Your Employees Do Not Need Another AI Tool

“Login counts and license usage don't measure ServiceNow adoption. # How to Measure ServiceNow Adoption — and Why Most Organisations Are Measuring the Wrong Thing By Iconica Editorial, Iconica 8 min read · Updated July 2026 Table of contents Summary Most organisations measure ServiceNow adoption by counting logins and licenses consumed — numbers that rise steadily right up until the platform quietly stops delivering value. Ask a platform owner how adoption is going on their ServiceNow instance, and the answer usually arrives as a number: 85% of licensed users logged in last month. Ticket volume is up. The change management module has been rolled out to three more business units. None of those numbers answer the question that was actually asked Adoption is not usage. A user who logs in once a week to close out a ticket they were forced to raise is not "adopted" in any meaningful sense — they are complying. A department that uses ServiceNow for incident tracking but has quietly rebuilt its approval workflows in spreadsheets and email has not adopted the platform either, no matter what the login dashboard says. Usage metrics count activity They do not tell you whether the platform has become how the organisation actually works, or whether it is being tolerated around the edges of how the organisation actually works ### Why Login Counts Survive as Long as They Do That's the trap. The metric looks healthiest at exactly the point where it is least informative. It's measuring novelty, not adoption. The real test comes twelve to eighteen months later, when the training has faded, the original project team has moved on, and users have had enough time to build workarounds if the platform didn't fit how they actually work ### What Adoption Actually Looks Like When It's Real Genuine adoption shows up as behaviour change, not activity volume. A few signals are far more reliable than login counts: Process compliance without enforcement. When users follow the intended workflow because it's genuinely the path of least resistance — not because a manager is checking — that's adoption. The moment enforcement stops and behaviour holds, you have real signal Employee experience score, tracked over time. This is one of the indicators Iconica tracks explicitly within Managed Indicators, and it exists because a platform can be technically "adopted" while making people's working lives measurably worse. If the experience score is flat or declining while login counts rise, that's not adoption — it's obligation ### The COO's Real Question Login counts cannot answer that question. Neither can license utilisation, ticket volume, or go-live sign-off. The only way to answer it is to define adoption as a business outcome before the platform is built, instrument the signals that actually predict it, and keep a continuous, accountable eye on those signals long after the project team has moved on to the next thing ### Top questions our clients ask Usage measures activity — logins, tickets raised, licenses consumed. Adoption measures whether the platform has genuinely changed how work gets done, without enforcement. A team can show high usage while quietly maintaining parallel spreadsheets or workarounds for the parts of their job the platform doesn't fit. Usage rises with novelty; adoption is only visible once the novelty has worn off and old habits either stayed gone or crept back. Reliable predictors include process compliance without enforcement, self-service deflection rates, employee experience scores tracked over time, the health of any platform champion network, and evidence of workaround systems being maintained alongside ServiceNow. These behavioural and outcome signals predict sustained adoption far more reliably than login counts or license utilisation, which mainly reflect short-term compliance.”

“# Why Are Your Enterprise Software Adoption Metrics Showing Green When Adoption Is Actually Failing? Your enterprise software adoption metrics are probably measuring the wrong things. Login rates, session duration, and active user counts tell you who showed up, not whether they are completing the workflows your software was built to support. The metric that actually reflects adoption health is workflow completion rate Most teams track enterprise software adoption metrics like daily active users and time in application, then wonder why ROI stays flat even when the numbers look good. The problem is not that these metrics are wrong; it is that they measure presence, not performance. If you want to know whether your software is actually working for your team, you need to look at what people are completing, not just how often they log in ## Key Takeaways Here is what this guide covers to help you build a more accurate picture of adoption health: - Why common adoption metrics like logins and session time do not reflect actual usage - What workflow completion rate is and why it matters more than active user counts - How to define, track, and act on completion data at the role level - A practical framework for building enterprise software adoption metrics that connect to business outcomes ## The Difference Between Product Usage and Adoption A user who opens a CRM every morning, navigates their pipeline and then switches to a spreadsheet to do the actual work is registered as an active user. Their session is counted. Their login is logged. The system reports high engagement. And your team continues to believe that adoption is on track because the difference between product usage and adoption is invisible in aggregate metrics The core issue is straightforward: usage measures how often someone accesses a system, while adoption measures whether they are completing the work the system was deployed to support. This gap often explains why many Feature Adoption in Enterprise Software Fall even when login rates appear healthy. These two things can look very different in practice, especially in the months after a major rollout when people are still finding their own workarounds ## Conclusion Enterprise software adoption is not about how many users log in-it is about how many successfully complete the workflows that drive business outcomes. While usage metrics such as active users and session duration provide visibility into activity, they do not reveal whether employees are achieving the goals the software was implemented to support ## Frequently Asked Questions (FAQs) ### 2. What is the difference between product usage and adoption? The difference between product usage and adoption is the difference between presence and outcome. Usage measures how often someone accesses a system. Adoption measures whether they are completing the workflows the system was built to enable. ### 3. Why does workflow completion rate matter more than active user counts? Active user counts measure access. Workflow completion rate measures whether that access is producing the intended outcome - task completion, process consistency, and reduced reliance on manual workarounds. A team can show strong login numbers and still have significant adoption failure underneath, because the two metrics are measuring fundamentally different things”

“Token consumption measures volume rather than value, raising the risk that companies optimize for AI usage instead of outcomes # AI Adoption Is Being Measured in Tokens, but the Metric Falls Short, Experts Say By PYMNTS | March 19, 2026 | Listen to Article Listen Pause AI tokens Highlights Companies are moving away from traditional per-user licensing toward token consumption as the primary metric for measuring AI adoption, workflow intensity, and enterprise spending Industry leaders, including Nvidia's Jensen Huang, suggest that employees may soon manage annual token budgets — a shift that treats AI compute power as a granular, real-time resource tied directly to worker behavior. While token tracking provides cost transparency, it risks becoming a misleading productivity proxy; high token usage often reflects inefficient prompting or “agentic” workflow leaks rather than high-quality business outcomes or ROI A growing number of companies are using a unit called the token to measure how much their employees and workflows use artificial intelligence, according to The Wall Street Journal. Companies that now regularly use AI are starting to track their workers’ use of tokens, AI’s unit of measurement ## Get the Full Story Every prompt a worker sends to an AI system and every response the system returns are measured and often billed in tokens. Every prompt and response consumes tokens and incurs charges. That direct relationship between usage and cost is what makes tokens attractive as a management tool. Unlike the seat-based pricing that defined earlier generations of enterprise software, token consumption is granular, real-time and tied directly to behavior ## Tokens Replace Seats as Adoption Proxy The shift from seat counts to token consumption mirrors how enterprise AI spending itself has changed. While the unit price of AI tokens is falling, overall enterprise spending on and scaling of AI systems is rising. The number of users, complexity of models, and intensity of workloads will likely drive greater token consumption and, consequently, higher costs OpenAI’s own data on its enterprise customer base illustrates how dramatically usage patterns have shifted. Average reasoning token consumption per organization has increased by approximately 320 times in the past 12 months, suggesting that more intelligent models are being systematically integrated into expanding products and services. That figure has become a headline metric in the company’s internal reporting on adoption progress ## Volume Without Value The appeal of token metrics runs into a fundamental problem: tokens measure volume, not outcome. Generating through packaged software abstracts tokens almost entirely, while consuming through APIs makes tokens explicit, but this can bring transparency and also volatility, as costs rise based on workload design, prompt length, and hidden choices of infrastructure providers The kind unit-economics mismatch is more visible as companies move from pilots to production deployments. As companies move from experimental chatbots to thousands of autonomous “agentic” workflows running around the clock, the sheer volume of tokens consumed has created a massive budgetary leak The dynamic draws comparisons to earlier enterprise metrics that proved easier to game than to interpret. Click-through rates once served as a proxy for advertising effectiveness; hours logged once functioned as a proxy for productivity. Both created incentives that diverged from the outcomes they were meant to track If token consumption becomes a performance indicator tied to employee evaluations, workers may optimize for AI interaction frequency rather than task quality. Knowing that “AI spend is up 40%” is not enough. Organizations need a single pane of glass that links every workload, tenant and token to their owners or business outcomes.”

“# 1. Why AI Agents Matter Now Not as a new interface. Not as a smarter chatbot. But as a different execution model for software. Agents matter now because large language models have crossed a practical threshold. They are not perfect, but they are reliable enough to reason across multiple steps, evaluate intermediate results, and decide what to do next. What they still lack is structure, control, and integration into real systems. Agents provide that missing layer Pic. Building an agent is a process of designing workflows and connecting pieces. Here an example of OpenAI (source) OpenAI frames this shift explicitly by defining agents as systems that “independently accomplish tasks on your behalf,” rather than systems that merely generate responses.2 Microsoft’s enterprise documentation makes a similar point, describing agents as long-running, goal-driven processes that operate across tools and workflows instead of inside a single UI.3 # 2. What Do We Mean by “AI Agent” #### A working definition (used throughout this series) For this series, an “AI agent” is a software system that: - pursues an explicit **goal**, not just a prompt - plans across **multiple steps**, not a single response - maintains **state** between steps - acts through **tools** (APIs, code, systems) - evaluates progress and decides whether to continue, adapt, stop, or escalate ``` Goal → Plan → Act (Tools) → Observe → Update State/Memory → (repeat until done) ``` #### Why the term “agent” became confusing The confusion is that the market uses “agent” for three different things: 1. **Assistant** (helpful interface) 2. **Workflow** (predefined steps, deterministic) 3. **Agent** (goal-driven, stateful, tool-using loop) If you don’t separate these, you can’t reason about reliability, cost, governance, or failure modes. Everything sounds “agentic” until it breaks Pic. *Growth of LLM-based autonomous agent research and agent categories (2021–2028). (Source)* **Diagram (fast comparison):** ``` Assistant: user ↔ chat (help) Workflow: input → steps → output (deterministic) Agent: goal → loop → tools → state → completion/escalation (adaptive) ``` # 3. From Apps to Agents: The Real Shift ### What agents change AI agents invert this relationship. Instead of waiting for instructions, an agent receives a **task**. It determines which steps are required, executes actions across tools and systems, evaluates intermediate results, and continues until the task is completed or explicitly escalated. The unit of interaction shifts: - from screens and features - to tasks and outcomes This is the core structural change. # 5. Structure ≠ Decision Capability At this point, the **structure** of AI agents is clear. They plan. They maintain state. They call tools. They execute workflows. That already marks a real architectural shift. What it does **not** imply is decision maturity”

“What is an AI agent? It's software that perceives data, reasons about goals, and acts autonomously across workflows — without waiting for step-by-step instructions. # What is an AI agent? A practical guide for business leaders If you are asking, “**What is an AI agent?”**, the simplest answer is this: an AI agent is software that can understand a goal, gather the context it needs, decide what to do next, and take action across one or more steps with limited human input. OpenAI describes agents as applications that can plan, use tools, collaborate with specialists, and maintain sufficient state to complete multi-step work. ## Key takeaways - **What is an AI agent**? Think beyond chat: an agent does not just respond; it can plan and execute work toward a goal. - The best early use cases are repetitive, high-volume workflows with clear success criteria, such as ticket triage, lead routing, status reporting, and meeting follow-up. - Data quality matters as much as model quality. Agents are only as useful as the context they can access ## What is an AI agent? An AI agent is a software system that can pursue a goal with some degree of autonomy. It takes in information from its environment, reasons about the task, chooses actions, uses tools or connected systems, and adjusts based on results. NIST’s new AI Agent Standards Initiative explicitly focuses on agents as systems capable of autonomous action on behalf of users, which is a good sign that the industry is moving toward a more formal definition ## What makes an AI agent different from traditional automation? An AI agent is different because it can deal with changing contexts. It can evaluate multiple inputs, interpret what matters, and choose between possible next steps. monday.com’s support materials make this distinction clearly: automations follow predefined if-this-then-that logic, while agents can evaluate context and choose actions dynamically within guardrails ## AI agents vs. chatbots, copilots, and assistants ### AI agents An AI agent is goal-directed. It can carry context across steps, use tools, and act with limited supervision. That means the difference is not just technical. It is operational. A chatbot answers. A copilot assists. This distinction also shows up in official platform guidance. OpenAI’s agent documentation centers on planning, tools, state, and multi-step task completion, while monday.com’s agent builder docs emphasize that agents can analyze information and choose next actions dynamically instead of following fixed scripts. ## Common types of AI agents ### Goal-based agents These are given an outcome and work through several steps to reach it. Unlike simple task agents, goal-based agents can navigate multi-step processes, make intermediate decisions, and adapt their approach based on what they encounter along the way. They maintain context across actions and understand when a step is complete versus when additional work is needed. ## Bottom line: what is an AI agent, really? An AI agent is software that can understand a goal, gather context, decide what to do next, and take action across multiple steps with limited human input It represents a fundamental shift in how work gets done. Unlike chatbots that respond to prompts or automation that follows fixed rules, agents can navigate complexity, adapt to changing conditions, and execute coordinated work across systems without constant human intervention ## Frequently asked questions ### Is ChatGPT an AI agent? Not in its standard form. ChatGPT is designed as a conversational assistant that responds to individual prompts. It becomes part of an agentic system only when it's connected to external tools, maintains state across interactions, integrates with business workflows, and can execute multi-step tasks with minimal human intervention.”

“# Companies Begin Deploying Internal AI Assistants March 20, 2026 Over the past two years, most organizations have experimented with AI tools like ChatGPT. Employees use them to draft emails, summarize documents, or brainstorm ideas. Now a new phase is beginning. Instead of using AI as a standalone tool, many companies are starting to deploy internal AI assistants that connect directly to company systems and data. ## Citigroup Testing AI Agents for Employee Workflows Citigroup is piloting AI agents that help employees complete research and operational tasks. Instead of manually searching through internal systems, employees can ask the system to gather information about clients, summarize research, or compile insights from multiple internal sources. The AI then retrieves relevant data and prepares a draft response or report, significantly reducing the time needed to complete the task.”

“# AI for work: 15 ways teams use AI to get more done ## How monday.com helps teams put AI to work ### A personal AI assistant built into your workflow monday sidekick is a context-aware AI assistant built directly into the workspace. It connects to your work data and integrated communication channels to help you think, create, and take action through natural conversation.”

“# What is an AI Assistant? Types & How They Work ## AI Assistant FAQs ### Is ChatGPT an AI assistant? ChatGPT is an AI assistant, but it’s software-based and prompt-driven. It works well for drafting, answering questions, and idea generation, but it doesn’t automatically capture meetings or shared context without manual input”

“# CRM AI Agents: A Complete Guide ## Traditional CRMs vs CRMs with AI Agents Instead of navigating complex menus or manually updating records, a salesperson might simply type, “Recommend the best cross-sell and up-sell opportunities,” or a service agent could ask, “Summarize the customer’s last three interactions.” **The AI agent interprets the intent and executes the task, often suggesting next steps** ## Why Should Businesses Consider Using CRM AI Agents ### 1. Efficiency and time savings AI agents handle **repetitive and administrative tasks**, such as updating records, scheduling follow-ups, or summarizing interactions, so employees can spend more time on high-value activities like building customer relationships or closing deals. ### 3. Better data accuracy CRM AI agents can **automatically log information and update records without manual input**. After a sales call, customer meeting, or support chat, the agent can capture key details, such as discussion points, decisions made, and next steps, and instantly update the CRM”

“# What is a CRM agent? The complete guide to AI-powered sales and service ## Key takeaways - CRM agents work 24/7 to qualify leads, answer questions, and manage follow-ups automatically, freeing your team to focus on closing deals and building relationships. - These AI-powered digital workers learn from every customer interaction and get smarter over time, unlike basic chatbots that only follow scripts. ## What are CRM agents? CRM agents are AI-powered digital workers that handle customer relationship tasks without human help. Think of them as an AI sales agent that can qualify leads, answer customer questions, and manage follow-ups automatically. Unlike regular CRM software where you manually enter data and track interactions, a CRM with AI can make decisions on its own. They learn from your customer data and get smarter over time ## 7 key benefits of CRM agent technology ### 5. Reduced manual data entry Nobody became a salesperson to type notes all day, yet according to HubSpot, sellers in the US and Canada spend over one-third of their time on administrative duties and updating their CRM. CRM agents capture information from emails, calls, and meetings automatically, aligning perfectly with CRM automation best practices.”

“# From Call to CRM Update Automatically Automated call-to-CRM doesn't eliminate that work. It compresses it. The AI does the extraction, drafts the update, and presents it to the rep for a 3-to-5-minute review-and-confirm rather than a 25-minute reconstruction. What used to be recollection becomes review. This is the Meeting Intelligence Pattern completing its full cycle: from audio capture through to a CRM record that actually reflects what happened ## What "auto-update from call" actually means - MEDDIC fields are chronically underfilled in manually-maintained CRM records; call-to-CRM automation closes that gap by extracting structured deal data from transcripts after every call, regardless of whether the rep manually updates ## The confidence threshold model **High confidence (auto-commit after configurable delay):** Statements that are explicit and unambiguous. "I'll send you the security review documentation by Thursday" is an explicit next step with a named owner, action, and date. The AI extracts it, maps it to a CRM task field, and auto-commits after a delay window (usually 30 minutes to 4 hours) to allow rep corrections ## Frequently Asked Questions Automated call-to-CRM update uses AI to extract structured data from call recordings and transcripts, then push that data directly into CRM fields rather than requiring the rep to reconstruct and type it manually after the call.”

“# Invoice Workflow Process: Steps, Roles, and Where It Breaks ## How AI and agentic automation change the invoice workflow Rules-based automation fixed the mechanical parts of this process years ago. It could route by threshold, match against a purchase order, and post to the ERP. What it could never do was handle judgment: the non-PO invoice with no code to copy, the layout it had never seen, the exception that needed a decision rather than a rule. - **Routing and approval.** Instead of a fixed rules table, AI-assisted routing learns who actually approves which invoices and sends them there, then reminds and escalates when they sit unactioned. The queue stops depending on one person's memory. - **Exception handling.** AI flags the unusual amount, the missing field or the mismatch early, before posting, so the exception is caught where it is cheapest to fix rather than after the payment has left - **The agentic leap.** The newest shift is from AI that suggests to agents that act. An agentic layer does not just recommend a code or flag an exception; it works the state machine, chases the missing PO, re-routes the stalled approval, clears the routine exception, and hands a person only what genuinely needs a human decision. This is the model Mindsprint SprintAP runs on, delivered as a managed service for complex, multi-entity operations ## How to automate your invoice workflow process If your team is working through what this looks like in your ERP environment, Mindsprint SprintAP is an agentic AP automation platform built for exactly this kind of phased deployment.”

“# Invoice Approval Workflow: Best Practices & Solution Guide 2026 ## TL;DR - An invoice approval workflow validates, routes, and approves supplier invoices before payment while improving financial control, compliance, and ERP accuracy. - Approval bottlenecks persist after AP automation due to limited visibility, fragmented approvals, exception handling, and complex enterprise workflows. ## What is an Invoice Approval Workflow? ### Invoice approval workflow vs. invoice processing vs. payment processing Process · Purpose · Outcome Invoice Processing · Captures, extracts, validates, and classifies invoice data · Invoice is ready for review Invoice Approval Workflow · Routes invoices for review, matching, approvals, and exception handling · Invoice is approved for payment ## Invoice Approval Workflow (Step-by-Step) ### Step 3: Three-way matching and invoice validation For example, **Serina** automatically validates invoices against purchase orders and goods receipts, identifies discrepancies in real time, and ensures only compliant invoices continue through the approval workflow while exceptions are isolated for further review. ### Step 4: Quick exception handling and approval routing For example, **Serina** uses configurable approval workflows to automatically route compliant invoices to the right approvers while directing exception invoices to finance, procurement, or business teams for resolution. This exception-based approach helps accelerate approvals without compromising governance. ## Best Practices for an Efficient Invoice Approval Workflow ### 7. Increase touchless invoice processing Focus on increasing the percentage of invoices that move through validation, matching, approval, and ERP posting without requiring additional intervention. Combining AI-powered invoice capture, three-way matching, approval routing, and automated validations allows finance teams to focus primarily on exception invoices ## Conclusion By combining AI-powered automation with standardized approval policies, approval matrices, intelligent routing rules, and continuous performance monitoring, you can reduce approval delays, improve compliance, strengthen financial governance, and increase touchless invoice processing without replacing your existing ERP ## FAQs **3. What is touchless invoice processing?** Touchless invoice processing refers to invoices moving through capture, data extraction, validation, approval, ERP posting, and payment preparation without requiring human intervention. AI-powered invoice automation makes this possible by automatically extracting invoice data, performing validations, routing approvals, and managing compliant invoices while directing only exceptions for review”

“# The Complete Guide to Invoice Processing and Accounts Payable Automation ## Why is intelligent automation the next step for accounts payable? ### How does efficient invoice processing improve cash flow? #### Faster approvals prevent late payments Every day an invoice sits waiting for approval reduces the time available to schedule payment. Slow, manual workflows often leave finance teams choosing between two bad options: pay late and face the penalty, or rush the payment to avoid one. ## Which technologies power invoice automation in 2026? ### How does optical character recognition improve invoice processing? While OCR is highly accurate, it can fail to recognize documents with unique layouts. That’s why an agentic procurement and spend centralization platform like Precoro, after seeing a high 96% accuracy rate with OCR, decided to take it further with Intelligent AP Automation, which handles every manual step between the invoice submission and final approval with AI ### How do RPA, workflow engines, and integration platforms work together? A typical automated workflow starts with intelligent document processing (IDP), which extracts and validates invoice data. RPA then uses that data to perform repetitive tasks, such as entering invoice information into the ERP or updating records. The workflow engine, a core part of BPM, routes invoices to the correct approvers, applies company policies, and manages exceptions ## What does an ideal end-to-end automated invoice workflow look like? ### How do modern invoice processing systems capture, validate, approve, and archive invoices? #### How automation solutions approve invoices Clean invoices are automatically routed to the appropriate approver based on business rules, such as invoice amount, department, cost center, or vendor. ## Quick recap of invoice processing - Manual invoice processing is slow, error-prone, and gets more expensive as volume grows. - Intelligent automation combines OCR, IDP, AI, and RPA to capture, match, and route invoices with minimal human touch. - OCR converts a scanned invoice into machine-readable text; IDP extracts and structures the data. - AI classifies documents and decides when a human needs to review an exception.”