ANSWER

How to Find Out Why Customers Are Contacting Support (at 10k–100k Tickets a Month)

Classify the conversations, find the real cause behind “Other”, and rank it by revenue before you act.

Try Rippit free
or request demo
By Mike Nucci · Published October 2, 2026

At 10,000–100,000 support conversations a month, the most reliable way to understand why customers contact you is to classify the conversations themselves, not count the tags agents select while closing tickets. Then find the real cause behind everything tagged “Other,” give each cause to the team that owns the fix, and rank the list by the revenue of the accounts it touches.

With Rippit, an agent you set up classifies every conversation you connect by contact driver and root cause, and joins it to the account, billing, usage and release data that shows who it touched and what changed.

At 50,000 conversations a month, the answer to “why are customers contacting us?” usually comes from a helpdesk dropdown picked under handle-time pressure.

The data exists, but a dropdown picked under time pressure doesn’t measure why customers called.

Key takeaways

  • Agent-applied tags fail at volume in four predictable ways: taxonomy drift, category collapse, first-touch bias, and an “Other” bucket that absorbs everything new.
  • “Other” is where your newest problems hide. Classify what’s in it by root cause, and give each cause an owner: product, billing, support ops or CS.
  • Rank drivers by the revenue of the accounts they touch, not only by volume, so the fixes that protect renewals go first.
  • Before acting on any driver list, run two tests: hand-label 50 conversations and check where people disagree, then re-run the classification and check the numbers hold.
  • A driver list is only useful with trend and example conversations attached.

Why do support tags stop working at scale?

1. Taxonomy drift

Two agents read the same conversation and pick different categories. A payment that failed because a card expired is Billing to one agent and Account to another. Both are defensible.

2. Category collapse

One broad tag becomes the default. “Login issue” absorbs password resets, SSO failures, locked accounts. Your largest category becomes your least informative one.

3. First-touch bias

The tag reflects what the customer said the problem was. A conversation that opens with “Where is my order?” and ends with a broken tracking page gets counted as Order status: the symptom, not the root cause. That’s a B2C example. In B2B, a conversation that opens with “Our SSO broke after the update” gets counted as an account issue, when the cause is the release.

4. The “Other” sink

Anything genuinely new has no tag yet, so it lands in Other or the closest bad fit. Your newest problem is structurally invisible. It’s also the one most likely to surprise CS at a renewal, because nobody has been asked to fix it.

You asked people to perform data entry as a side effect of helping customers.

What should you measure instead of support tags?

Classify the actual conversation, on several fields at once, across the full dataset:

  • Why the customer contacted you, and what they were trying to accomplish
  • The underlying root cause, and the product or workflow involved
  • Whether it was resolved, repeated or preventable
  • How sentiment changed
  • Which account it came from, that account’s plan and revenue, and which team owns the fix
  • What changed just before it started: a release, a plan or usage limit, or a policy

One payment conversation might carry contact driver Payment failure, root cause Expired card, product area Mobile checkout, repeat contact Yes and resolution Unresolved, and, joined to billing data, the account’s plan and revenue. Reducing that to Billing throws away most of what you learned.

WHAT THE HELPDESK RECORDS

“My payment keeps failing in the app. Second time this week.”

“Looks like the card on file expired. Can you update it?”

Agent tag

Billing
WHAT RIPPIT CLASSIFIES · SAME CONVERSATION
Contact driver · Payment failureRoot cause · Expired cardProduct area · Mobile checkoutRepeat contact · YesResolution · Unresolved+ Account plan & revenue
One conversation, two records: a single agent tag versus five fields classified from the conversation text.

Can you change the taxonomy later?

Yes, if you keep the conversations, not just the labels.

Split Billing → Payment problem into expired card, bank decline and processor error, and old tagged tickets don’t follow. Someone retags them, or history breaks.

When the original conversations stay in a dataset, you can change the definition and reclassify historical conversations using the new logic. Your taxonomy can evolve with the business without losing the past.

How do you find the real cause behind contacts tagged “Other”?

Read what’s in the bucket instead of counting it, then hand each cause to the team that can fix it.

Surveys won’t find it for you. Only 16% of CX leaders say surveys give them data granular enough to find root causes, yet 93% use a survey-based metric as their main CX measure (McKinsey, 2021).

Start with every conversation tagged “Other,” plus the ones forced into the closest bad fit. Read each one in full and match it to a category and reason from what the customer said. Then do one of three things with each:

  • Retag it when an existing category fits and the agent picked the wrong one.
  • Add a subtag when it belongs to a category but needs a finer reason, such as Billing › Seat count.
  • Propose a new category when a theme fits nowhere and crosses the volume you set, such as 200 conversations in a week. Give it a name, a definition and example conversations, and approve it before it goes live.
  1. 1

    Start

    Every conversation tagged “Other” or a bad fit

  2. 2

    Classify

    Driver and root cause, from the text

  3. 3

    Join

    Account, billing, usage and releases

  4. 4

    Assign

    An owner for each cause

  5. 5

    Rank

    By revenue touched, conversations attached

Steps 1–3 replace the tag with what the conversation says and the data that confirms it. Step 4 is what turns a finding into a fix.

The conversation tells you the cause. Your business data tells you whether it’s real and who should fix it:

Root causes hiding in “Other,” the data that confirms each, and who owns the fix
Root cause typeWhat the customer saysData that confirms itOwns the fix
Product defect or regression“Our SSO broke after the update.”Release notes and the date the contacts startedProduct and engineering
Usage limit or plan change“We hit the limit again and nobody told us it changed.”Plan, limits and usage from product and billing dataProduct (pricing and packaging)
Billing or payment issue“Invoice shows the wrong seats.”Invoices, seat counts and plan from billingBilling or finance ops
Missing help content or unclear policy“How do I add a teammate?”Help center articles, macros and policy docsSupport ops and knowledge
Promotion or campaign confusionPromo code contacts at 6x normal, all tagged “Other”Launch and promo announcements in internal SlackMarketing
Account-level riskRepeat contacts from one account close to renewalARR and renewal date in the CRMCustomer success

The owner mapping will differ by company. What matters is that every cause has one, so “Other” stops being a place where problems wait.

Here’s the difference in practice. Both outputs come from the same week of conversations:

Helpdesk report

“Other” is up this week

Tells you something new is happening.

No cause, owner or account.

New category proposed · Usage limits

640 conversations this week, 4x last week

Ready for review

“We hit the limit again and nobody told us it changed.” Admin, email ticket, tagged “Other”

“Can you raise our API limit?” In-app chat, tagged “Other” → Usage limits › API limit

Check: plan and limit changes in billing and the release log. Owner: product for the limits, support ops for a macro in the meantime.

The second output gives you a cause, the customers’ own words, the data to check and an owner.

What should a contact-driver report contain?

For each driver in a ranked list, include:

What to include in a contact-driver report
IncludeWhy it matters
Volume and shareHow large is the issue?
Change vs. prior periodIs it getting better or worse?
First-seen / emergenceIs this an established problem or something new?
What changed before itDid a release, plan change or policy start it?
Resolution/outcomeWhat happens to these customers?
Accounts and revenue touchedIs it reaching the customers you most need to keep?
OwnerWho fixes it: product, billing, support ops or CS?
Example conversationsCan someone verify what the category actually means?
Top contact drivers · 50,000 conversations · last 30 days
DriverVolumeShareVs. priorRevenue touched (ARR)Owner
Payment failure · expired card4,2108.4%+12%$0.4MBilling
Login · SSO failure First seen this month3,1806.4%New$2.1MProduct
Order status · broken tracking page (B2C)2,9405.9%−5%$0.3MProduct
Refund timing2,1204.2%+1%$0.2MSupport ops
EVIDENCE · CHAT #20917

“The SSO button just loops me back to the login page.”

Resolution: unresolved
EVIDENCE · TICKET #48321

“Our whole team got locked out after the SSO change.”

Repeat contact
A ranked driver list with volume, share, trend, emergence, revenue touched, owner and example conversations attached.

Ranked by volume, payment failures lead. Ranked by revenue touched, the new SSO failure does, because in this example it lands on larger accounts. That’s the driver product needs to fix first, and the one CS needs to hear about before renewals.

If nobody can open the conversations behind a number, the meeting becomes an argument about the metric.

How do you validate AI-generated contact drivers?

Run two tests before you act on any driver list, including one an AI produced.

Test 1: Human disagreement

Have two experienced people independently classify 50 conversations against the same definitions. If disagreements cluster in some categories, fix those definitions before automating.

Test 2: Classification stability

Rerun your classifier against a known validation set. If classifications move without an intentional change to the definition or model, investigate. The goal isn’t 100% determinism. It’s enough reproducibility that a movement in your dashboard reflects customer behavior, not classification behavior.

Why analyze all conversations instead of a sample?

Sampling is useful for validation.

Your largest contact driver will probably appear in a sample. A new problem, or a low-volume issue that drives churn or compliance risk, may not.

Sample to validate the method; analyze everything to measure the business.

How does Rippit analyze contact drivers?

Rippit treats your customer conversations, and the account, billing, usage and release data around them, as one dataset that stays current.

Connect your help desk, chat and call tools with pre-built integrations, including Zendesk, Intercom and Gong, and add your CRM, billing and product data plus your own sources, such as release notes. Snowflake is a ready connector, and other tools can come in through MCP. Rippit handles ingestion, then enriches each conversation with structured and AI-derived fields as it arrives: driver, root cause, product area, resolution.

Someone on the support ops team sets up the analysis in plain language, without needing an engineer to start, and approves any new category before it goes live. The resulting fields become reusable parts of the dataset, so next month’s question doesn’t start from scratch. When a cause grows fast, the agent sends it to the team that owns the fix in Slack or email, with the conversations and the accounts behind it.

At Klaviyo, Rick Franke, VP of Global Support Services, puts it this way: “The product team was running at a five-minute mile, and we were running at a seven-minute mile. What Rippit has done is sped up the process so now we’re all running a four-minute mile.”

Which agents should you run once you know your drivers?

A monthly driver list tells you what happened. Refund requests can double while CSAT holds steady, and a new cause can sit in “Other” for weeks before anyone looks. These agents keep watching, and each one sends its finding to the team that owns the fix:

Documented agents and custom agents that catch what “Other” hides, and who gets each finding
AgentWhat it catchesWho actsRuns
Anomaly-detection agent · Custom agentA contact spike the day it starts, even when it’s mis-tagged, with the likely cause: a promo, a missing macro or a product regressionSupport ops, with marketing or product as the cause requiresDaily
Churn-Risk Bug Watch · Documented agentBugs hitting accounts with upcoming renewals, ranked by the ARR at stake, with repro steps and the conversationsProduct and engineering, with CSWeekly
Launch Feedback Agent · Documented agentWhat customers said about a release, by account: who got stuck, where, and the revenue behind itProductDaily or weekly after a release; each readout shows days since release
Policy-friction agent · Custom agentRefund, cancellation and SLA policies ranked by the negative sentiment they cause and the revenue behind them, with quotesSupport and ops leads, with the policy ownerMonthly
Macro and help-center gap agent · Custom agentQuestions agents answer by hand again and again with no macro or article behind them, with a draft for each gap to reviewSupport ops and knowledgeWeekly

Start with the one that matches your biggest blind spot. If “Other” is growing, start with the taxonomy itself (below). If you can’t tell which drivers threaten renewals, start with Churn-Risk Bug Watch.

Who is this approach best for?

BEST FOR
  • CX, support ops and product teams at 10k–100k conversations a month whose tagging taxonomy has drifted, and the CS and product teams who own the fixes
  • Leaders who need a defensible answer to “why are customers contacting us?”
  • Teams whose conversations live in Intercom, Zendesk, Gong or the other tools Rippit connects, with account, billing and usage data alongside
NOT A FIT
  • Teams under a few hundred conversations a month, where reading them is faster and more accurate
  • Teams that only run surveys and have no conversation volume to analyze
  • Teams that want a general BI tool rather than analysis of conversation text

Comparing customer insights tools? See how Rippit stacks up against Enterpret, Unwrap and Zendesk.

FAQ: finding out why customers contact support

Can’t we just fix our tagging taxonomy instead?

Improve it, but tags stay a quick judgment by a rotating set of people, so drift and first-touch bias return. Classifying the conversation text removes the data-entry step and lets you re-derive drivers historically when definitions change.

What should we do with contacts tagged “Other”?

Read them in full, classify each by driver and root cause, then retag, add a subtag or propose a new category, and give each cause an owner. See the “Other” section above.

How many conversations do we need to analyze?

All of them. A sample gets your biggest driver roughly right. Rare, severe and brand-new issues live in the long tail, and a sample is designed to exclude it.

How long does it take to get a first driver list?

Connect your conversation source with a pre-built integration, plus your account and billing data if you want revenue weighting, describe what you want the agent to find in plain language, and it reads every conversation you connect and delivers results on a schedule. Budget more time for the two validation tests above than for setup.

Does this replace CSAT or NPS surveys?

No. Surveys hear from the customers who respond; conversation analysis covers everyone who contacted you. Most teams run both.

Can we ask these questions from ChatGPT or Claude?

Yes. Rippit hosts its MCP server, so an admin adds one connector URL and there is nothing to deploy. Each person sees only what they can already see in Rippit, and the assistant reasons over the drivers and root causes already classified on every conversation you connect, instead of reading a sample of tickets pasted into the chat.

What’s the bottom line?

The conversation is the source data. Keep it, structure it, and make every number traceable to the conversations behind it, with the account data around it and an owner for each cause. Then you’re answering from what customers actually said, not from a dropdown.

To keep that classification current as new causes appear, run it as a weekly agent, starting with what lands in “Other.”

Documented agent · runs every Monday

Taxonomy Builder

Contact reasons from what customers say, not a preset dropdown. Every week it reads each new ticket, chat and AI agent conversation against the tags you have today, and sends:

  • Your top contact reasons with the change from last week, each linked to example conversations
  • Mis-tagged tickets with a suggested retag, in a weekly list for your review; writing corrections back to your help desk needs a custom MCP connection with the write tool on
  • New-category proposals when a theme in “Other” crosses the volume you set, with a name, definition and examples to approve
  • A note to the team that owns each fast-growing reason, with the conversations behind it
Sends to
SlackEmail

What is Rippit?

A contact driver classified once is a record Product can prioritize, Billing can fix and CS can check before a renewal.

Rippit is a Growth Data Platform: it keeps enriching your customer conversations and other sources into structured data, connects them to CRM, billing and product data and each team’s own sources, and lets anyone set up agents on it, without needing an engineer to start. Connect with pre-built integrations or through our MCP, and ask from Rippit, Claude, ChatGPT or Cursor.

Support conversations (tickets, chat, call transcripts)Accounts & billingProduct usageProduct updates (releases)and more

Build your first agent on your customer and business data.

Try Rippit free
or request demo

Sources

  • McKinsey & Company, survey of 260+ US CX leaders (published February 2021)

Keep reading

More from the Rippit Library.

View the library →