Use cases

Stop escalations that reach tier 2 without what it needs

When an agent presses Escalate, checks the note and chat against the receiving team's checklist, so the agent adds what is missing. They can still send it.

Try it on this example

Example · A stopped store sync escalated with "please check", before the reconnect was tried

Team the agent is escalating to: Tier 2 - Integrations

Agent's handoff note: Customer's Shopify sync not working. Please check.

Conversation with the customer so far, with speaker labels

[10:02] Customer (Mara Ellery): Hi, since yesterday afternoon none of our new Shopify orders are coming through to Quillbox. Everything from before about 3pm yesterday is there. We haven't changed anything on our side. [10:03] Agent (Rafael): Hi Mara, sorry to hear that. Can you tell me your store address so I can find the account? [10:03] Customer (Mara Ellery): loomandwick.myshopify.com [10:05] Agent (Rafael): Thanks, I've found it. Are your Etsy orders coming in as normal? [10:05] Customer (Mara Ellery): Yes, Etsy is fine, it's only Shopify. Stock levels aren't updating on Shopify either, so we've already oversold two candles. [10:06] Customer (Mara Ellery): I emailed about this last night and heard nothing, which is why I'm on chat now. [10:07] Agent (Rafael): I'm sorry nobody got back to you. Do you see an error message anywhere in Quillbox? [10:08] Customer (Mara Ellery): Not that I can see. The Shopify tile on the connections page just says "Last synced 14:52 yesterday". [10:09] Customer (Mara Ellery): We're packing today's orders straight from the Shopify admin for now, but it's slow and we get about 60 orders a day. [10:10] Agent (Rafael): Understood. I'm passing this to our integrations team now and they'll have it fixed by the end of today. [10:10] Customer (Mara Ellery): OK, thanks. Please do, it's our busiest week. [10:11] Agent (Rafael): You're welcome. You'll get an email from the team.
  1. What should the agent do with this escalation before it moves to the receiving team?Try the standard fix first87%
  2. Does the escalation note state the customer's problem clearly enough for someone who has not read the conversation?No87%
  3. Does the escalation note say what was already tried and what happened?No84%
  4. Does the escalation include the error text, reference or log the receiving team's checklist asks for?Missing98%
  5. Does the escalation give the device, version, store or account details the receiving team's checklist asks for?Only in the conversation78%
  6. How badly does the conversation show the problem affects the customer?Degraded96%
  7. Is the team named in the escalation the right one for this problem, under the team descriptions in the context?Yes98%
  8. Does the escalation note leave out or contradict something important the customer said?Yes94%
  9. Was the standard fix from the context tried, when one applies to this problem?Not tried100%
  10. Does the customer say they have contacted the firm before about this same matter?Yes98%
  11. Does the agent promise the customer a fix, a time for a fix, a credit or a refund on the receiving team's behalf?Yes98%

These are real answers stored from one run on this example.

The prism behind it

Stop escalations that reach tier 2 without what it needs11 questions

Fields

  • Team the agent is escalating to
  • Agent's handoff note
  • Conversation with the customer so far, with speaker labels

Context

We are Quillbox, an inventory and order management app for small online shops. Tier 1 agents answer customers by chat and email. When an agent presses Escalate, the handoff note and the conversation are checked here against the checklist of the team the agent picked. The agent sees what is missing before the ticket moves, and can still send it with a reason. Tier 2 reads the note first and the conversation only when it needs to, so the note must stand on its own. Every escalation needs: what the customer is trying to do and what goes wrong, since when, what tier 1 tried and what happened, and how badly the customer is affected. Receiving teams, what each needs, and the standard fixes tier 1 tries first: - Tier 2 - Integrations. Connections to store platforms and marketplaces, order and stock sync, the API and webhooks. Needs: the platform and the store address or account ID; which records are affected (orders, stock, products) and whether old records are affected or only new ones; the exact error text, or the sync log reference (SL- followed by digits) from Settings > Connections > Sync log. Standard fixes: when a sync has stopped with no error, reconnect the store from Settings > Connections > Reconnect, then press Sync now. When one product will not sync, check its SKU is not blank. - Tier 2 - Billing. Plans, invoices, failed card payments, refunds. Needs: the invoice number or the date and amount of the charge, and what the customer expected to pay. Standard fix: for a failed payment, ask the customer to update the card under Settings > Billing and retry. - Tier 2 - Mobile app. The iPhone and Android apps. Needs: the phone model, the operating system version and the app version; the steps that lead to the problem. Standard fixes: update the app; sign out and back in. - Tier 2 - Data and reports. Missing, duplicated or wrong records inside Quillbox, imports, exports and reports. Needs: an example record (an order number or SKU) and what it should show. Standard fix: for a failed import, download the error file from the import screen and correct the rows it lists. - Account security. Sign-ins the customer did not make, a suspected takeover, locked accounts after failed two-step codes. Needs: when the customer noticed and what they saw. No standard fix: escalate at once. What tier 1 may tell the customer: that the issue has been passed to the named team, and that the team replies within one working day. Tier 1 may not promise a fix, a time for a fix, a credit or a refund on the receiving team's behalf.

Questions

  1. What should the agent do with this escalation before it moves to the receiving team? Choice

    Judge the note and the conversation against the receiving team's checklist and standard fixes in the context. This is a suggestion for the agent, who can still send the escalation with a reason. When more than one option fits, pick the one lowest in the list.

    • Send as it is The note states the problem, what was tried and the impact, everything on the team's checklist is in the note or the conversation, no standard fix is left untried, and the team is right.
    • Add the missing information first The team is right and no standard fix is left untried, but the note is unclear or something on the team's checklist is missing.
    • Try the standard fix first A standard fix in the context applies to this problem and the conversation does not show it was tried, so tier 1 may still solve it.
    • Send to a different team The problem belongs to another team in the context, whatever else the note lacks.
  2. Does the escalation note state the customer's problem clearly enough for someone who has not read the conversation? Yes / No

    Read the note on its own, as the receiving team will. Clear means it says what the customer is trying to do, what goes wrong and since when. Words such as "not working, please check" do not describe the problem. Yes: A reader of the note alone would know what is wrong and since when. No: The reader would have to open the conversation, or ask the customer, to learn what is wrong.

  3. Does the escalation note say what was already tried and what happened? Yes / No

    Count steps taken by the agent or the customer, such as reconnecting, retrying, updating the app or checking another channel, when the note says what the result was. Steps that appear only in the conversation do not count here. Yes: The note names at least one step tried and its result. No: The note names no step tried, or names one without its result.

  4. Does the escalation include the error text, reference or log the receiving team's checklist asks for? Choice

    Look in both the note and the conversation. Use the checklist for the team named in the target team field. A description of what the screen shows is not the error text unless the checklist accepts it.

    • Included The error text, reference or log the checklist asks for appears in the note or the conversation.
    • Missing The checklist asks for an error text, reference or log, and neither the note nor the conversation gives it.
    • Not required for this team The receiving team's checklist asks for no error text, reference or log.
  5. Does the escalation give the device, version, store or account details the receiving team's checklist asks for? Choice

    Use the checklist for the team named in the target team field. Look for details such as the platform, the store address or account ID, the phone model and app version, or an example record.

    • In the note Every set-up detail the checklist asks for is written in the note.
    • Only in the conversation The details are all given, but at least one appears only in the conversation and not in the note.
    • Missing At least one set-up detail the checklist asks for appears in neither the note nor the conversation.
    • Not required for this team The receiving team's checklist asks for no set-up details.
  6. How badly does the conversation show the problem affects the customer? Choice

    Judge from what the customer says about their work or their shop, not from how serious the fault sounds.

    • Blocked The customer cannot use the service for the job they use it for, and has no way round it.
    • Degraded The service works badly, or the customer has a slower way round the problem.
    • Minor A cosmetic problem or a question, with no effect on the work.
    • Not stated The conversation says nothing about how the problem affects the customer.
  7. Is the team named in the escalation the right one for this problem, under the team descriptions in the context? Yes / No

    Compare the problem in the conversation with what each team in the context handles. Yes: The problem belongs to the team named in the target team field. No: The problem belongs to a different team in the context.

  8. Does the escalation note leave out or contradict something important the customer said? Yes / No

    Compare the note with the customer's words in the conversation. Important means a fact that changes how the receiving team would investigate: when it started, which records or devices are affected and which are not, what the customer already did, or a second problem. Yes: The note leaves out or gets wrong at least one such fact. No: The note carries every such fact the customer gave, or the customer gave none beyond what the note says.

  9. Was the standard fix from the context tried, when one applies to this problem? Choice

    Use the standard fixes listed for the receiving team, and for the team the problem belongs to if that is different. Judge from the conversation and the note.

    • Tried A standard fix applies, and the note or the conversation shows it was tried.
    • Not tried A standard fix applies to this problem, and nothing in the note or the conversation shows it was tried.
    • No standard fix applies No standard fix in the context fits this problem, or the team has none.
  10. Does the customer say they have contacted the firm before about this same matter? Yes / No

    Count any earlier call, chat, email or ticket about the same problem that the customer mentions. Yes: The customer says they contacted us before about this problem. No: The customer mentions no earlier contact about it.

  11. Does the agent promise the customer a fix, a time for a fix, a credit or a refund on the receiving team's behalf? Yes / No

    Our rule: tier 1 may say the issue has been passed to the named team and that the team replies within one working day. Tier 1 may not promise a fix, a time for a fix, a credit or a refund on that team's behalf. Yes: The agent promises any of these, such as "they will have it fixed today". No: The agent only says the issue is passed on, or gives the one-working-day reply time.

Lens columns

suggested_step, suggested_step_probability, issue_clear, issue_clear_probability, note_steps_tried, note_steps_tried_probability, evidence_included, evidence_included_probability, setup_details, setup_details_probability, impact, impact_probability, right_team, right_team_probability, note_misses_facts, note_misses_facts_probability, standard_fix, standard_fix_probability, repeat_contact, repeat_contact_probability, promised_for_team, promised_for_team_probability

Run it on your own text

Add this prism in the app, change any question, and test it on a file of your own.

Ask for an invite