The Limits of First Contact Resolution as a Measure

Wednesday, 16:50, at Bluewater Bank. Ade has nine tickets left and forty minutes. The desk's first contact resolution figure has been on the team's monthly page for three months, rising each time. It has a note from the branch operations manager saying "great work". Ticket 3317 is from Tom in mortgages: his loan calculator spreadsheet has started freezing for a minute at a time. Ade has seen this before. The honest fix is to watch the machine for a day, because it is one of three things. The quick fix is to clear the application's cache and tell Tom to restart, which makes it go away for a while. He clears the cache, writes "this should resolve it", and marks the ticket resolved. Tom replies "seems fine, thanks". Six working days later it freezes again, outside the reopen window. Tom opens a new ticket, which Ade's colleague fixes properly after a day of watching. The first ticket counted as resolved on first contact. The satisfaction survey on the second one came back with the words "second time of asking".

Nothing on the monthly page was wrong, and the page was lying anyway. The limits of first contact resolution are the things the number cannot see. They include whether the fix was the right one, and whether the problem needed two contacts and was forced into one. They include how much of the requester's time was spent, and whether the person on the other end would describe the experience as help. Every measure has a blind side. This one's blind side is quality, which is exactly where a desk under pressure to raise the figure will go to find the points. This article is part of our First Contact Resolution series. The definition is in What First Contact Resolution Means. This article is about reading the number with the two or three others that cover its blind side. That way, a rise means what it appears to mean.

It covers what the number sees and what it does not, and the pressure to close fast and the behaviours it produces. Then come the ticket that should take two contacts, and the quality the number does not see. Next is pairing it with reopens, with satisfaction, and with time and escalations. It ends with when not to read it at all, a short war story, and the paired-metric check to copy, in that order.

What the Number Sees and What It Does Not

First contact resolution sees four things about a ticket: that it existed, who owned it, how many exchanges it took, and whether it was reopened inside the window. Those are all facts about the shape of the ticket. They are good facts, because the shape is what the requester experiences as "how many times did I have to ask". The number is honest about shape. It has no opinion about content.

What it does not see is everything inside the exchange. It cannot tell a complete fix from a workaround that bought a week. It cannot tell a reply that taught the requester something from one that left them unable to recognise the problem next time. It cannot see that the requester spent forty minutes on the phone being walked through something a technician could have done remotely in four. It cannot see that the "resolved" ticket produced a different ticket three days later about a side effect. And it cannot see the reopen that arrives on day eleven of a ten-day window, which is how Tom's ticket got away.

None of this is a flaw in the number; it is the number's job to measure shape, and it does. The flaw is in reading it alone, as though shape were quality. A tape measure is not wrong because it cannot weigh anything. It is wrong only in the hands of someone who reports the weight.

The Pressure to Close Fast

The oldest rule about measurement is that a figure people are judged on stops measuring what it was meant to. It starts measuring their ability to move it. First contact resolution is unusually easy to move the wrong way. The moment it appears on a page with "great work" beside it, every technician on the desk feels a small pressure. It comes at the point where Ade felt it: the ticket that could be closed now, or fixed properly later. The pressure is not dishonesty. It is a tired person at ten to five being told, by a number, which of two reasonable choices the organisation prefers. The cures for a gamed metric in general belong to Metric Gaming and Its Cures. This section is about what this one looks like when it bends.

The behaviours arrive in a known order. First the workaround presented as a fix: clear the cache, restart, "this should resolve it". Then the first-reply close, where a reply that asks the requester to try something is marked resolved before they have tried it. Then comes the discouraged reopen: the resolving message that says "please open a new ticket if the problem recurs". That moves the recurrence out of the window and into a fresh ticket that counts as a new problem. Then comes the technician who stops escalating tickets that should be escalated, because a handoff breaks their chain. They spend two hours on something the second line would have fixed in ten minutes. Each of these raises the figure. Each of them is worse for the requester than the behaviour it replaced.

The last one deserves its own warning, because it looks like diligence. A first-line technician who will not hand off is a technician whose first contact resolution is high and whose resolution time is terrible. They are quietly teaching the requester that the desk is slow. Escalation is a good act when the ticket needs it, and a desk whose number punishes it will get less of it. Say so to the team, out loud, before the number goes on the page. A handoff that the ticket needed is not a failure. The number will count it as one. The number is not in charge.

The Ticket That Should Take Two Contacts

Some problems need two contacts, and a desk that forces them into one does worse work. Tom's spreadsheet was one: an intermittent fault that needs a day of watching before anyone can say which of three causes it is. A device that has to run overnight before a change can be confirmed is another. So is the request where the fix depends on a decision the requester's manager has to make. Another is the ticket from a requester who is upset and needs the first contact to be a conversation and the second to be the fix. The right number of contacts is the number the problem needs, and not one more. Sometimes that number is two.

The honest way to handle the two-contact ticket is to make the second contact the desk's. Say so in the first: "I have cleared the cache, which will help for now. But I would like to watch the machine tomorrow to find out which of three things is causing it. I will call you at ten." The ticket stays open. It will not count as first contact resolution. It will count as a fixed problem, a requester who was told the truth, and a technician who did the job. The figure takes the hit and the desk takes the credit, which is the right way round.

A desk that understands this writes it into the counting rules sheet from the previous articles as a plain statement. Tickets that the technician deliberately keeps open for a second contact are not first contact resolutions and are not failures either. Some desks give them a note or a tag so that the monthly review can count how many there were. That count is useful in its own right. A month with no deliberate second contacts is a month in which nobody watched anything. A month with sixty is a month in which "I'll call you tomorrow" has become the new "try restarting". (It will try. Habits always do.)

The Quality It Does Not See

Beyond the two-contact case, there is a kind of quality the number is simply blind to. It is the quality of the fix itself. A ticket fixed at the symptom counts the same as one fixed at the cause. Clearing the cache and replacing the faulty add-in that corrupts it are both one contact, both resolved. Both are still fixed when a ten-day window closes. One of them will be back in a month and one of them will not. The difference shows up, eventually, as a repeat ticket, which is the territory of problem management. But it will show up there under a different ticket number, from the same requester or a different one. The first contact resolution count will never connect them.

The same blindness covers the resolution note, the requester's time and what the requester learned. A ticket resolved in one contact with the note "fixed" is one contact. So is the same ticket with a note that says what was wrong, what was done and how to recognise it next time. Only one of them helps the technician who gets the next one. A ticket resolved in one forty-minute call is one contact; so is one resolved in four minutes by remote control. The number scores them alike. The requester does not, and neither should the desk lead, which is why the number's partners matter more for this measure than for most.

Pairing With Reopens

The reopen rate is the share of resolved tickets the requester came back on to say the problem was not fixed. It is the number that catches the hurried fix, because a workaround that lasts a week comes back as a reopen or a repeat. Read beside first contact resolution, it turns a rising figure into one of two stories. If first contact resolution rises and reopens hold steady or fall, the first line is finishing more tickets and the fixes are holding. If both rise together, the desk is closing faster, not fixing more, and the extra "first contact" tickets are the ones coming back. What a single reopen is telling you, and how to read them in bulk, is the subject of What a Reopen Is Telling You. Here the point is only that the two numbers are read on the same page or not at all.

The pairing works because reopens are hard to suppress without being noticed. Where a requester's reply to a resolved ticket reopens it, as it does in QuerySolv and in many tools, the reopen is recorded without anyone having to admit to it. In that case, the only way to hide it is to tell requesters to open new tickets instead. That is visible in the resolving messages and in the repeat-ticket search from the counting rules. The one gap is the reopen after the window, which Tom's ticket slipped through. The repeat-ticket count by requester, over a longer period than the window, is the net for that. It belongs on the monthly page beside the other two.

Pairing With Satisfaction

The satisfaction score is the requester's own rating of how the ticket went, collected by a short survey after resolution. It is the number that catches the fix the requester did not like. That could be the forty-minute call, the reply they did not understand, the "fixed" that was a workaround they could tell was a workaround. Its own series begins with What a Satisfaction Score Measures, and this article borrows one idea from it. The survey rides on the resolving reply in many tools, so a ticket closed silently, with no reply, produces no rating. A desk that closes silently will have a satisfaction figure made only of the tickets it bothered to reply to. Avoiding the Silent Close is the fix, and it is a prerequisite for this pairing to mean anything.

Read together, the two numbers ask one question of each other: did the requester agree? A rise in first contact resolution with a flat or falling satisfaction score means the desk thinks it is fixing more and the requesters do not. That gap is where the workarounds live. A rise in both is the thing itself. The rating request goes out with the resolving reply and the result lands in the standard reports in QuerySolv and in many tools. Where that happens, the pairing is a matter of putting two figures on one page. No tool makes the reading for you, and the reading is the point.

Pairing With Time and Escalations

Two more numbers round out the picture, and they catch the technician who will not hand off. Resolution time by category, read beside first contact resolution, shows whether one-contact fixes are taking longer than they should. A first line that has stopped escalating has a high first contact figure and a rising resolution time in the categories it should be handing off. The trade-off between the two is the subject of Speed Versus Quality. The escalation rate, the share of tickets handed to the second line, shows the same thing from the other side. A desk whose escalations fall to nearly nothing in a month has not become brilliant; it has become stubborn.

The table below is the reading grid Grace, who runs the Bluewater desk, keeps beside the monthly page. It is not exhaustive and the "likely meaning" column is a prompt to go and read tickets, not a verdict. Every row ends in the same instruction, which is to open the tickets that moved and find out.

When Not to Read It at All

There are months and cuts in which the number should not be read, and knowing them is part of using it. Below about a hundred eligible tickets in the period, two or three tickets changing sides move the figure by several points. In that case, a month-to-month comparison is noise. Report the quarter, or the counts without a share. In a month with a major incident, the incident's tickets are often many and often one-contact once the fix is known. They swamp the figure in whichever direction the incident ran. The honest reading excludes them with a note or waits for an ordinary month. In the first month after the counting rules change, the figure is not comparable with anything before it, and should be published with the change written beside it.

The cut that most needs this caution is the one by technician. A per-person first contact resolution figure is read for coaching, never for ranking, and not below a minimum sample. At Bluewater the floor is forty tickets of a comparable kind in the period, and Grace prefers sixty. Even above the floor the figure compares people with different queues, different shifts and different departments. The only honest use of it is the conversation "here are the tickets you handed off; what would have let you finish them?", which is the coaching method from Raising First Contact Resolution Without Cherry-Picking. A desk that ranks its technicians by this number will get a higher number from every one of them. It will get that by every means in the second section of this article, and will deserve what follows.

A Short War Story: The Cache-Clear Quarter at Kestrel Media

At Kestrel Media, a regional publisher, the desk lead Ruth was asked to raise first contact resolution and did. Over one quarter the desk-wide figure moved from about half to nearly three in four. The figure went on a slide to the board. What had moved it was a reply template, "please try these three steps and let us know if the problem continues". The first line had begun sending it as the first and final reply to most application tickets, marked resolved. Roughly one in three requesters came back inside the window and another share came back after it as new tickets, which nobody connected. Reopens doubled over the same quarter and the satisfaction score fell by most of a point. But the two figures were on a different page from the first contact figure. Nobody read them together until the editor of the largest title complained in person that her reporters were "being sent a leaflet". The cost was two months of double-handled tickets, a template withdrawn, and a board slide corrected. It was also a newsroom that for a long while described the desk as the place that sends you three steps. A number that is read alone will be raised alone, and the requester is the one who finds out how.

The Paired-Metric Check

The check below is filled in once a month, before the first contact resolution figure leaves the desk. It takes about ten minutes once the other figures exist. Section A is the figure with its counts. Section B is the three partners on the same page. Section C is the reading, chosen from the grid above. Section D is the honesty questions, which must all be answered before the figure is reported. Section E is the sentence that goes on the page. If any question in D is answered "no" or "don't know", the figure is reported with that written beside it. The check does not make the number better. It makes the number mean what it says.

FIRST CONTACT RESOLUTION — PAIRED-METRIC CHECK, ONE PAGE PER MONTH
Desk: ______________________   Month: ____________   Prepared by: ______________________

A. THE FIGURE (from the counting rules sheet)
   First contact resolution: ____ of ____ tickets, about ____      Last month: ____ of ____, about ____
   Mix-adjusted at last quarter's mix: about ____
   Tickets deliberately kept open for a planned second contact: ____

B. THE PARTNERS (same page, same month)
   Reopen rate: ____ of ____ resolved tickets          Last month: ____
   Repeat tickets from the same requester, same fault, within 30 days: ____
   Satisfaction: average ____ from ____ ratings        Last month: ____     Ratings sent: ____ of ____ resolutions
   Escalation rate: ____ of ____ tickets               Last month: ____
   Resolution time, median, in the three categories most often escalated: ____ / ____ / ____

C. THE READING (one row from the grid)
   First contact resolution went ______, reopens went ______, satisfaction went ______
   Likely meaning: _______________________________________________________________
   Tickets read to confirm it: ____ ____ ____ ____ ____ ____ ____ ____ ____ ____

D. THE HONESTY QUESTIONS (all answered before the figure is reported)
   [ ] Same counting rules sheet as last month; same window; same date rule
   [ ] Reopens did not rise with the figure (or the rise is explained in C)
   [ ] Satisfaction did not fall with the figure (or the fall is explained in C)
   [ ] Escalations did not collapse (a first line that stopped handing off)
   [ ] No reply template that closes with "let us know" is being marked resolved
   [ ] No resolving message tells requesters to open a new ticket if it recurs
   [ ] Per-technician figures, if read, were for coaching and above ____ comparable tickets
   [ ] Volume above the floor for a monthly share (otherwise the quarter is reported)

E. THE SENTENCE FOR THE PAGE
   "First contact resolution ____ of ____ (about ____), reopens ____, satisfaction ____;
    this means ________________________________________; next month we ____________________."

Grace added the check to the Bluewater monthly page the month after Tom's second ticket. The first one she filled in said, in section C, "figure up, reopens up, satisfaction flat; likely closing faster". Ticket 3317 was among the ten she read. The conversation with Ade was not about his figure. It was about the three things that make a spreadsheet freeze. It was about whether the desk would rather he watched a machine for a day or cleared a cache in a minute. He said, with some relief, that he had been wondering the same thing. The figure dipped the following month. The reopens dipped more. Tom's calculator has not frozen since.

Frequently Asked Questions