"Let me check on that." "Give me a moment to look this up." "I'll need to check with our team on this." "Hang on, let me pull up your account."

Your agents say some version of this 20 to 40 times per day. Each time, the customer waits 1 to 5 minutes while the agent searches for information across tabs, systems, and Slack channels.

Let's do the math on what those minutes cost.

The Direct Cost

A support agent earns $20/hour (loaded cost). They type "let me check on that" 30 times per day. Each check takes an average of 3 minutes (some are quick system lookups, some involve asking a coworker or checking documentation).

30 checks × 3 minutes = 90 minutes per day per agent spent looking things up.

90 minutes at $20/hour = $30/day per agent.

For a 5-agent team: $150/day, $750/week, $39,000/year.

$39,000 per year in agent time spent searching for information they should have at their fingertips.

The Customer Cost

Each "let me check on that" adds 3 minutes to the customer's experience. Over 30 interactions per day per agent, that's 1,500 minutes of customer waiting per day (across the team). 25 hours of cumulative customer wait time. Every day.

Customer time has value too. A business customer earning $50/hour who waits 3 minutes has lost $2.50 in productive time. Multiply across all your interactions, and your "let me check" habit costs your customers tens of thousands of dollars per year in aggregate.

More importantly, every wait introduces friction. The customer's satisfaction drops a little each time they hear "hang on." Four waits in one conversation drops satisfaction significantly more than one wait, even if the total wait time is the same.

Why It Happens

The information agents need is scattered across multiple systems:

The help desk has the ticket history. The CRM has the customer's account details. The billing system has their payment info. The product's admin panel has their feature access. Slack has the answer from the engineer who fixed a similar issue last week. The knowledge base has the documentation (maybe).

Switching between 4 to 6 systems to gather information for a single ticket takes time. Every system has a different login, a different search interface, and a different data model. The agent is doing manual data aggregation across fragmented tools.

The Fix: Information at the Point of Need

The goal: when an agent opens a ticket, everything they need to resolve it should already be visible. No switching tabs. No "let me check."

Customer context. Account type, plan, billing status, recent orders, previous tickets. All visible in the same view as the current conversation.

Classification context. What does the customer want? AI classification determines this in milliseconds. The agent sees: "Intent: billing dispute. Priority: high." They don't need to read the message twice to figure out the category.

Suggested response. Based on the classification, show the agent a draft response or relevant knowledge base article. They customize and send instead of composing from scratch.

Internal notes from previous tickets. If a coworker handled a similar issue last week, the resolution should be findable without Slack-searching.

Supp provides the classification layer: 315 intents classified in under 200 milliseconds. The agent sees the intent, the priority score, and the full conversation history before they type a single word. For automated intents, the response goes out without an agent at all, eliminating both the "let me check" and the wait.

The Realistic Savings

If you reduce "let me check" time from 90 minutes per agent per day to 30 minutes (by putting information at the point of need):

Time saved: 60 minutes per agent per day. 5 agents: 300 minutes saved per day, 25 hours per week. Annual savings: $26,000 in agent time.

Those 25 recovered hours per week can handle an additional 75 to 100 tickets. That's either a capacity increase (you handle more with the same team) or a cost decrease (you handle the same with fewer people).

The investment: better tooling, AI classification, and unified customer context. For most teams, this costs $100 to $500/month. The payback period is measured in days.

The Phrase That Should Disappear

In a well-instrumented support system, "let me check on that" should be rare. Not because agents are rushing, but because they don't need to search. The information is already there.

When an agent does need to check (rare situations requiring cross-team consultation), they should say so specifically: "I need to check with our engineering team about this specific behavior. I'll follow up within 2 hours." That's honest and specific. "Let me check on that" is vague and anxiety-inducing.

The difference between the two responses is the difference between a team that has its information together and one that doesn't.