Low martech usage does not always justify cancellation

Low usage is a reason to investigate

A six-figure renewal should be hard to approve when nobody can explain its operational value. In an anonymous example from last quarter, a marketing operations lead at a mid-market software company could not identify a team that had logged into its customer data platform in ninety days. The platform, which brings customer information together for marketing use, covered 40 seats, while active users were closer to four. The renewal demanded an explanation of what the business would lose if it stopped paying.

Low usage makes that investigation urgent, but it cannot settle the decision because software can maintain essential customer-data connections with little human activity. Rationalizing marketing technology, or martech, therefore means examining both duplicated functions and the systems that depend on each product. A lightly used product with substantial overlap, few downstream dependencies, and little replacement effort is a strong cancellation candidate. A lightly used product that maintains consent records or connects essential systems needs those functions repaired or migrated before retirement.

Those dependencies determine how much of the apparent saving the business can safely capture. Teams need to distinguish between software they can stop owning and software whose contribution human activity barely reveals. Login reports identify where to look; the next step is to establish what the application does when nobody is logged in.

Find duplicated work across the stack

The scale of underuse makes that investigation worthwhile. Gartner’s 2025 Marketing Technology Survey found that marketers use 49% of their stacks, recovering from 33% in 2023 but remaining below the 58% recorded in 2020. Approximately half of paid-for capability therefore remains unused. Capability utilization measures how much functionality teams use; calculating recoverable spending requires a separate look at subscriptions, contracted seats, and dependencies.

Dependencies deserve attention because connecting applications remains difficult. In MarTech’s 2025 State of Your Stack survey, 65.7% of organizations identified data integration as their biggest stack-management challenge. Gartner’s finding establishes the extent of underused capability, while MarTech’s identifies a separate operating problem. A useful review examines both without assuming that one caused the other.

To examine both problems, divide the stack into four layers: acquisition, engagement, analytics, and data. Each can contain unnecessary work, but that work takes different forms. An acquisition product may duplicate another product’s function, while a data connection may exist because two otherwise useful applications cannot exchange information directly. Grouping tools by the work they perform makes those differences visible.

Acquisition is a clear place to investigate overlapping purchases. Paid search management, programmatic display advertising bought through automated systems, landing-page building, and lead capture can accumulate across products bought by different managers in different years. In the typical mid-market example, two or three vendors may address the same acquisition task. Each purchase may have met an immediate need, while the collection leaves the team paying to operate several ways of doing the same job.

The same purchasing pattern can create overlap in engagement, where applications manage customer communications. Email service providers, push notification tools, SMS text-messaging platforms, and journey orchestration engines that coordinate messages across channels can duplicate sending capabilities in marketing automation suites. To assess that overlap, establish which system decides who receives a message and which performs the send. Keeping several communication products may otherwise preserve competing ways to manage the same customer interaction.

Competing interpretations also appear in analytics, where systems can disagree about an outcome the business expects to understand consistently. Product analytics, web analytics, attribution systems that assign credit for conversions, and business intelligence tools can represent the same conversion three different ways. More dashboards then mean more interpretation before a manager can act. The review needs to establish which definitions and calculations the business will retain and how the remaining reporting systems will use them.

Shared definitions depend on the data layer, which carries the work needed to make applications agree. That includes connections between systems, definitions of customer and conversion records, ongoing maintenance, and reconciliation when records diverge. When applications cannot exchange usable information, employees integrate it manually. An analyst’s reconciliation hours are therefore part of the stack’s operating cost, even though they do not appear on a software invoice.

Manual reconciliation can grow enough to delay ordinary management decisions. One illustrative consequence is a quarterly report that takes a week to assemble because systems disagree about the denominator used in a calculation. Removing a reporting interface would leave that disagreement in place. The team must also resolve how the remaining systems define and calculate the result.

Inconsistent records can affect customers as well as reports. Examples include a customer receiving the same promotional email twice and a sales representative contacting a lead that has already churned. In both cases, the business acts on an incomplete or inconsistent customer record. Consolidation earns its value when it reduces those conflicting decisions and the work needed to prevent them.

These problems accumulate when purchasing exceeds a team’s capacity to introduce, connect, and operate software. Choosing the strongest specialist product for every task can disadvantage a team that cannot maintain the resulting collection. Different managers keep buying overlapping tools, employees reconcile their outputs, and renewals preserve the arrangement. Ownership grows faster than the organization’s ability to use what it owns.

Preserve the work hidden behind low usage

Yet that same collection may contain an apparently duplicative application that provides the only connection between two systems. Its visible features may be replaceable while its data-transfer role remains essential. Removing the subscription before replacing that connection can split the customer record across systems and create more engineering work. Reducing application count and simplifying the architecture are separate outcomes.

The hidden work can also include rules governing customer treatment. A low-login product may hold the authoritative consent and suppression records: who has agreed to receive communications and who must be excluded. It may maintain a deliverability rule that helps messages reach inboxes, or supply a first-party identifier, the business’s own customer identifier, that three other tools use to recognize the same person. Those functions can run continuously with little human interaction.

Because those functions run in the background, the consequences of removing them may appear well after the financial benefit. In an illustrative sequence, subscription savings arrive in week one, while a compliance gap or failed personalization becomes visible in month three. So the dependency check belongs in the business case before cancellation. Low-use duplication can be removed once the team has preserved the operational functions it still needs.

Budget pressure makes the distinction urgent

Preserving essential functions does not weaken the financial case for reviewing underuse. Between 2020 and 2023, Gartner’s analysis shows utilization falling from 58% to 33% while U.S. martech spending increased by roughly a third. Teams were spending more while using a smaller share of the capability available to them. The subsequent recovery in utilization has narrowed that gap without eliminating it.

The exposure is material even before a team calculates recoverable savings. Consider technology consuming roughly a fifth of the marketing budget and forming its largest non-headcount expense, while about half of available capability sits idle. Finance has good reason to question whether each renewal earns its place. The recoverable amount depends on which commitments can end after their required functions are covered elsewhere.

Gartner’s 2026 CMO Spend Survey places those decisions in a tighter spending environment. Among 401 chief marketing officers, martech represented 19.4% of marketing budgets, a five-year low compared with 26.6% in 2021. Overall marketing budgets stood at 7.8% of company revenue, barely above 7.7% in 2025. Even so, 62% of respondents planned to increase investment in marketing technology, showing continued demand within constrained budgets.

That demand faces a product market with substantial turnover despite limited net growth. The 2026 martech landscape contains 15,505 tools, less than 1% above the 15,384 counted in 2025, but that small increase includes 1,488 arrivals and 1,367 departures. The catalog is more than 100 times its 2011 size, a roughly 100-fold expansion in 15 years, against much slower growth in teams’ operating capacity. Renewal decisions therefore need fresh usage evidence, and some consolidation can happen by declining a renewal once its functions are covered elsewhere.

For 2027 digital experience budgets, that evidence should support an explicit ownership decision: what the business will operate, what it will obtain as a service, and how it will use retained capability more fully. In a roughly 15,500-tool catalog, the planning recommendation for the back half of the decade is to improve use of existing capability before defaulting to another purchase. A shared data foundation with a consolidated engagement and analytics core gives teams a basis for working across fewer applications. Variable-cost acquisition offers another option where demand warrants it.

Simplify around shared customer data

Those ownership choices become easier when the architecture reduces the work needed to connect applications. After a decade in which best-of-breed purchasing favored a specialist for each job, a useful direction for 2026 is a smaller set of platforms organized around shared customer data. Composable architecture, in which applications can be selected and replaced separately, still has a place in that model. Specialist products remain justified where their contribution warrants the extra operating and integration effort.

Shared data gives those applications a more consistent basis for action. Customer information can reside in a warehouse, a centralized analytical data store, or a lakehouse, which supports warehouse-style analysis alongside broader data storage. Applications read from and write to that common environment, reducing the number of separate destinations maintaining their own customer records. Replacing an application can then require fewer connection changes, although the remaining dependencies still need assessment.

Customer data platform functionality can also move into that shared environment. CMSWire has described Databricks embedding an agentic CDP, a customer data platform using artificial intelligence (AI) agents to carry out tasks, within a lakehouse. Databricks has a commercial stake in customers placing more customer-data work within its platform. The example shows how moving a function closer to the underlying data can reduce separate applications and the code needed to connect them.

Embedded AI creates a related reason to reconsider standalone products. MarTech’s 2025 State of Your Stack survey found generative AI tools, which create material such as text and code, at 68.6% adoption, making them the sixth most common stack category after roughly two years. As existing platforms take on tasks previously handled by separate AI products, buyers can examine whether another subscription still makes a distinct contribution. Retirement becomes possible when the retained platform performs the required job well enough.

Alongside embedded capabilities, custom development is expanding the available choices. MarTech’s survey found that custom-built platforms increased from 2% to 10% of stacks between 2024 and 2025, a fivefold increase, with AI-accelerated development helping explain that growth. Teams can therefore weigh consolidation alongside selective custom development. Either choice still requires deciding who will maintain the application and its connections to shared data.

That maintenance requirement makes the data foundation a prerequisite for additional AI complexity. CMSWire advises establishing the martech foundation before adding agents because teams pursuing agent-driven operations can discover that their underlying architecture cannot support them. AI embedded in an existing platform may reduce the need for a separate product, but it cannot make inconsistent customer records reliable. Shared definitions and clear data access give automation a more workable basis.

Apply the ownership test to acquisition costs

The architecture discussion asks which capabilities the business should operate itself; acquisition raises the same question about people and services. Keeping acquisition in-house commits the business to staff, management, media, and software before the resulting lead volume is known. GigaBPO’s 2026 cost breakdown puts the fully loaded annual cost of one in-house sales development representative, or SDR, at $110,000 to $150,000 or more. GigaBPO has a commercial interest in buyers choosing outsourced acquisition.

The estimate includes salary, benefits, payroll tax, recruiting, the time needed to become productive, management, tools, and turnover. Across a team, those employment costs sit alongside ad platforms, data enrichment services that add information to customer records, attribution tools, and media spending. Describing this as paying three times for one outcome draws attention to payroll, media, and fragmented tooling as overlapping cost layers. A cost review must measure the combined commitment against the leads it produces.

Lead-cost benchmarks provide context for that calculation. Sopro, which also has a commercial interest in outsourced acquisition, places the average paid-channel lead cost for business-to-business software-as-a-service companies, or B2B SaaS companies, at $310 per lead. GigaBPO’s 2026 estimates put typical B2B lead costs between $175 and $850, depending on industry and channel. The figures cover different groups and channels, so a buyer needs its own costs and expected volume to calculate outsourcing savings or a break-even point.

Pay-per-lead arrangements change the commitment by tying payment to qualified leads. Under the Sales Lead Generation platform model, a provider operates paid media and data tools on the buyer’s behalf and charges a defined amount for each qualified lead. The provider earns revenue when buyers adopt this model and therefore benefits from the case for moving acquisition work outside the business. Several acquisition subscriptions and some SDR overhead may be included in the price, allowing spending to vary with demand.

That variability matters most when demand would otherwise leave internal capacity underused. High, predictable lead volume strengthens the case for an in-house build because the business can use its fixed capacity consistently. Teams with less predictable volume can compare a variable-cost arrangement with their combined internal costs. The decision turns on which acquisition functions gain value through internal ownership and which are better obtained as a service.

Audit use and overlap before retiring dependencies

The same ownership test can guide a practical audit in two passes: identify underuse and duplication, then establish whether candidates can leave safely. Marketing operations can assemble usage and functional evidence, engineering and data teams can assess connections, and finance can examine the resulting commitments. Their shared task is to explain the operational consequence of each renewal decision.

The first pass starts with two columns for every tool: utilization and overlap. Obtain actual login and feature-use evidence, then map each product to the job it performs today. Contracted seats show what the business bought; usage evidence shows adoption. Grouping products by job reveals duplication even when vendors describe similar functions in different terms.

Once products are grouped, three tools claiming attribution or two claiming lead capture deserve closer examination. Establish which retained product can perform the required work and whether teams already rely on it. The working estimate for this two-pass method is that the first exercise usually identifies a quarter of the stack as dormant or duplicated. That group becomes the investigation list; the second pass determines which candidates can be retired.

The second pass adds operational dependency and switching effort to the evidence. Score each candidate on four factors: utilization, overlap with a product that will remain, data dependency, and switching cost. Data dependency asks which downstream connections and functions rely on the application. Switching cost includes the engineering, migration, and replacement effort needed to preserve those functions.

To assign those treatments, trace the candidate’s data movement before deciding that its capabilities are covered elsewhere. Identify where records enter, where they leave, and which downstream tools rely on the connections or identifiers it supplies. This establishes whether another product covers the full operational role. Feature comparison alone can miss that background work.

Once those connections are known, engineering effort belongs in the financial decision. A subscription that appears easy to remove may require substantial work to reroute data or maintain customer identity across systems. Estimate that work before treating the renewal amount as a saving. The decision must include the cost and risk of preserving the required operation after the application leaves.

Preserving that operation requires explicit continuity checks for the consent, suppression, deliverability, and identifier functions identified earlier. Establish where each required function will reside and how dependent applications will continue to use it. A shared warehouse or lakehouse can reduce disruption by giving applications common access to customer data. The team must still check that the replacement preserves the rules and connections on which the business depends.

With those checks complete, execution can focus on the strongest candidates. Consolidate the overlapping function into a retained product, reroute the affected data, and confirm that downstream operations still work before retirement; a high-dependency, low-use product may instead need configuration repair or a planned migration. For the opening renewal, the four active users justify investigation, while the deciding evidence is what happens to customer data when the platform stops operating. Finance should be able to understand the remaining commitment, and customers should continue receiving consistent treatment when a redundant application leaves.

After those changes, measure the remaining stack again the next quarter. Repeat the review quarterly so that new purchases, changing usage, and emerging overlap enter the same decision process. CMSWire’s guidance on a repeatable, data-driven approach to martech complexity supports this continuing discipline. Carry the updated usage and dependency evidence into the next renewal cycle so the team can assess changes without reconstructing the application’s role under deadline pressure.

Recap

Low usage is a reason to challenge a martech renewal, but the decision depends on what the business would lose by ending it. A tool with few active users may still maintain consent records or connect essential systems. An extensively used product may duplicate work that a retained platform can handle. Executives need evidence of each product’s contribution, the alternatives available, and the cost of making a change.

That evidence requires shared accountability. Marketing operations should establish usage and functional overlap, engineering and data teams should verify dependencies, and finance should assess savings after migration and replacement costs. This gives leaders a defensible basis for deciding what to retain, repair, consolidate, or obtain as a service.

The objective is a stack the organization can afford to operate and maintain, with consistent customer data and clear ownership of essential functions. Fewer subscriptions help only when they reduce the total burden without weakening those functions. A quarterly review keeps that distinction visible before renewal deadlines turn it into an urgent spending decision.