qDesk

Halving first-response time without adding headcount.

Helix consolidated six support channels into one queue and let the help centre absorb the repeat questions.

Outcome

45-minute first response

Customer
Helix Instruments
Sector
Laboratory instruments and consumables
CEDX Desk, as the team works in it.
Rollout

4 phases, in the order they happened.

  1. Weeks 1–2

    Contact study. All six routes instrumented and a month of traffic classified by question rather than by channel.

  2. Weeks 3–5

    qDesk live with a single queue and one SLA. Ring replaces the two regional numbers.

  3. Weeks 6–7

    Help centre rebuilt around the eleven repeat questions. Beacon deployed on the documentation site.

  4. Weeks 8–9

    Named-engineer lines routed through Ring; Poll added for post-resolution satisfaction.

What changed, measured.

Every figure came from Helix Instruments. Where one is a median or a sample it says so.

45 minutes

Median first response

Across every channel, from nine hours twelve minutes on the one route that was measured.

2,230

Repeat contacts no longer raised

A month, against the 5,880 identified as one of eleven questions.

211

Help centre articles

Up from 34, written by support engineers in scheduled time.

0

Additional support headcount

Contact volume rose 9% over the same period.

The programme

Six queues, six definitions of answered.

Support reached Helix by six routes: a shared mailbox, a web form, two regional telephone numbers, a chat widget, and (for the forty largest accounts) the mobile numbers of three named engineers.

Only two of the six were measured. First response on the mailbox was reported at four hours. Once we counted messages that arrived over a weekend and the ones a colleague answered without recording anything, the real figure was nine hours and twelve minutes.

The named-engineer route was the worst of it. Fast for the customer, invisible to everyone else, and a single point of failure that took three weeks of leave every August.

The queue was not the problem. The repeats were.

Of 14,300 contacts in a representative month, 5,880 were one of eleven questions. Calibration procedure after a firmware update accounted for 1,340 on its own.

Answering the same question 1,340 times is not a staffing problem, it is a publishing problem. The help centre had 34 articles, most written at launch and none updated after the firmware release that was generating the calls.

Hiring would have worked. It would also have cost four engineers a year and made the repeats permanent, because a question that is cheap enough to answer never gets written down.

One queue, and something to point at.

All six routes now land in one qDesk queue against one SLA clock. Ring carries the telephone lines, so a call becomes a ticket with a recording attached rather than a note somebody may write up later. Beacon sits on the documentation site and hands over to a person the moment it cannot answer.

The named-engineer route was not switched off, it was absorbed. Those three engineers kept their numbers; the calls now open a ticket automatically, which is what made the change acceptable to the accounts that depended on them.

The help centre went from 34 articles to 211 in eleven weeks, written by the support engineers themselves in two scheduled hours a week rather than by a technical writer afterwards. The eleven repeat questions were the first eleven articles.

We were about to hire four engineers to answer the same eleven questions faster. The uncomfortable finding was that the questions were our fault, not the customers'.

Tomás FerreiraDirector of Global Support, HelixRunning CEDX Desk
Next

See what Desk does before you read another story.

Tickets with full customer context. There is a full trial on desk.cedxsystems.com, and an engineer who has run a migration like this one will walk it with you.