How helpdesk scoreboards stay green while silent churn grows

Help desk metrics, with formulas

Help desk metrics are the counters a support team uses to measure how fast and how well it handles incoming tickets: response and resolution speed, resolution quality, backlog, customer satisfaction and cost per ticket. Almost all of them start counting when the shopper contacts you.

These helpdesk metrics are worth running well, and I would never tell a support lead to stop. Here is the standard set with a formula for each. The response time and resolution time figures are Gorgias Ecom Lab medians for brands at $10M GMV in March 2026, and the CSAT range comes from the same Gorgias research; that is one vendor's platform data, not an industry average. The first contact resolution figures are SQM Group's 2024 call center benchmark, and the SLA and NPS lines are Gorgias rules of thumb.

Three habits keep them honest. Report response time in bands (the share answered inside 1, 6 and 24 hours, the bands Zapier tracks), because one slow weekend skews a mean. And read backlog as a trend, as Gorgias advises: one that clears overnight is a volume spike, one that grows week over week is structural. Last, split every row by contact reason before you act on it. Gorgias makes the point itself: "a low FCR on fulfillment-related tickets may simply reflect your operational reality rather than a support problem," and a consistently high average resolution time on one ticket type "usually points to a missing self-service resource, a broken integration, or a workflow that needs automation." When the slow, multi-touch rows are order problems, the fix belongs in operations before it belongs in the queue. For cost per ticket, an illustrative team of three at $45,000 loaded cost each, plus $12,000 of platform and $18,000 of overhead, costs $165,000 a year; across 22,000 resolved tickets that is $7.50 a ticket.

Where the clock starts

Nine of the ten count only contacts that reached you, and the tenth, NPS, is a relationship survey that cannot tell you which order broke. So a team can improve every row while the operation behaves exactly as it did last quarter. Customer service measures how well problems get fixed, not how well they are prevented, and no shopper ever wanted to ask where their order is. That is why I read CSAT as a recovery score and nothing more. The survey only reaches the shopper who had a problem, contacted you, got it fixed and rated the interaction. It never reaches the one who had the same problem, decided the hassle was not worth it, left for a competitor and never came back. What CSAT actually samples matters more than its score.

Three upstream counters for ecommerce teams

• Contact rate: tickets in the period / orders in the same period x 100, so 10,000 tickets on 100,000 orders is 10 per 100. Unlike the ten above, it moves when the operation improves, not only when the team works faster. Read it by reason: the order-linked rows are the ones operations can remove. Vertical benchmarks are on tickets per 100 orders.
• Orders in breach: the count that question two of the audit below produces, taken now rather than when someone complains.
• Share caught before contact: orders in breach that your operation flagged before the shopper wrote in / all orders in breach x 100. Each one caught and fixed is a ticket prevented, the north star in the last section.

A support ticket is evidence that an operations promise already broke, so the fix for most order contacts sits upstream of the help desk. Our about page puts it in six words: "Helpdesk manages complaints. Keeyu prevents them." The ten metrics stay on the board for the questions only a person can answer, in whichever helpdesk you run. Keeyu integrates with Gorgias and Zendesk and works upstream of them: it detects the break in payment, fulfillment, shipping, delivery or returns, decides what should happen, and acts before the shopper writes in.

Every order is a promise. Keeyu keeps it. So the ecommerce scoreboard needs counters that start before the ticket.

What green actually certifies

First response time is green. Handle time is respectable. CSAT on closed tickets looks polite. Queue depth is inside target. Then a founder asks why repurchase feels softer than the slide claims. The scoreboard is not lying about what it measures. Most CX and ops teams I see still measure success by speed. For two decades helpdesk software has focused on faster response times, and reply time and tickets closed per hour are easy to track and easy to show improving. That gives everyone an illusion of progress. A fast reply does not fix the problem underneath. It only masks it.

Every tile on that dashboard starts counting at the ticket, which means every tile is silent on everything that happened before it, and on everyone who never filed. First response time proves how fast you acknowledged a complaint. It says nothing about the hours or days of silent break before it, the same gap first response time versus first detection time measures directly. Handle or resolution time proves how efficiently you closed a conversation. It says nothing about whether the root cause still exists. CSAT on closed tickets proves the mood of people who wrote in and stayed engaged. Queue depth proves you staffed to visible demand. Ticket averages falling after triage AI prove the inbound ticket got triaged faster. What I see is a CX team left to triage it, because they have no visibility of what is happening upstream and find out after the fact. Not one of those proves the promise was kept.

The cleanest example I know

On a discovery call my team ran with an omnichannel footwear retailer, helpdesk AI had pulled average tickets from roughly 600 toward 400. The triage scoreboard greened. Their biggest reach-out stayed WISMO, specifically split delivery and tracking. Underneath it, their warehouse system was still not passing tracking numbers back to the storefront, with about 40% of orders shipping from store and 60% from the warehouse.

A greener average grades how efficiently you take the test results. The complete-order promise was still broken while Monday's slide looked better. A board celebrating ticket decline without measuring silent breaches is celebrating a thinner sample of the same pain.

What sits outside the frame

Three scenes show what the tiles structurally cannot see. On a scoping call with a fashion apparel brand, I pulled up an existing customer's data set to show how the dashboard reads. The broken-service-level view showed 175 potential complaints with no proactive measures taken, and the issue rate was running at 43% of all orders in that geography. Not carelessness. Those breaches simply were not a first-class object until someone complained. That is the silent cohort on glass, before it becomes either tickets or lost customers.

On a scoping call with a consumer brand, my contact told me they count it a victory when more than 98% of orders ship on time. The blind spot, in their words, is whether the customer is actually getting it. They review the negative feedback every week, and by their estimate roughly 20% of it is the carrier not delivering when it was supposed to. Pick SLA is necessary and it is not the customer promise. The promise spans warehouse, carrier, collection and delivery scan.

And on a call with a supermarket and marketplace-style retail group, I told them what we usually find when a brand comes onto our platform: the issues run at about twice the tickets, and that gap is churn. Their answer was that NPS is the golden metric there, everyone's bonus is tied to it, and the score they cannot explain comes from the shopper who never raised a ticket. Treat the ratio as framing rather than a constant, then measure your own. The question I put to every board is the same: have you measured your silent cohort, or are you surveying smoke while the plant runs?

Why the scoreboard resists change

A beauty retailer I met on a discovery call had no automated comms about carrier delays, and their early warning after Boxing Day (December 26, the biggest single sale day of the Australian retail year) was still somebody scanning a report by hand. Courier late, missed, damaged and return to sender were all discovered reactively. Roughly fifty onshore agents sat across five systems, measured on first response time and average handle time, with CSAT as the north star. Fulfillment rate, partial fill, SLA violations and value at risk never made the board.

That is not a failure of effort. The way I put it, behind every late order is one of two customers, and both are dangerous. The one who asks where it is becomes a helpdesk ticket and lands on a tile. The one who got the order later than promised and is upset, but never writes in, lands on none, and then they are gone. That is the silent problem. A multi-brand footwear group I demoed to had never had platform reporting on fulfillment rate or SLAs at all, and told me a General Manager would care about that gap. You will not fund what the scoreboard does not name. Every order is a promise. Keeyu keeps it. The board has to name the orders where that promise is at risk.

Audit your own scoreboard this week

You do not need a new platform to find out whether your tiles are hiding something. Four questions. For each tile, when does the clock start? Write it down. If every answer is "when the ticket is created," you have confirmed the scoreboard cannot see prevention. How many orders are in breach right now, not tickets, but orders past their promised dispatch, parcels with no scan past your threshold, returns aged past your credit clock? If you cannot produce this number, that is the finding. How many of those breached customers have contacted you? The gap between question two and question three is your silent cohort, measured rather than assumed. What is repurchase rate for that cohort against everyone else? You already hold both datasets, and joining them takes an afternoon and changes the meeting permanently.

Most teams discover the answer to question two is a number nobody has ever produced. That absence is the whole argument.

What replaces it

Two north stars, and neither starts at the ticket: tickets prevented, contacts that never formed because a breach was caught and fixed or honestly communicated first, and lifetime value protected, repurchase preserved among customers whose promises were kept. Both sit inside the full contact-center to commerce-operations KPI migration. Keep first response time, handle time and CSAT for the residual contacts that genuinely need a person. Just stop letting a clock that starts at the complaint decide what gets funded.

When a brand moves to proactive e-commerce operations, its scoreboard does not go greener. It starts measuring something different: orders in breach, caught and fixed before the shopper writes in. Our customer stories show what that looked like for EHP Labs, Budgy Smuggler and Clutch. Points and VIP tiers cannot out-spend a late parcel. For me, operations are increasingly the customer experience: an order that shows up on the day you promised is the loyalty program, and no tile on your current dashboard is measuring it. Every order is a promise. Keeyu keeps it. Keeyu gets shoppers what they want, on time, as promised.

• 1. Keeyu discovery call run by the Keeyu team (prospect, not a customer), Australian omnichannel womens footwear and accessories retailer, internal call transcript, 2026-04-23.
• 2. Keeyu product demo (now a customer), fashion apparel DTC account, internal (date not recorded).
• 3. Keeyu discovery call (prospect, not a customer), outdoor and adventure gear brand, internal call transcript (date not recorded).
• 4. Keeyu call notes, undated, supermarket / marketplace-style retail group, internal call transcript (date not recorded).
• 5. Keeyu discovery call (prospect, not a customer), Australian beauty retail / omnichannel business, internal call transcript, 2026-02-11.
• 6. Keeyu demo call (prospect, not a customer), Australian multi-brand footwear and fashion group, internal call transcript, 2026-03-12.
• 7. Gorgias, "14 Customer Support Metrics Every Ecommerce Team Should Track", https://www.gorgias.com/blog/customer-support-metrics. Formulas for FRT, resolution time, FCR, backlog, CSAT, CES, NPS, SLA compliance and cost per ticket; Gorgias Ecom Lab medians at $10M GMV, March 2026, for FRT and resolution time; backlog trend advice; the fulfillment-related FCR and missing-self-service-resource caveats; CSAT range 4.45 to 4.66 out of 5; CES wording; SLA 90% and NPS 50 as Gorgias rules of thumb.
• 8. Zapier, "9 help desk metrics to guide your customer support", https://zapier.com/blog/customer-support-data/. Reporting response time in time bands rather than a single average.
• 9. Qualtrics, "What is CSAT", https://www.qualtrics.com/articles/customer-experience/what-is-csat/. CSAT formula (4 and 5 responses / total responses x 100).
• 10. SQM Group, "Call Center FCR Benchmark 2024 Results by Industry", https://www.sqmgroup.com/resources/library/blog/call-center-fcr-benchmark-2024-results-by-industry. 69% all-industry FCR average, 73 to 75% retail, 80% world class; North American call center data.
• No dollar lifetime-value or NPS-point uplift is claimed on this page. Published customer outcomes are on keeyu.com/customers.