How to reduce SaaS software costs without contract talks

A medium-sized company in São Paulo renews a stack of international software subscriptions every January. The IT director knows it is expensive. The path he sees is to ask the vendor for a discount, contract by contract, and this path consumes the first quarter, yielding meager results, because the conversation with the vendor tinkers with the unit price and leaves untouched the two numbers that determine the invoice: how many licenses exist and through which channel they are acquired.

There are four levers that reduce the cost of the software stack without going through the negotiation table. These are the purchasing channel, the licensing model, the recovery of idle seats, and spend consolidation. None depend on the vendor's goodwill, and all attack the base of the expenditure instead of the unit price.

The ordering criterion for this text is the time until the cash flow feels the difference, crossed with how much the lever depends on third parties to happen. The first two change the purchasing structure and take one to three months to show up in the results. The last two recover money that has already been spent or is spent every month, and the third of them is the only one that returns cash in less than thirty days.

How We Chose and Ordered These Levers

The four were chosen by a single test: the cost reduction must happen without the vendor needing to agree to it. Excluded were renegotiation, switching vendors for a competitor, and the threat of cancellation. These are known tactics, and all share the same flaw, which is to put the outcome in the hands of another company.

The order of execution follows three variables. The first is the implementation effort, which measures how much internal work the lever requires before yielding results. The second is vendor dependency, which measures whether an external party can hinder the initiative. The third is the time to cash, which measures the lag between deciding and seeing the effect on the payment flow.

Recent data provides the dimension of the problem. According to the ABES/IDC study, software developed abroad accounted for 73.7% of the Brazilian software market in 2023, and software produced domestically accounted for 24.8%. The majority of a Brazilian company's stack, therefore, is purchased from outside, which means that most of the expenditure carries exchange rates, import taxes, and international contracting processes within its price.

A buyer averse to renegotiation is not disarmed in the face of this. They have four instruments, and the first one changes where the purchase is recorded.

1. Purchasing Channel: Where the Same Software Costs Less

The purchasing channel is the lever that decides the route international software takes to enter the company, and it usually yields more than any one-off discount because the cost is not only in the license price. Import charges and currency conversion increase the effective cost of an international solution by 50% to 70%, which turns a US$100,000 invoice into up to US$155,000 disbursed. Changing the channel does not change the vendor and does not change the product; it changes who processes the remittance, in what currency, and with what document.

The standard route is direct international contracting. The company signs with the foreign vendor, receives the charge in dollars or euros, processes the remittance through the bank, and absorbs the exchange rate variation between purchase approval and payment settlement. In this route, the import cost is added on top of the license price, and the buyer does not have, at the time of purchase, predictability over the final value. The problem is not the vendor. It is the route.

The alternative is to contract through a channel that already handles international purchases and issues domestic tax documents. Nexforce Marketplace is this path: the buyer starts receiving an invoice in Brazilian Reals, with an exchange rate lock and local payment via boleto or Pix, and the license value stops fluctuating after the purchase is approved. Companies under the Lucro Real tax regime recover PIS/COFINS credit of 9.25% (Laws No. 10,637/2002 and 10,833/2003) through the domestic invoice. In direct international contracting, although Law No. 10,865/2004 provides for PIS/COFINS-Importation credit for Lucro Real, the Brazilian Federal Revenue Service frequently restricts the crediting of general software, understanding that it does not constitute a direct input. In the horizon of the Tax Reform (Complementary Law No. 214/2025), PIS and COFINS will be abolished in 2027 by the CBS (with a 2026 test of 0.9% CBS and 0.1% IBS), ensuring broad non-cumulativeness. Companies under the Lucro Presumido tax regime do not take this credit in the current model, and the direct benefit is exchange rate lock, invoice in Brazilian Reals, and operational simplification.

The channel is the lever with the greatest base effect and the slowest implementation, because it requires changing the contracting route for vendors already in use. For those who already purchase dozens of international solutions, the gain accumulates with each renewal. For those who purchase two, the gain exists and is small. The decision on how to compare these routes before choosing is detailed in direct purchase or software marketplace, and the comparison criterion between channels in how to compare software marketplaces.

The limit of this lever is clear. It reduces the import cost and eliminates exchange rate exposure, but it does not reduce the price the vendor charges for the license itself, and it does nothing for licenses that are already paid for and idle. These two fronts belong to the subsequent levers.

2. Licensing Model: Pay-per-Use Instead of Provisioned Seat

The licensing model decides whether the company pays for the number of people who have access to the tool or for the number of people who actually use it. The difference between the two counts is the monthly waste. Switching from provisioned seat to metered consumption, when the vendor offers both modalities, usually frees up more cash than annual renegotiation, because the savings apply to the entire base and not just the discount agreed upon in the contract.

The provisioned seat model is the most common in corporate contracts and the most generous to the vendor. The company buys a block of licenses, distributes them among departments, and renews the block on the contract anniversary. Those who join the company gain access, those who leave do not always return the license, and the block grows by inertia. The cost does not follow usage, it follows the historical team size.

The metered consumption model charges for what was processed, called, or executed during the period. It transfers the risk of idleness from the buyer to the calculation basis. In observability tools, data processing, and AI infrastructure, consumption is already the natural unit of billing, and the platform exposes spending per team and per project. What the company needs to do is read this number before renewing, not after.

The model change is a contractual decision and, therefore, depends on the vendor accepting the modality. This is where the lever reaches its limit: when the vendor only sells by seat, the path is to reduce the block on the next anniversary based on actual measured usage, not to ask for a percentage discount on a block whose size no one can determine. The calculation that supports this reduction is in how to calculate the total cost of foreign SaaS.

There is a restriction worth stating. Migrating to metered consumption exchanges a predictable cost for a variable cost. In teams with concentrated peak usage, the metered model may cost more than the fixed block. The right choice depends on the dispersion of usage within the month, and the company only has this data if it measures before migrating.

3. Idle Seats: Recovering What's Already Paid For and Unused

Idle seats are active licenses that no one uses, and recovering them is the only lever on this list that returns cash in weeks. The typical waste comes from three sources: the license of someone who left the company and did not have their access revoked, the duplication of two tools that perform the same function, and the seat provisioned for a project that has ended. None of these appear on the invoice as a separate item, and that is why the cost survives accounting audits.

Idle doesn't ask for a discount. It asks for revocation.

The diagnosis requires crossing two sources that usually live separately. The first is the list of active subscriptions, with the value of each. The second is the access log for each tool, showing who logged in and when. Cross-referencing the two produces the list of paid and unused licenses, and this list is what the company takes to renewal.

The gain here is from recovery, not negotiation. Each idle seat the company cancels comes off the recurring expense the following month, and the effect is immediate because it does not depend on the vendor granting anything. It is also the lever with the lowest execution risk: canceling a license that no one opened does not alter anyone's operation.

The complete method for subscription auditing, with the four scanning stages and the treatment of duplicate licenses, is in unused SaaS licenses: recover this spend. This section does not repeat that method, but rather positions it: it is the lever that finances the other three, because the cash it returns pays for the work of changing the channel and migrating the licensing model.

The limit is the size of the waste. A lean and well-governed base has few idle seats to recover, and in that case, the lever yields little. It also does not solve the underlying problem, which is fragmentation: fifty vendors with fifty idle seats is a management problem, not a licensing one.

4. Spend Consolidation: Fewer Vendors, More Volume Leverage

Consolidating spend means reducing the number of software vendors and concentrating purchasing volume, which changes the company's position at the table. The gain does not come from a discount requested vendor by vendor. It comes from the total volume becoming visible and negotiable as a block, and from the administrative effort of maintaining dozens of contracts decreasing along with the number of contracts.

Fragmentation exacts a toll in three places simultaneously. In price, because each vendor sees only their slice and none of them see the company's total spend. In operations, because each contract carries its own renewal, its own payment process, and its own due date. And in control, because spending without a formal purchase order does not enter anyone's budget and reappears at closing, which is addressed in software purchasing without a PO.

Consolidation is also the lever that gives bargaining power to the other three. With spend concentrated on fewer vendors, the usage reading for each tool becomes comparable, licensing model migration becomes a portfolio conversation, and channel negotiation ceases to be case-by-case. The internal policy that transforms these decisions into a rule is in SaaS purchasing policy for companies in Latin America.

The limit is the risk of concentration. Reducing too many vendors creates dependence on the remaining vendor, and this dependence again makes the next renewal more expensive. Healthy consolidation reduces the number of contracts without eliminating alternatives. It is also the slowest implementation lever, because reorganizing vendors requires renegotiating terms and migrating data, and this work does not fit into a single quarter.

Which Lever Yields the Most, and in What Order to Execute?

The interpretation of the table changes according to the company's situation. Those with many idle seats start with lever 3, because it finances the others with recovered cash. Those with few vendors and a large international contract tackle lever 1, because the import cost applies to the entire volume. Those with fifty vendors start with lever 4, because without spend visibility, previous decisions lack a basis.

The order of execution that works in most cases is this:

1. Audit the subscription base and cross-reference usage with value paid, to identify where the waste is.
2. Cancel all idle or duplicate licenses and record the cash freed up.
3. Measure the actual usage of each remaining tool, to size the correct block of licenses.
4. Consolidate the remaining vendors and review each one's licensing model.
5. Change the channel for international purchases that remain active, to eliminate exchange rate exposure.
6. Transform the five previous decisions into internal policy, with a review trigger before each renewal.

Figure 1: The recommended order of execution for SaaS cost reduction levers.

The order matters more than intensity. Executing all four simultaneously in a lean IT company halts the entire process, because each one requires a different owner. Executing in sequence, with lever 3 first, creates the cash and credibility that support the larger work of changing the channel.

Frequently Asked Questions

Is it possible to reduce SaaS costs without renegotiating with the vendor? Yes. Purchasing channel, licensing model, idle seats, and spend consolidation change the account structure without going through the negotiation table. Renegotiation is left for the end, when volume and usage are already visible, and the conversation with the salesperson is data-driven.

How much does it cost to import international software directly? Import charges and currency conversion increase the effective cost of an international solution by 50% to 70%. In practice, a US$100,000 invoice can amount to up to US$155,000 disbursed. The exact percentage varies with the type of operation and the tax classification of each remittance.

Does CIDE apply to imported SaaS? Yes. The 10% CIDE (Contribution for Intervention in the Economic Domain) (Law No. 10,168/2000, Article 2, Paragraph 2) applies to the import of SaaS because the Brazilian Federal Revenue Service formally classifies it as a technical service (Cosit Consultation Solution No. 191/2017 and Cosit Consultation Solution No. 99/2018). The non-incidence of Article 2, Paragraph 1-A, of Law No. 10,168/2000 only covers pure software licenses without technology transfer, a contractual and tax category distinct from SaaS.

Which lever returns money fastest? The recovery of idle seats. It is the only one that returns cash in less than thirty days, because canceling a license that no one uses does not depend on anyone's approval. The channel and consolidation levers take one to three months to show up in the results.

Do companies under the Lucro Presumido tax regime recover PIS/COFINS credit? No. The PIS/COFINS credit of 9.25% on the domestic invoice benefits only legal entities under the Lucro Real tax regime, operating under the non-cumulative system (Laws No. 10,637/2002 and 10,833/2003). For Lucro Presumido, the benefit of contracting locally is exchange rate lock, predictability, and operational simplification without customs bureaucracy. This system will undergo a transition starting in 2026 with the abolition of PIS/COFINS and the introduction of CBS/IBS under Complementary Law No. 214/2025.

References and Further Reading

• ABES/IDC, Brazilian Software Market: Overview and Trends, 2024 edition, with 2023 software origin data (software developed abroad: 73.7%; developed domestically: 24.8%).
• Brazilian Federal Revenue Service, Cosit Consultation Solution No. 191/2017, on the incidence of CIDE (10%) on Software as a Service (SaaS) remuneration as a technical service.
• Complementary Law No. 214/2025, which institutes the IBS and CBS, regulating the abolition of PIS and COFINS and the tax transition.
• Law No. 10,168/2000, Article 2, Paragraph 1-A, as amended by Law No. 11,452/2007, on the non-incidence of CIDE on software license fees.
• Law No. 11,452/2007, Article 20, which added Paragraph 1-A to Article 2 of Law No. 10,168/2000.
• ABES, Legal Guides, tax analysis of the software sector, including software import and remittances abroad.
• Brazilian Supreme Court (STF), Theme 914, RE 928943/SP, judged on 08/13/2025, on the constitutionality of CIDE-Technology under Law No. 10,168/2000.

The Lower-Cost Stack Wasn't Negotiated, It Was Recomposed

Renegotiation is the last lever, not the first, and there is a practical reason for this order. It is the only one of the five that puts the outcome in the vendor's hands, and therefore the only one the company can entirely lose. The other four produce an effect regardless of whether the vendor agrees, and that is why they affect the base of the expenditure while negotiation only affects the unit price.

The buyer who reorganizes the purchase before sitting at the table comes to the conversation with an argument the vendor cannot circumvent: actual usage, consolidated volume, and the alternative channel are already measured. The conversation shifts from being about a discount to being about how much of the total spend makes sense to maintain.

The lower-cost stack wasn't negotiated. It was recomposed.

This is the point where Nexforce Marketplace enters as a path, not as an argument. By receiving international contracts with invoices in Brazilian Reals, with an exchange rate lock, local payment, and PIS/COFINS credit for those under the Lucro Real tax regime (and full CBS and IBS credit under LC 214/2025), the company exchanges exchange rate exposure and import costs for a predictable value from the moment of purchase approval. The other three levers remain internal work, and it is internal work that sustains the result in the long term.