Skip to content
Faisal Mahmud TusherDigital marketing, SEO & analytics
with Faisal Tusher

Customer Experience, Retention & Loyalty › CX Measurement

Tagging Support Contacts by Reason

How to build a contact reason list for a support team: draft it from real contacts, separate reason from root cause, tag at close, audit and report monthly.
Tutorialintermediate13 min readUpdated

Why a reason list is the cheapest research a business has

Support hears from customers every day, unprompted, at the moment something mattered enough for them to get in touch. No survey reaches that many customers at that moment. The service and support hub sets out the loop that turns this into insight: tag every contact by cause, count the causes monthly, and send the top ones to the teams that can remove them. It also shows where the causes lead. A spike in order-tracking contacts is a delivery or communication problem, a cluster of "I thought it included" contacts is a promise problem marketing created, and a run of "how do I" questions is a product or help centre gap.

What the hub does not do is build the list agents tag with, and that is where most tagging fails. Lists grow by accretion until agents pick the first plausible tag; reasons mix what the customer asked with what the agent suspects went wrong; "other" swallows a large share of contacts and nobody reads it; and each change to the list breaks the comparison with last month. The result is a report nobody trusts and nobody acts on.

The discipline that fixes this already exists for survey answers. The tutorial on coding open-text answers by hand builds a codeframe from a sample, defines each code, checks agreement with a second coder and counts against a base. This tutorial applies the same discipline to an operational list that agents use on every contact, under time pressure, which changes the design: the list must be short, quick to choose from and stable for months.

The reasons then feed the rest of the measurement work. The voice of the customer guide reads survey comments alongside support contacts, the churn and retention metrics guide puts contact rate and complaint volume by cause on the retention dashboard, and the service recovery and complaints guide traces complaints to root causes with the five whys. All of them assume a reason list that agents apply the same way. Remarks that customers make to staff at a counter or on the phone, without asking for anything, reach no help desk unless someone writes them down, and the tutorial on recording what customers tell your staff shows how to write them down and tag each with a reason from this list.

Before you start

  • Know your channels. List every channel contacts arrive through: email, chat, phone, messaging and social replies. The support channels and self-service guide describes them. The reason list should be the same across channels so reasons can be compared, with the channel recorded separately.
  • Read the coding tutorial. The codeframe method, definitions, examples and agreement checks, is the base this tutorial adapts. Link it for the team rather than restating it.
  • Agree what the list is for. Write down the decisions the monthly report should inform, such as which help articles to write, which pages or emails to change, which operational faults to fix. A reason that informs no decision does not need its own tag.
  • Handle personal data with care. Contacts contain names, addresses, order numbers and sometimes health or financial details. These are Personal Data, and the GDPR, the EU framework for processing personal data, is applied on the site's data guides as processing only what is needed. Work from copies with identifying details removed when you draft the list, and keep reports to counts and short anonymised quotes. This is general direction, not legal advice.
  • Name an owner. One person in the support team owns the list, its definitions and its change log, and runs the monthly report with whoever owns marketing and product.

The steps

  1. Read a sample of real contacts

    Take a sample of recent contacts from every channel, spread across days and agents, and remove identifying details before anyone reads them, as the coding tutorial does for survey answers and the guide to minimising data before prompting does for AI tools. Read each contact and write one line on why the customer got in touch, in plain words. Do not start from the tags already in the help desk: they reflect old assumptions, and the point is to see what customers actually ask about now. Include contacts from quiet and busy periods, and from a campaign if one ran, so the list covers more than one kind of week.

  2. Draft reasons in the customer's terms

    Group the one-line notes into reasons and name each reason the way a customer would describe the problem: "parcel not arrived", "item does not fit", "charged twice", "cannot log in". Avoid internal names for teams, systems or processes, because agents choose faster when the tag matches what the customer said, and because the report will be read by people who do not know the internal terms. Keeping the customer's wording also helps later, when the tutorial on turning customer language into page copy lifts phrases from the same contacts.

  3. Keep it to two levels

    Arrange the reasons under a small number of groups, such as orders and delivery, products, billing, account and access, and pre-purchase questions. One group and one reason under it is enough for almost every decision. A third level multiplies the choices an agent faces and lowers agreement, and the detail it promises is better captured in the contact itself or found in the monthly review. Merge reasons that would lead to the same action; split a reason only when it would send work to two different owners.

  4. Separate the reason from the root cause

    The reason is what the customer raised; the root cause is why it happened. "Item does not fit" is a reason; an unclear size guide, a product cut differently from the rest of the range or a mislabelled item are causes. Keep the two apart. Agents record the reason at the end of every contact, which they can do reliably. Root causes are found later, for the largest reasons, with the five whys, asking why until the answer is something a team can change, as the service recovery guide does for complaints. Where an agent already knows the cause, let them add it as a separate optional field or a note, never as a replacement for the reason.

  5. Write a definition and an example for each

    For every reason write one sentence that says what belongs in it, one that says what does not and where that goes instead, and an example contact with identifying details removed. Definitions settle the borderline cases agents argue about, such as whether a late parcel that arrived damaged is a delivery or a product reason. Put the list, the definitions and the examples next to the tag field in the tool agents use, not in a separate document.

  6. Tag at close, one main reason per contact

    Ask agents to tag when the contact is resolved, not when it arrives, because the real reason is often clear only after the conversation. Record one main reason per contact, chosen as the reason the customer got in touch in the first place, so that counts add up and every contact is counted once. A contact that raised a second, unrelated problem can carry a secondary reason, reported separately. When the same customer comes back about the same problem, keep the original reason, so repeat contacts can be counted by reason rather than disappearing into general categories.

  7. Set the rule for 'other'

    Every list needs an "other" tag and every "other" needs a rule. Agents who choose it add a short note on what the contact was about. The owner reads those notes each month and does one of three things: moves contacts that fit an existing reason and tightens that reason's definition, adds a new reason when the same theme keeps appearing, or leaves genuinely one-off contacts where they are. The AI in customer service guide gives the warning sign for automated tagging, and it applies to agents too: a growing share in "other" means the categories need revisiting.

  8. Audit a sample for agreement

    Each month, take a random sample of tagged contacts and ask a second person, without seeing the original tag, to choose a reason from the definitions. Compare the two. Where they disagree, read the contact together and decide whether the definition was unclear, two reasons overlap, or the agent chose quickly. Fix definitions, merge overlapping reasons and share the borderline examples with the team. Track agreement over time as a quality measure for the list rather than as a score for individual agents.

  9. Report reasons monthly with owners

    Count contacts by reason each month, by channel and against a base the business already tracks, such as orders, active customers or bookings, so a rise in volume is not mistaken for a rise in problems. When the rate itself has risen, the diagnosis for support ticket volume that is rising uses this monthly count to rank the reasons behind the rise and name the team that owns each. Read small counts with the caution the reading small numbers guide recommends. Put an owner beside each of the largest reasons, the team that can remove its cause, with the five whys chain, a few anonymised quotes in the customer's words and an agreed action with a date. Promise problems go to marketing, page and help gaps to whoever owns them, operational faults to operations. Where contacts arrive in several languages, report by market on the same list, as the guide to supporting customers across languages and time zones asks.

  10. Change the list without breaking the trend

    The list will need to change, and every change risks making this month incomparable with the last. Change it on a fixed cycle rather than whenever someone asks, record each change with its date in a change log, and map old reasons to new ones so the history can be restated. When a reason is split, report the old combined figure beside the new parts for a few months. When a reason is retired, keep it in the history with a note. A list that never changes stops describing the business; a list that changes every week describes nothing.

The reason list: an example layout

The table shows the columns a working reason list holds, filled for a generic online retailer. The groups and reasons are examples of the shape, not a list to copy; each business drafts its own from its own contacts.

GroupReason (customer's terms)Belongs hereDoes not belong hereLikely root causes to investigateUsual owner
Orders and deliveryParcel not arrivedCustomer asks where an order isParcel arrived damaged (Products: arrived damaged)Courier delay, unclear delivery estimate, no tracking messageOperations, with marketing for the delivery promise
Orders and deliveryChange or cancel an orderCustomer wants to change items, address or cancel before dispatchReturn after delivery (Returns)No self-service change option, order confirmation unclearProduct or web team
ProductsItem does not fitSize or fit is wrong after deliveryCustomer asks about size before buying (Pre-purchase)Unclear size guide, product cut differentlyWeb team, buying team
BillingCharge looks wrongCustomer questions an amount or a duplicate chargeRefund not received (Returns)Confusing invoice line, payment retry, price shown differently at checkoutFinance, web team
Account and accessCannot log inCustomer cannot reach their accountWants to close the account (a separate account reason)Password reset email delayed, unclear login pageProduct team
Pre-purchaseQuestion before buyingCustomer asks about a product, stock or delivery before orderingQuestion about an existing order (Orders)Missing information on product pageMarketing, web team
OtherOther, with a noteDoes not fit any reasonAnything that fits a reasonReviewed monthly by the list ownerList owner

Each reason is a Touchpoint problem waiting to be located: the contact shows where in the journey the customer got stuck, and the owner column shows who can change that step. Treat the share of contacts tagged to the top reasons as a KPI only when an owner and an action stand behind it; otherwise it is a count that nobody manages. The measurement hub shows how operational measures like these sit beside perception and behaviour in one plan.

A school uniform retailer tags its contacts (illustrative)

Illustrative example — not a real result

Tinzaro School Uniforms builds its first reason list

Tinzaro School Uniforms, an invented online school uniform retailer, has 4 agents answering email and chat. Its help desk has 46 tags, built up over years, and the monthly report shows "general enquiry" as the largest at 38% of contacts. The support lead and 2 agents read 300 recent emails and chats with names and order numbers removed and draft 14 reasons under 5 groups.

They separate "wrong size" as the reason from the causes behind it. Reading the 41 wrong-size contacts in the sample, they find most mention the size guide, which gives chest measurements only, while customers measure children by age and height. That cause goes to the web team with 6 quotes in customers' words, such as "the age on the label and the size on your chart don't match".

In the first two months "other" grows from 7% to 12% of contacts. The notes show the same theme repeatedly: parents asking whether orders placed in August will arrive before the school term starts. The lead adds a reason, "delivery before term", under orders and delivery, and the share in "other" falls back to 6% the following month. A second agent re-tags a sample of 60 contacts each month; agreement rises from 41 of 60 in the first audit to 52 of 60 by the third, after two overlapping delivery reasons are merged. The web team adds age and height columns to the size guide, and wrong-size contacts per 100 orders fall from 3.1 to 2.2 over the next term.

Verify it worked

  • Open the tag field in the agents' tool. It shows the two-level list, with a definition, a "does not belong" note and an example for each reason.
  • Read a sample of tagged contacts. Each has one main reason, set at close, in the customer's terms rather than internal names.
  • Check where causes live. Tags hold reasons only; any cause sits in a separate optional field or in the monthly review.
  • Have a second person re-tag a sample blind. Disagreements lead to changed definitions, not blame.
  • Open the monthly report. Counts sit against a base, the largest reasons have an owner and an action, and the change log is dated.

If it goes wrong

  • Audit agreement is low. The list is probably too long or too deep. Merge overlapping reasons and tighten definitions.
  • The share in "other" keeps growing. The list is probably missing a theme. Read the notes and add a reason when the same theme keeps appearing.
  • Tags look confident but prove unreliable. Agents are probably guessing causes. Record the reason only, and find causes for the largest reasons with the five whys.
  • The monthly report moves nobody. It probably lacks a base or an owner. Add orders, customers or bookings as the base and name the team that owns each large cause.
  • The trend breaks after a list change. The change was probably unmapped. Map old reasons to new and show the old combined figure beside the new parts for a few months.

Next steps

Common mistakes

  • Starting from the old tags. A list inherited from the help desk reflects what someone assumed years ago. Draft from a sample of current contacts.
  • Internal names for reasons. Tags named after teams or systems slow agents down and mean nothing to the people who read the report.
  • Mixing reason and root cause. Asking agents to guess the cause on every contact produces confident, unreliable tags. Record the reason; find the cause for the largest reasons in the review.
  • Too many levels and tags. A deep tree or a long list lowers agreement and invites the first plausible tag.
  • Tagging on arrival. The reason is often clear only once the contact is resolved.
  • An unread "other". A growing "other" is the list telling you what it is missing.
  • Counts without a base. A rise in contacts during a busy month is not a rise in problems. Report against orders, customers or bookings.
  • Reports without owners. Reasons that reach no team with the power to act are counted and forgotten.
  • Changing the list quietly. Unlogged changes break the trend and turn the monthly review into an argument about the data.
  • Copying personal data into reports. Quotes need the meaning, not the identity. Remove names, numbers and details before anything leaves the support tool.

Frequently asked questions

What is the difference between a contact reason and a root cause?

The reason is what the customer got in touch about, described in their terms: a parcel has not arrived, an item is the wrong size, a charge looks wrong. The root cause is why that happened: a courier delay, an unclear size guide, a confusing line on the invoice. Agents can record the reason reliably at the end of every contact. The root cause usually needs investigation across many contacts, so it is worked out for the largest reasons in the monthly review rather than guessed by each agent.

How many reasons should a contact reason list have?

Few enough that an agent can choose the right one quickly at the end of a contact without scrolling, and enough that the largest reasons point to a team that can act. There is no correct number. Start from a sample of real contacts, merge reasons that would lead to the same action, and split a reason only when it grows large and hides two different problems. Agreement between agents falls as the list grows, which the audit will show.

Can an AI tool tag contacts instead of agents?

It can suggest or assign tags, and the reason list is what it applies, so the list still has to be written, defined and audited by people. Sample its tags against a person's reading in the same way as agents' tags, keep sensitive contacts such as complaints and customers who may be in difficult circumstances routed to people, and watch the 'other' tag, because a model that cannot place a contact will put it there.

Terms used on this page

Frameworks

Where this page is used

Also related

Sources

  1. Regulation (EU) 2016/679 (GDPR) — EUR-Lex

Related