Licensing Assurance 2026 — LATAM
The new discipline of the digital asset in Latin America. How the 2026 enterprise licensing paradigm affects EBITDA for mid-sized and large companies in the region — and what to do before July 1.
Tags: Microsoft, Oracle, LATAM, Software Licensing, Enterprise, FinOps
The new discipline of the digital asset in Latin America
Executive Whitepaper · Vol. 01, No. 05 · May 2026 · Guillermo Sandí · Principal
LATAM Panorama · Raw 2026 Data
Six figures define the context before the argument begins. This is the operating terrain where the next three-year contract is signed — or postponed.
- 391% — ROI from digitizing frontline workers in Mexico and LATAM. Source: Nextcore · Microsoft.
- 137B — Cyberattack attempts recorded in LATAM in one semester. Source: Fortinet FortiGuard · Tigo Business.
- $2B — Medical device exports · Costa Rica 2023. Source: PROCOMER · Coyol Free Zone.
- 340% — Growth in LATAM fintech startups since 2017. Source: Facephi · Latin America.
- 60% — Increase in automated intake of banking requests. Source: Bantrab / AuraQuantic · Guatemala.
- 30% — Reduction in KYC onboarding time with LLMs in production. Source: GJR Labs · Near Coding · Costa Rica.
Executive Summary — What changed, why it matters, and what to do before July 1
The implicit contract that governed the relationship between Latin American enterprises and their software providers for two decades was rewritten quietly. There was no announcement. There was no negotiation. What used to be a predictable budget line — the Enterprise Agreement, the three-year SaaS contract, annual maintenance — became a stochastic variable governed by user behavior, autonomous agents, and consumption algorithms that no native ERP records with precision.
Three vectors drive this change, and they operate simultaneously. Organizations that do not model it in 2026 will discover it in their P&L in 2027, when negotiation leverage with the vendor is practically gone.
The first is structural: consumption-based licensing has replaced seat-based licensing as the dominant model. Tokens, AWUs, Flex Credits, Assists — four different accounting grammars, four separate pools, four sources of monthly variance. The CFO who signed a three-year EA with nominal predictability is now managing IT spend whose real variance may be 2x or 3x the estimate.
The second is pricing: on December 4, 2025, Microsoft activated its largest commercial-line adjustment since 2022, effective July 1, 2026. Increases range from 5.3% on E5 to 43% on frontline F1 SKUs. For a regional bank with 3,000 F3 employees, a retailer with 5,000 point-of-sale seats on Business Basic, or a BPO with 2,000 F1 agents, the impact is not theoretical: it is an additional annual line item of USD 200,000 to 600,000 that nobody budgeted.
The third is FX: in Costa Rica, the CRC/USD band between January and May 2026 was 9%. In Colombia, post-election peso/dollar volatility introduces a political factor into the calculation. A company that models software spend in local currency but contracts in USD carries unquantified FX exposure in its TVO.
A fourth variable sits on top of those three, and no channel partner has an incentive to reveal it: the structural conflict of interest in the VAR/ISV model. Forrester documented it more than fifteen years ago. The difference between the optimal portfolio for the client and the portfolio that maximizes channel revenue can represent 18% to 43% of recurring enterprise software spend.
The central thesis is simple: in 2026, license management stopped being a procurement function. It is financial strategy. Organizations that treat it as such will capture structural margin. Those that do not will transfer that margin in full to their vendors — permanently, with no retroactive recovery.
§ 01 — Diagnosis: Why SAM is no longer enough
For two decades, Software Asset Management in Latin America was reactive: respond to audits, reconcile inventories, maintain contractual compliance. A perimeter discipline, housed in procurement or the CIO office, with modest budget and annual reporting. That conception did not evolve slowly. It expired all at once.
The reason is that the central object of that management — the license — dissolved.
A software license used to be a discrete right to use an installable binary. It was measured in users and devices. It was audited by counting installations and renewed on annual or three-year cycles. The model was operationally tedious but financially predictable. A Microsoft EA, an Oracle contract, an Adobe subscription: known numbers, known dates, predictable contractual increases.
That model no longer exists. What replaced it is a heterogeneous bundle of rights over infrastructure, generative models, autonomous agents, connectors, and orchestration capabilities — each with its own cost function. A single corporate identity in 2026 can simultaneously activate: a Microsoft 365 E5 seat with a monthly recurring charge, Copilot consumption billed against an entitlement pool, an Agent 365 invocation that draws from a governance pool, a Salesforce Agentforce session that consumes Flex Credits at USD 0.10 per action, a ServiceNow Now Assist query that draws from the tenant’s annual Assist allowance, and an API call to Claude Opus 4.7 at USD 5 per million input tokens and USD 25 per million output tokens. Each component is billed in a different unit, against a different balance, with different overage rules, and with delayed visibility.
The result is what the practice calls financial Total Value of Ownership (TVO): a metric that no longer stops at acquisition and maintenance cost, but incorporates consumption variance, FX risk, internal demand elasticity, opportunity cost from shelfware, and the security premium associated with the use of unmanaged generative models.
The FinOps Foundation reported in 2025 that 75% of mature organizations already apply FinOps principles to SaaS and AI spend, not only to IaaS. The premise is the same one that drove FinOps in cloud a decade ago: granular visibility, business-unit attribution, iterative optimization, and shared governance across Finance, IT, and Operations. The perimeter changed. The principle did not.
For the board of any mid-sized or large company in Latin America, the implication is concrete: an annual licensing review cycle is not insufficient — it is dangerous. A firm that reviews its portfolio once a year discovers deviations six to twelve months late. By then, the financial damage is irreversible and negotiation leverage with the vendor is practically zero.
Bantrab, Guatemala — 60% increase in automated request intake after digitizing physical processes with AuraQuantic. Reactive SAM did not enable that transformation; continuous governance did.
§ 02 — Four accounting grammars: Tokens, AWUs, Flex Credits, and Assists
The consumption licensing model emerged in four primary variants. Each has its own accounting grammar. Together, they produce the same effect on the P&L: unbudgeted variance in a line item the CFO believed was controlled.
Tokens — the atomic unit of generative compute
In May 2026, pricing for frontier models is public and verifiable:
| Model | Input · USD/M tokens | Output · USD/M tokens |
|---|---|---|
| Claude Opus 4.7 · Anthropic | 5.00 | 25.00 |
| Claude Sonnet 4.6 | 3.00 | 15.00 |
| GPT-5.4 · OpenAI | 2.50 | 10.00 |
| GPT-5.2 Pro · reasoning | 21.00 | 168.00 |
| Gemini 3.1 Pro · Google | 2.00 | 12.00 |
| Gemini 2.5 Flash-Lite | 0.10 | 0.40 |
| DeepSeek V3.2 | 0.14 | 0.28 |
The paradox of the model: price per token falls while total consumption scales. Reasoning models bill every token of hidden thinking at output-token price. A single GPT-5.2 Pro query can generate 50,000 output tokens before returning one paragraph. For the CFO, unit price is a deceptive metric. What matters is cost per business outcome — a metric no native ERP records.
30% reduction in KYC onboarding time in Costa Rican fintechs. GJR Labs and Near Coding already integrate models like Claude and GPT into production back office and customer service workflows. LLM adoption in the region is not a projection: it is measured operation.
AWUs — Agent 365 and Microsoft 365 Copilot
Microsoft consolidated its agentic model around the concept of the Agentic Work Unit (AWU) and the Agent 365 control plane, integrated within M365 E7 Frontier Suite. The unit combines fixed monthly allocation with metered consumption for long-running tasks. Agent 365 is also available as a USD 15/user/month add-on for companies that have not yet migrated to E7. IDC projects 1.3 billion AI agents in circulation by 2028. For the CIO, the implication is direct: every deployed agent is an active cost unit that traditional procurement did not contemplate.
Flex Credits — Salesforce Agentforce
The 2026 consolidated model replaces the “$2 per conversation” scheme with a granular metric: 100,000 Flex Credits cost USD 500; each standard action consumes 20 credits (USD 0.10); voice actions consume 30 credits (USD 0.15); any action processing more than 10,000 tokens counts as multiples. A complex conversation consumes between 5 and 15 actions — a real cost between USD 0.50 and 1.50, with the possibility of exceeding USD 2.00 in multi-step reasoning. The critical assumption most prospects do not validate: Conversations and Flex Credits models cannot coexist in the same Salesforce organization. The choice is structural and costly to reverse.
Assists — ServiceNow Now Assist
ServiceNow bills every generative AI action against an annual Assist allowance. An incident summary consumes 1 Assist; creating an application may consume 20. The AI Starter Pack includes 25 ITSM Pro Plus users with 150,000 total Assists. The opacity of the model is documented: ServiceNow does not publish prices and, according to analyses from UpperEdge and Redress Compliance, the uplift over base ITSM to activate Now Assist ranges from 25% to 60%. In the Foundation/Advanced/Prime cycle of April 2026, the model incorporated retroactive increases when a user on a Standard license invokes Pro functionality.
Attention point · Gartner: In early 2026 Gartner estimated that 85% of organizations underestimate their AI costs by more than 10%. The emblematic case is Uber: Claude Code adoption scaled from 32% to 84% of its 5,000 engineers between December 2025 and March 2026, exhausting the annual AI budget in April. For LATAM, replicating that experiment is not risky. It is financially unviable.
§ 03 — The structural reset: Microsoft, Salesforce, ServiceNow, Google
Microsoft 365 is not just another IT budget line in Latin America. It is the operating infrastructure for millions of knowledge workers in the region. In banks, BPOs, insurers, agribusinesses, and Latin American retail chains, total operational dependency on M365 exceeds 70% of the workforce. In frontline roles — contact centers, manufacturing plants, retail cashiers, field technicians — it reaches 100%: if M365 does not work, the employee cannot produce.
When that cost rises between 5% and 43% in a single round, the blow does not land on an optional luxury. It lands on the broadest base of the payroll.
The structure of the adjustment — effective July 1, 2026
| SKU | Previous price | Jul-2026 price | Change |
|---|---|---|---|
| M365 Business Basic | USD 6.00/u/mo | 7.00 | +16.7% |
| M365 Business Standard | 12.50 | 14.00 | +12.0% |
| M365 Business Premium | 22.00 | 22.00 | 0% |
| Office 365 E1 | 10.00 | 10.00 | 0% |
| Office 365 E3 | 23.00 | 26.00 | +13.0% |
| Microsoft 365 E3 | 36.00 | 39.00 | +8.3% |
| Microsoft 365 E5 | 57.00 | 60.00 | +5.3% |
| M365 F1 frontline | — | — | up to +43% |
| M365 F3 frontline | — | — | up to +29% |
| Windows Enterprise per device | — | — | +31% |
What it means by industry
Riopaila Castilla, Colombia — 27% growth in operating income — COP $1,508,683 MM driven by sugar, alcohols, molasses, and energy, with 2,324 workers distributed between urban headquarters and rural plants in Valle del Cauca. This is the exact profile: 4,000 mixed employees on E3/F3, with the +25% F3 adjustment hitting the broadest base of the operation.
Profile · BPO / contact center (2,000 F1 agents, no Teams): the 43% increase over a monthly baseline of roughly USD 44,000 represents USD 19,000 additional per month — USD 228,000 per year unbudgeted.
Profile · Central American retailer (5,000 point-of-sale seats on Business Basic): the 16.7% increase over an annual baseline of roughly USD 360,000 represents an additional USD 60,000. Multiplied across five countries with separate contracts, the figure exceeds USD 200,000.
Profile · Colombian agribusiness (4,000 mixed employees · E3 + F3): the combined impact of +8.3% on E3 and +25% on F3, over an estimated annual baseline of USD 1.8M, produces a delta of USD 150,000 to 280,000 depending on seat distribution.
The Golden Window — the period between the current date and June 30, 2026 — allows companies, through an early renewal under NCE or EA, to freeze current prices for twelve or thirty-six months. For companies renewing after July 1, the adjustment is full and immediate.
The five large providers have synchronized a pricing move that the Latin American economy will absorb with a delay. The client that renews before June 30 with independent advisory protects EBITDA through 2029. The one that renews later absorbs the full jump.
§ 04 — Frontier Suite: Microsoft 365 E7, anatomy of the most expensive decision of 2026
Microsoft 365 E7, announced on March 9, 2026 and available since May 1, marks the first new enterprise tier since E5 launched in 2015. At USD 99 per user/month, E7 unifies four components previously sold separately: full Microsoft 365 E5; Microsoft 365 Copilot with Work IQ; Microsoft Entra Suite; and Microsoft Agent 365 — the control plane for AI agent governance, observability, and security. Purchased separately, the four components exceed USD 117/u/month.
The real math by profile
CIO · Central American enterprise (1,000 employees on E5): current annual cost of USD 684,000. The jump to E7 implies USD 1,188,000 — a delta of USD 504,000, equivalent to a 73% increase. If Copilot is used only in sales and marketing, buying E7 across the board destroys 40% to 50% of the value paid.
CFO · regional bank (3,000 employees across E3 and E5): independent analysis reveals that 60% of users will not activate 70% of E7 functionality over the next 24 months. The cost of that unrealized value promise: USD 800,000 to 1.2M annually in paid but unused capabilities.
IT Director · agribusiness (4,000 frontline employees): E7 does not exist for them. F1/F3 SKUs do not have access to the tier. Pressure toward E7 is, in this case, an irrelevant conversation. The F1/F3 increase of 25%-43% is urgent and real.
The right decision — three exercises before signing
The E5 → E7 decision is not solved with the Microsoft representative’s spreadsheet, whose incentive is total suite booking. It requires three exercises that no channel reseller has an incentive to complete honestly: (1) real adoption matrix by role, (2) AWU/Agent 365 projection with scenarios calibrated against LATAM market benchmarks — not vendor projections — and (3) negotiation of protection clauses: ramp-down rights, model swap rights, and exit clauses with data portability and no penalty.
Committing to three years on E7 without exit clauses is a technology bet in an environment that changes every 18 months.
Productivity increase in complex tasks in Guatemalan banking — 50%. Bantrab reduced product delivery times from 48 hours to 30 minutes by selectively integrating Azure OpenAI and Salesforce, without migrating the entire organization to a single bundle. The lesson: surgical adoption beats the bundle in real ROI.
E7 does not arrive as a billing line. It arrives as a three-column proposal designed to look equivalent in year one. The next section decomposes that mechanism through a Central American BPO case: how equivalence is built, how it erodes, and how much it is worth in colones under an adverse fiscal-cycle exchange rate.
§ 04.5 — Anatomy of the MEO: The three-column offer and the trap for a Central American BPO
The Multiple Equivalent Offer is not a pricing proposal. It is a leverage-removal mechanism: the year-one discount is what Microsoft pays the client to disarm itself before the next negotiation.
The proposal arrives late — almost always later than is convenient — and in presentation format. The deck mentions Frontier Firm, Digital Labor, Agentic AI, and Copilot Augmentation. The buzzwords do the work numbers alone cannot: no CIO wants to be responsible for leaving the organization behind the competition in AI. Fear of technological lag is part of the pressure mechanism.
In the room sits the company’s Microsoft Gold partner — the same one that has managed the EA for three years and that, according to the compensation model documented by Forrester, earns commission on what it sells, not on what it saves the client. Its incentive in that meeting is not the same as the CFO’s incentive across the table.
The scenario: Central American BPO · 2,000 agents on M365 F1
The company: BPO headquartered in Guatemala, with operations in Costa Rica and El Salvador. 2,000 agents on F1 licenses (contact center, no Teams). 120 knowledge workers on E3. 15 executives and senior supervisors on E5. EA renewal in October 2026. The partner presents the MEO in May, citing urgency from the July price adjustment.
The deck arrives with three columns. Left column: renewal “as is” — F1 at Level A list price with the 43% adjustment already included, E3 at list, E5 at list. No discount, no argument. This column exists so the other two look generous. Middle column: upgrade to F3 + E5 for all knowledge workers — presented as the “productivity” path. Right column: selective migration to E7 for the 135 highest-value users — presented as the “AI” path. In year one, all three columns land at roughly the same number. That is the first artifice.
What the deck does not show: year three, year four, and year five — when the initial discount has eroded to zero and the client is trapped in a stack dependency that replaced its alternatives during the contract term.
The table the partner never quite shows
The following table reconstructs five years across the three MEO paths. Values indexed to the current baseline (F1 at $2.50/u/month · E3 at $23/u/month · E5 at $57/u/month, pre-July prices).
| EA year | Path A · Current renewal | Path B · F3+E5 upgrade | Path C · Selective E7 migration |
|---|---|---|---|
| Year 1 | ~USD 640K | ~USD 638K | ~USD 641K |
| Year 2 | USD 640K | USD 748K | USD 810K |
| Year 3 | USD 640K | USD 858K | USD 1.01M |
| Year 4 | USD 640K | USD 940K | USD 1.18M |
| Year 5 · Renewal | USD 640K | USD 1.02M (+59%) | USD 1.41M (+120%) |
Year 5: discount eroded to 0%. The client renews at full list with zero alternative leverage because the MEO displaced its vendors during the term. Column A, presented as the worst option, turns out to be the least damaging.
FX dimension for the Guatemalan BPO with Costa Rican operations
On the Path B commitment in Year 5 (USD 1.02M), the CRC/USD band from January-May 2026 (₡451-₡497) produces a swing of ±₡46M on the colones invoice. In an adverse scenario (+15% over the average), that commitment equals ₡589M annually — versus ₡310M for the current baseline.
Three moves to dismantle the trap
01 · Reset the “as is” column. Require the base column to include the real discount earned from volume and relationship tenure. That adjustment collapses year-one equivalence and forces the deal desk to decide how much additional discount it is willing to finance.
02 · Not-to-Exceed clause on the next renewal. An NTE clause written into the current contract, limiting the permitted increase over the term’s exit price in the next renewal — typically a 3%-5% annual cap.
03 · Decouple displacements from the discount. Every MEO displacement — Defender, Entra, Purview, Teams — must be evaluated as an independent architecture decision, with its own Proof of Concept, total cost analysis, and migration calendar.
The discount is real. The trap is real too. Both are true at the same time. A client that sees only one of the two will sign an agreement whose true cost does not arrive until the next renewal.
Documented return per dollar invested in independent pre-signature advisory — 8 to 25 times. The range consolidates public cases from Redress Compliance, US Cloud, and Softimiza Colombia in the Latin American mid-market segment.
§ 05 — Macro and FX: Exchange rate, free trade zones, and real cost
For any Latin American company that contracts software in USD but earns revenue in local currency, exchange rate is an IT variable, not only a finance variable. A TVO model that does not incorporate FX sensitivity is incomplete.
In Costa Rica, the CRC/USD band between January and May 2026 was 9% (₡497 to ₡451). On an annual licensing commitment of USD 5M, that band represents a colones swing close to ₡300 million.
Coyol Free Zone, Costa Rica — 40% reduction in lead times in medical manufacturing. In 2023, medical devices generated USD 2.0 billion in exports, surpassing coffee and bananas as the country’s primary source of foreign exchange. Software is not a support cost: it is export infrastructure.
The TVO identity Deep Sight Consulting uses to model real cost
C_local = P_usd × E × (1 + i)
- C_local · total executed cost in local currency
- P_usd · negotiated price in dollars
- E · effective exchange rate weighted by payment window
- (1+i) · uplift factor for unaccounted consumption
In unmanaged environments, the factor (1+i) ranges from 1.15 to 1.35: realized cost exceeds budgeted cost by 15% to 35%. In environments with FinOps and Licensing Assurance discipline, the factor falls to 1.02-1.05.
§ 06 — Tokenmaxxing: The financial and cultural crisis of AI adoption without governance
Tokenmaxxing — AI token consumption above any business need, driven by internal incentives that reward usage over value.
The phenomenon was first documented at Meta, where 60.2 trillion tokens were consumed in 30 days. At Salesforce, internal widgets showed every engineer their monthly “minimum expected spend.” Uber is the most verified case: 95% of its 5,000 engineers using AI monthly, 70% of committed code generated by AI, the full annual budget consumed in four months.
The crisis has three dimensions. Financially, it turns a contained spend variable into an explosion with no modeled ceiling. Culturally, it distorts productivity incentives, confusing token consumption with value delivery. Strategically, it drains resources from legitimate initiatives into unproductive consumption.
For LATAM, the risk is asymmetric. A Colombian fintech with 300 employees that implements customer-service agents on Agentforce without budget circuit breakers can triple its software spend in one quarter.
The interventions that mitigate tokenmaxxing are technical, not only cultural: intelligent routing, prompt caching, batch APIs with 50% discount, budget circuit breakers, and Inference Yield as a KPI — measuring value per token, not tokens per user.
Price per token falls while total consumption scales. That is not a paradox — it is the business model.
§ 07 — VAR / ISV: The structural conflict of interest in the channel model
The reseller earns commission on what it sells, not on what it saves the client. Forrester documented it more than fifteen years ago. The mechanics have not changed.
The consequences are verifiable across four planes: portfolio composition, negotiation, audit, and innovation. A Microsoft Gold partner will not propose Google Workspace or Notion even when either is the optimal solution.
Public cases measure the magnitude: USD 1.2M recovered from a Fortune 500 EA with US Cloud; USD 12M in unused Azure MACC recovered by Redress; USD 7.2M avoided in a Copilot rollout through a controlled pilot of 2,000 users instead of the proposed 25,000; USD 5.7M in an Oracle ULA through Softimiza Colombia; USD 3.7M in a corporate SAP contract.
For every dollar invested in independent advisory, clients capture between 8 and 25 dollars of net savings in the first year.
§ 08 — Licensing Assurance Methodology: the three pillars
The methodology operates on three interdependent pillars. None works in isolation; all three are necessary to capture the full savings.
Pillar I · Continuous Compliance
Permanent visibility between what is contracted, what is deployed, and what is consumed. Continuous inventory, utilization rate by SKU, proactive true-up management, audit defense, and Shadow IT governance. A firm that reviews its portfolio once a year discovers deviations six to twelve months late.
Pillar II · Digital Asset Perimeter Security
Misprovisioned identity and security vulnerability are the same problem. Right-sizing by role and privilege, agentic governance, DLP policy in prompts, and API key segregation. In Costa Rica, crossed with Law 8968 and PRODHAB; in Colombia, with Law 1581 and the SIC.
Pillar III · Operational Continuity
A poorly governed software portfolio creates single points of failure. Exit and portability clauses, multi-vendor coverage by critical category, alternative support with potential savings of 50%-70%, and continuity plan for provider failure.
§ 09 — Savings arithmetic: documented ROI by company profile
Profile 1 · Central American mid-sized enterprise (800-2,000 employees · USD 500K baseline): aggregate release of USD 90,000 to 215,000 annually — 18% to 43% of baseline.
Profile 2 · Multinational subsidiary or large corporation (2,500-5,000 employees · USD 1.5M baseline): aggregate release of USD 270,000 to 645,000 annually.
Profile 3 · Large multinational / regional hub / tier-1 banking (5,000+ employees · USD 5M baseline): aggregate release of USD 900,000 to 2.15M annually.
§ 10 — EA/NCE tactics for 2026
The window between May and June 2026 is the most critical Latin American CIOs have seen so far this decade.
- Pre-July 2026 lock-in. Negotiate early with a new term closed before June 30.
- Right-sizing by role. Optimal distribution: E5/E7 in 8%-12%, E3 in 35%-45%, F1/F3 in 35%-45%.
- Controlled Copilot pilot. 8%-12% of the workforce for six months with adoption KPIs before scale-up.
- Quantitative fair-use clauses. In every offer that mentions “unlimited,” require a quantitative definition of “typical use.”
- Model swap clauses. Right to reroute consumption between Claude/OpenAI/custom Foundry models without penalty.
- Vendor fiscal calendar. Synchronizing signature with the vendor’s fiscal window generates an additional 5% to 15% over the standard discount.
- Agentic governance from day one. Maximum monthly pool by business unit, owner by agent, circuit breaker at 110% of quota.
§ 11 — Implementation roadmap: 90 days · 6 months · 12 months
Phase 01 · Days 1-30 · Diagnosis and baseline. Automated discovery of SaaS/EA/cloud state. TVO baseline in USD and local currency. Quick wins identified (typically 5%-8% of baseline in immediate cancellations).
Phase 02 · Days 31-90 · Quick wins and renewal preparation. Cancellation of inactive licenses. Right-sizing. Shadow IT recovery. Renewal calendar analysis with pre-July 2026 priority.
Phase 03 · Months 4-6 · Master negotiation and new contract. Financial modeling with three scenarios. Vendor-specific negotiation playbook. Contract close with price lock, fair use, model swap, ramp-down, and exit clauses. Documented reduction of 18%-30% against baseline.
Phase 04 · Months 7-12 · Permanent governance. Quarterly Licensing Assurance committee with CIO, CFO, and business units. Financial KPIs published monthly.
Phase 05 · Year 2 and beyond · Permanent function. Incremental savings in years 2 and 3 range from an additional 5% to 12% of baseline, versus the natural 8%-12% annual increase accumulated by unmanaged firms.
§ 12-13 — Deep Sight’s three engagement modalities
Deep Sight Consulting does not sell licenses from Microsoft, Salesforce, ServiceNow, Oracle, Adobe, or any ISV. We do not receive channel discounts. We do not have preferred partnerships with manufacturers. Our only compensation comes from the client, in one of three modalities:
Modality I · Closed project. Initial diagnosis (Phases 1 and 2), with fixed fees in USD payable in local currency at the day’s exchange rate. Duration of 90 days, predefined deliverables, no scope surprises.
Modality II · Success fee on verified savings. 8%-15% of documented net savings during the first 12 months, with joint third-party audit. The client does not pay if there is no measurable savings.
Modality III · Monthly Licensing Assurance retainer. For organizations that adopt the methodology as a permanent function. Guaranteed response SLA and participation in every portfolio renewal.
Every engagement begins with a 90-minute confidential session. We map the current portfolio, identify the highest-exposure points, and build a preliminary savings hypothesis. That session does not generate an invoice.
Closing — July 1, 2026
July 1, 2026 is not a date on Microsoft’s fiscal calendar. It is the moment when the asymmetry between those who acted and those who waited becomes permanent.
Companies that renew earlier absorb the adjustment with protection for 12 or 36 months. Those that renew afterward absorb the full jump without cushioning.
What changes with Licensing Assurance is not an event. It is a posture. The posture of an organization that stops being a passive buyer of software and becomes an informed actor that negotiates with data, governs with metrics, and protects EBITDA with the same discipline it applies to any other financial asset.
The difference between an informed and an uninformed decision is measured, in 2026, in millions of dollars per year. Deep Sight Consulting exists so that difference happens in your favor.
Epistemic posture
✓ Verified against canonical source: Microsoft price adjustment effective 1-Jul-2026 · M365 E7 Frontier Suite (9-Mar-2026) · Salesforce Flex Credits · public LLM model pricing · Uber Claude Code case · FinOps Foundation 2025 · Forrester on VAR/ISV channel.
◐ Inferred from operating context: Savings ranges by company profile · specific tokenmaxxing percentages · ServiceNow Now Assist uplift (25%-60%) · optimal SKU distribution · 2026 FX sensitivity.
◯ Undetermined: Client-specific renewal calendars · DAAP contractual terms in regional LATAM context · E7 adoption velocity in Central American mid-sized enterprises.
Deep Sight Consulting has no commercial partnerships with Microsoft, Oracle, SAP, Salesforce, or ServiceNow. It does not sell, resell, or receive commission on software licenses from any manufacturer. DS · LA-26.05 · MS-FY26 · deepsightconsulting.com