadline.

What Kotter says. Often people want to help, but something gets in the way: the structure, or a compensation or appraisal scheme that forces them to choose between the change and their own interest. Worst of all, he says, are the bosses who keep demanding the old way [1].

In the e-invoice 4.0 case. This is the step the case illustrates best. The obstacle was getting the correct data for each worker: while each one had to go and get their certificate, there were lines [9]; when companies could request the data directly, more than 100,000 used the channel in two months, according to the SAT [14].

What your implementation must have. A list of obstacles, reviewed every week by whoever makes the decisions. The typical ones are unglamorous and very expensive: duplicate customers, products with three codes, unreconciled balances, and commissions paid per order captured instead of per order captured correctly.

What win will you show in the first year? (step 6)

An implementation that promises everything for the same day has no early wins: it has a single exam day.

What Kotter says. Most people will not keep marching unless they see, within 12 to 24 months, evidence that it is producing results. Those wins have to be clear, and they are planned: you do not wait to see whether they come [1].

In the Cemex Go case. Cemex, the Mexican cement company, launched in November 2017 a platform for its customers to place orders, track their deliveries and handle invoices and payments online: Cemex Go [17][23]. It started in the United States and Mexico, with the rest of its countries planned for 2018 [17]. It went on to report milestones with numbers:

These are milestones Cemex reported, not independent verifications. Our reading: they have the shape of visible wins, with a number per stage. Whether they were planned from the start, the sources do not say.

What your implementation must have. A first win that can be counted, with its number written down before launch: one branch, one product line or one process. If a regulatory deadline forces everything to launch at once, the early win is planned within that launch.

Did you already declare victory on launch day? (step 7)

Launch day is not the end of the implementation: it is the beginning of the hard part.

What Kotter says. Celebrating progress is fine; declaring the war won can be catastrophic [1]. Until the new way becomes part of the culture, something that can take five to ten years, it is fragile [1].

In the Cemex Go case. In May 2019, with the platform in 21 countries, Cemex's CEO said the company was still only in the early stages of its digital transformation [20]. That year it began expanding what the platform could do [22].

In Bimbo's ERP. Grupo Bimbo has spent years moving its operations onto a single management system (ERP) in the cloud. According to its annual reports:

• 2020: it completed the implementation in Argentina, Colombia and Peru [24].
• End of 2021: the system was in 38% of its operations [25].
• End of 2022: in 70% [26].
• 2025: it reported 90% deployment [27]. That report no longer says "of its operations," so the figure may not measure the same thing.

The figures come only from its reports: at least five years of work, and in 2025 it was still moving forward [27]. The same report announces acquisitions such as Wickbold in Brazil and Karamolegos in Romania, without saying whether they will join the ERP [27]. Our reading: in a company that keeps acquiring, 100% can be a moving target.

The sources cited do not say how much either project cost: Bimbo reports its total investment, not that of the system [27]; Cemex reports customers and sales, not amounts [20][23].

In the e-invoice 4.0 case. The SAT moved the date several times, but once it set it, the old version stopped counting: from April 1, 2023, an income, expense, payment or transfer invoice in version 3.3 has no tax effect and may be penalized [8].

What your implementation must have.

• Support after launch and measurement of real usage: how many orders come in outside the system?
• A written condition, with someone accountable, for switching off the old system.

Which system do the numbers for Monday's meeting come from? (step 8)

If the numbers for Monday's meeting come from Excel, the change did not stick.

What Kotter says. Change sticks when it becomes the normal way of doing things. For that, you show how the new way improved results and make sure that the people who get promoted embody it: a bad succession at the top can undo a decade of work [1].

In the Cemex Go case. In 2019 the company said the platform was continuing to transform the role of its sales force, which could spend more time with customers [20]. And in November 2022, five years after the launch, more than 50,000 customers were using it: 93% of its cement customers [23].

Our reading: these are signals, not proof that the change took root in the culture; Cemex does not publish that.

What your implementation must have.

• The new system as the reference source for leadership's numbers, with documented exceptions.
• When someone asks "how is this done?", the answer is "like this, in the system," not "ask Juanita."

What numbers should you ask for before launch and before switching off the old system?

Without written thresholds, "we are ready" is an opinion; with thresholds, it is a decision.

Some come from the model or from the SAT; the indicative ones are our own judgment, which your committee should adjust.

Which mistakes cost the most?

The costly mistakes documented by Kotter and by these cases are not technical: they are about who decides, about data and about when the old system is switched off.

Our reading: those reengineering efforts are the closest thing to the title of this article: a system that already works and that, little by little, nobody uses.

What should you take to Monday's meeting?

Eight questions, one per step, with a measurable answer; and a one-page decision sheet.

Both are our own tools, not part of Kotter's model. Each question asks for a piece of data and comes with its green and its red.

How to read it. It is not a score, and anything that is not green counts as red. If step 1 or step 2 comes out red, it is advisable not to launch yet: more than half of the companies Kotter observed got stuck on the first, and without the second the group does not gather the power it needs [1]. If the date is regulatory, it is not postponed: you narrow the scope of the launch and name the owner that same week. If steps 5 to 7 come out red, a narrower launch or a clearer switch-off date is advisable.

And if your change does not alter anyone's work, for example a version upgrade with no process changes, steps 3, 4 and 8 carry little weight. Do not oversize the project.

The decision sheet. One page the committee signs before launch. The right-hand column is an Example scenario, not a real case:

Key takeaways

Three questions for your next executive committee:

1. Who on the business side, named and with authority in writing, owns this change? If the answer is "IT," the project has no owner in operations.
2. Do we launch in stages or all at once, and with what number will we know it went well? Either can work; launching without the number cannot.
3. What condition allows us to switch off the old system, and who answers for it? Without it, the old system is never switched off.

What does Hábil do?

We build and integrate systems with a way of working that forces us to name who decides, against what the work is accepted and how it is rolled back.

Hábil is a Mexican engineering consultancy that has been building and integrating critical systems since 2006 [30]. The change is led by someone in your company; what we look after is that the systems project does not get in its way. This is how we work:

• Who decides: from the start, we agree with you on the person from the business who validates and prioritizes.
• What exists today: before designing, we measure processes, systems, data and rules; if the new system connects with another system, the connection comes with tests your team can run again.
• What it is accepted against: before building, you know what will be delivered, in what order and against what you will accept it.
• See it before using it: in stages with acceptance at each one, or in two- or three-week cycles with working progress.
• Your people validate: acceptance testing is done by your team, in your own environment, before release.
• After launch: an agreed rollback, support during the first days, and the user and operations manuals.

See How we work.

If you are about to launch an implementation, or you already launched it and it is not moving forward, let's start there. Request your first conversation, at no cost. You come away with a concrete map of your case: scope, who takes part on your side, what evidence is reviewed and what decision it lets you make. Write to us on WhatsApp.

Frequently asked questions

Did Hábil work with the SAT, Cemex or Bimbo?

No. This article is based solely on public sources: SAT documents and press releases, the companies' reports and the press cited in the references.

Sources opened and consulted on October 22, 2026. Quotations from English-language texts are given in the original; paraphrases are our own.