Skip to content
Help Center

Support: why customers contact you and how well you answer ​

The Support tab turns your Gorgias helpdesk tickets into a dashboard. Open it at Insights > CX, then select Support.

Commander puts each ticket next to the order it is about. The tab counts only tickets from customers who need something, and it says how many tickets it left out. It opens with one line on the data, then a short list of what to do about it. Below, it ranks the reasons customers write in as a list with the selected reason beside it, and keeps the rest behind collapsed rows.

Problem and intended user ​

The ecommerce manager (the person who runs the store) wants to know why customers contact the store, and whether a product or the delivery is the cause. The media buyer (the person who runs your paid ads) and the creative strategist (the person who plans and makes ad creatives) want to know whether an ad promises something that customers then complain about.

The problem arises when tickets pile up. A rising reason for contact stays hidden until the complaints become loud. Automated mail and notifications make the pile look bigger than the real customer demand.

Current workflow ​

Most merchants read tickets in the helpdesk's own reports. To connect a ticket to a product or an ad, they export tickets to a spreadsheet and match them to Shopify orders by order number.

Prerequisites ​

  • A Gorgias helpdesk account. The Gorgias connection is not open yet. The preview on this tab says: "Gorgias is still reviewing the Commander app. You can connect as soon as it is approved."
  • Your Gorgias account linked to your Shopify store, inside Gorgias. Without this link, Commander cannot match tickets to orders.
  • Your order data, for the per-order and product views.
  • For the ads card: a connected ad account. See Meta Ads and Google Ads.
  • For the NPS card: NPS answers. See NPS.

A section that lacks one of these leaves the page and counts in one link at the end of the data line, for example 3 sections not available yet. Select it to open a list headed Not available yet. The list says what each section needs, in plain words, and shows how close it is when Commander can count it. The page does not show zeros.

How Commander works ​

Once Gorgias is connected, Commander imports the last 12 months of tickets first, then keeps reading new and updated ones. It reads your tickets, messages and satisfaction surveys. It never sends messages and never changes anything in Gorgias.

Fixed rules compute every number, headline and signal. AI writes only the short notes under a signal, from your ticket text. You approve nothing here. The claims block only flags copy. You edit the copy yourself.

Customer issues and excluded tickets ​

Commander puts every ticket in one group, by fixed rules:

  • Customer issue. A customer wrote in and needs something. Only these tickets count in the headline numbers and the ranked issues.
  • Not from customers. Automated mail, a notification from another service, a ticket with no message from the customer, a ticket your helpdesk tags as automated, a vendor pitch or spam.
  • Subscription records. Notices from a subscription app and retention calls. Commander counts them on their own, even when an automated address sends them.
  • Thanks with no request.

A ticket that your helpdesk tags "auto-close" or "Automatic" still counts when Commander reads it as a real customer request. A ticket that Gorgias marks as not support never counts, whatever Commander reads.

Commander imports every ticket. Two kinds carry a label: a ticket that Gorgias marks as not support, and a ticket from an automated sender. An automated sender is an address that sends many automated emails and rarely gets an answer. Recent tickets lists these tickets with the label Not support or Automated sender. The left out breakdown shows them under the same two names. No number, signal, chart or AI note counts them.

Subscription records are the one exception. Commander counts them on their own, even when an automated address sends them. A store whose cancellation notices come from one automated address keeps its subscription records.

Each customer issue is then sorted into an issue, such as Where is my order or Return, refund or exchange. A few issues are filled in from the helpdesk's own labels for tickets Commander has not read yet. Those issues carry the word Provisional. Select it to read what it means.

Read the data line ​

One quiet line at the top tells you what the headline numbers and the ranked issues count, for example "678 customer issues · 91% linked to orders · 5-week baseline · 387 left out".

  1. Customer issues: how many tickets are real customer requests.
  2. Linked to orders: the share of customer issues tied to a Shopify order. Select the (i) beside it for what it means.
  3. Baseline: for example 5-week baseline. While it says Building baseline, 2 of 4 weeks, trends are not ready yet.
  4. Left out: select the number to see what Commander left out, by kind and by reason, and how many issues are not sorted yet. Tickets labelled Not support or Automated sender are in this number, each with its own reason. Select it again to close it. The (i) beside it says what left out means.
  5. Comparison unavailable: earlier data still syncing ends the line while the previous period is still syncing.

INFO

Tickets that were left out are out of the headline numbers and the ranked issues. The findings from signals and claims and the Ads and NPS row still count every ticket, except the tickets labelled Not support or Automated sender. Nothing is hidden: the count and the reasons stay one tap away.

Read the dashboard ​

  1. Open Insights > CX.
  2. Select Support.
  3. Set the date range at the top of the page.
  4. Read the data line and the headline numbers under it, then start at What to do about it, then What customers contact you about. The detail behind the numbers is at the bottom, in More detail.

Dependencies ​

The dashboard appears after the first tickets arrive. Until then the tab shows Importing your Gorgias history. While the import runs, totals for earlier periods can grow, and the data line says so.

A trend needs weeks of history. A change against the previous period needs that period to be imported too. Until it is, the tab says Comparison unavailable: earlier data still syncing once, in the data line, and the headline numbers show no change.

Exact output ​

The dashboard ​

The tab holds these blocks, from top to bottom.

BlockWhat it shows
Data lineCustomer issues, linked to orders, the baseline, left out (select the number to see why) and, while the earlier period is syncing, Comparison unavailable: earlier data still syncing. A link such as 3 sections not available yet opens the list of sections that cannot be shown yet. The note that your Gorgias history is still importing sits under the line.
What to do about itUp to five findings, Act first, then Fix copy, then Watch. Each has a bold title, a muted line, one figure and See why, which opens the evidence in place. The first one is open.
Headline numbersOne line under the data line: Satisfaction, Median first response, Median resolution and Open, each with its change against the previous period. Where it is known, Median first response says "up from 2h 40m" or "down from 5h". While many tickets are still being sorted, the line says Partial while tickets are being sorted.
What customers contact you aboutThe issues of the date window, ranked by customer issues. A list on the left, the selected issue on the right, with collapsed rows of evidence under its actions. Switch to a table with Ranked and Table.
More detailOne card of closed rows, each with a one-line summary: Tickets per order (with when customers ask where their order is, and the shipping status), Customer segments, Product complaints, Ads and NPS, and All tickets and team detail. Each appears once there is data for it. Until then it is counted in the sections not available yet link.

Under a signal's See why, an AI note can appear, with What customers say, What to try and a toggle for customer quotes. It ends with "Check it before you act on it."

What to do about it ​

The card at the top has up to five rows, always in this order: Act, Fix copy, Watch. Each row has a badge, a bold title, a muted line, one figure and a See why button. Select See why to open the evidence in place, and Hide why to close it. The first row is open when you arrive. Commander builds the rows by fixed rules from data already on the page:

  • Act: a High or Medium signal, such as "Subscription tickets spiked this week", with this week next to the weeks before ("44 vs 27"). See why shows what customers say and what to try (the AI note, when there is one), the customer quotes behind a toggle, and the line that AI wrote it.
  • Fix copy: the claims in your ads and landing pages that your tickets contradict, for example "3 claims in your ads and pages that tickets contradict". It links to the first flagged ad or page. See why lists the claims, the products and the generated copy that makes them.
  • Watch: a Low signal; an issue that is up, such as "Where is my order is up" with "25 vs 6" (this week against a usual week), whose title selects that issue in the list below; and an ad that brings more tickets than your average, such as "One ad brings 2.5× the tickets".

With nothing to report, the card shows up to three headlines instead: the issue that is up the most, the top reason customers write (left out while most customer issues are not sorted yet), and the issue that leaves customers least satisfied. While many tickets still wait to be sorted, it shows one line, Still sorting tickets, with how many are sorted. With no customer issues it says No customer issues in this window.

The ranked issues ​

What customers contact you about is ranked by customer issues. Ranked is the default. Table shows the same issues as a table. Commander remembers your choice in this browser.

Ranked is a split view. The list is on the left, and the issue you select is on the right. The top rising issue is selected when you arrive, or the first issue when none is rising. A link with an issue in it opens on that issue. The detail stays in view while you scroll the list, and it fits one screen.

Each row of the list shows a rank number (gold for 1, silver for 2, bronze for 3), the issue, the trend and the tickets this week. The trend is three small bars and plain words, such as Rising, about 4× usual (this week against a usual week). The more bars are lit, the further this week sits from the issue’s usual level. Point at it for the number in standard deviations. Stable is one grey bar, and Building baseline has none.

Select a row to read that issue. On a phone the list comes first. Select a row to open its detail, and select All issues to go back to the list.

The detail of an issue shows:

  • the issue, a small rank number, a trend chip such as Rising with about 4× usual (select it to read the number in standard deviations), Start here on the top rising issue, and Provisional when it applies;
  • one sentence on what changed;
  • three facts: This week (tickets of the last 7 days), Usual week (or Building, 2 of 4 weeks while the baseline is building) and Satisfied ("88%, 34 rated", or Few ratings);
  • a chart of the last 9 weeks. Select the (i) above it to read what Tickets per 100 orders and Usual range mean;
  • the actions: Copy a brief, Copy link (an icon), See tickets, and the suggested owner, such as Owner: Operations;
  • collapsed rows of evidence for this issue, each with a one-line summary, only when there is data: When customers ask (for where is my order), Who writes (customer segments whose top reason matches), Claims at risk and What customers say.

The actions ​

  • Copy a brief copies a short plain-text summary of the issue to paste to a colleague: the issue, the date range, this week against a usual week, the trend in words, tickets per 100 orders, satisfaction and the suggested owner. A message says Brief copied.
  • Copy link copies the address of this page for this issue. A message says Link copied. Whoever opens the link lands on the Support tab with that issue selected.
  • See tickets opens All tickets and team detail on Recent tickets and scrolls to it. It does not filter the tickets to this issue.

The trend ​

The trend compares each issue with its own recent weeks, per 100 orders, so a busy shipping week does not hide a smaller issue that is growing.

You seeIt means
Rising with words such as about 4× usualThis week is well above the issue's usual level. Select the words to read how far in standard deviations (the typical week-to-week swing), for example "2.4 standard deviations above usual for this issue". A small count does not count as rising.
FallingThis week is well below the issue's usual level.
StableThis week is within the issue's usual range.
Building baselineThe issue has too few full weeks of history. No trend is shown yet.

Rising is amber, never red: it is a signal to look into, not an error. Falling is green.

The weekly chart ​

  • Each grey bar is one week’s tickets per 100 orders. The dashed line is the issue’s usual rate, and the light band around it is its usual range: its average plus or minus two standard deviations. A week outside the band is unusual for this issue.
  • This week is picked out, in amber when the issue is rising.
  • Point at or tap a week to read its tickets, orders and rate. A screen reader reads the same weeks as a table. Table shows every issue’s numbers.

An issue that is still building its baseline draws its weeks without a band. A missing bar means the week is not imported yet, had no order data or had too few orders.

Table keeps the table with Tickets, Per 100 orders, vs baseline, Last 9 weeks and Satisfaction. Select the small chart in a row to read each week's tickets and rate. On a phone, the vs baseline line has its own (i) with the numbers behind it.

Satisfaction is the share of rated tickets that scored 4 or 5 stars, with the number of rated tickets beside it, for example "88% of 34 rated".

While many tickets are still being sorted, a line above the issues says Sorting tickets: N of M done, with a bar, and placeholder rows stand in for the issues. The real ones return when most tickets are sorted.

All tickets and team detail ​

The last row of More detail, All tickets and team detail, holds the blocks behind the numbers. It is one closed row. Select it to open it, and again to close it. Select any other row of More detail to read it: Tickets per order, Customer segments, Product complaints and Ads and NPS are the same blocks as before, one row each.

Open, it is the same split view as the issues, without rank numbers. The list on the left has six sections, each with an icon and one figure. The section you select is on the right. On a phone the list comes first, and All 6 sections goes back to it.

  • Ticket volume: tickets Created and Closed per period, as a chart.
  • Channels and tags: where tickets come from, and the top tags.
  • Satisfaction survey: the survey scores.
  • Team performance: One-touch resolution and Messages per ticket.
  • What customers said: low survey ratings with comments, and first messages from the biggest themes. Commander removes emails and phone numbers.
  • Recent tickets: each ticket with the order it is about. It lists every ticket, including a ticket labelled Not support or Automated sender. Those two labels mean no number counts the ticket. "Likely" means nothing names an order, so Commander picks the latest order before the ticket. The list is read only when you select Recent tickets, or when you select See tickets on an issue.

A line inside says that Channels and tags count every ticket except those labelled Not support or Automated sender, and that Recent tickets lists all of them.

Before Gorgias is connected ​

A store without Gorgias sees two quiet invitations.

The Support tab preview. The tab draws the whole dashboard from sample data, blurred and faded, under a card headed See why customers contact you. The card lists what you get after you connect. A line says: "The dashboard behind this card shows sample data." While the connection is not open, the card shows the waiting-for-approval message instead of a Connect Gorgias button.

The Top findings row. At Insights > Top findings, one row sits after the last finding when no filter is active. It is headed Add findings from your support tickets. It shows no sample findings and shows the same waiting-for-approval message.

Both previews have a Hide this preview button. Select it and both go away for your store in that browser. A message says: "Gorgias previews hidden. You can connect Gorgias any time in Settings › Integrations." The Support tab then leaves the page until Gorgias is connected.

What to do with it ​

What you seeDecision or actionOwner
A Rising issue, or an Act findingSelect Copy a brief to hand the issue to the person who owns it. Find the cause in your store: a product, a page, a delivery partner or a policy. Fix it outside Commander.ecommerce manager
Above store average under Product complaints in More detailCheck the product's quality, size guide or packaging. Decide whether to keep advertising it.ecommerce manager
A Fix copy finding, or an item under Generated copy that makes these claims in its See whyChange the wording of the copy, outside this tab.creative strategist
An ad finding under Watch, or Ads and NPS in More detailCompare those ads with the promise they make. Review their budget in Meta Ads Manager or Google Ads, outside Commander.media buyer
A large number next to left outOpen it and read the reasons. If real customers are in it, check how your helpdesk tags tickets.ecommerce manager
A low Customer satisfaction (4–5 stars) or a slow Median first responseReview staffing and replies in Gorgias, outside Commander.ecommerce manager

Support findings also reach Insights > Top findings, under the CX chip.

Expected value and measurement ​

This tab can help you notice a growing reason for contact earlier, and stop ad claims that tickets disprove.

To check a fix, open Insights > CX > Support. Pick a date range after the fix and one of the same length before it. Compare the tickets and the Tickets per 100 orders of the issue in What customers contact you about, and check that its trend moves towards Stable or Falling. The chart in the issue's detail shows the last 9 weeks. For delivery fixes, compare Tickets per 100 orders.

You cannot isolate the effect of one change. A season or a promotion can move tickets too.

Limitations and exceptions ​

  • Gorgias only. Commander reads no other helpdesk.
  • Not open yet. No store can connect Gorgias until Gorgias approves the Commander app.
  • AI notes can be wrong. Read the quotes before you act.
  • English copy only for claims. For another language, the tab says the copy was not checked.
  • Labelled tickets are not counted. Gorgias marks some tickets as not support, such as auto-replies and newsletters. Commander also labels the tickets of an address that sends many automated emails and rarely gets an answer. Recent tickets shows these tickets with a label, the left out breakdown counts them as Not support or Automated sender, and the tab leaves them out of every other number. Subscription records are the exception: Commander counts them even when an automated address sends them. Commander does not link an automated sender's tickets to orders. Commander cannot change a label.
  • Ticket matching is partial. A "Likely" order is a guess. A product complaint counts against the product, not the variant. The share linked to orders shows how much of the data is matched.
  • Sorting is partial. Customer issues that Commander has not sorted into an issue are counted in the data line and the headline numbers but not in the ranked issues or the table. A Provisional row can change when Commander reads more tickets.
  • Trends need history. An issue shows Building baseline until it has enough full weeks. A store that connected recently sees fewer trends.
  • One Gorgias account, several stores. Per-order rates can read high. The tab then shows a note.
  • A lost connection. The tab asks you to reconnect in Settings > Integrations. Your existing support data stays.

When a block is empty ​

MessageWhat to do
Importing your Gorgias historyWait. The dashboard appears when the first tickets are in.
"No tickets were created in this period. Pick a longer date range to see your support data."Widen the date range.
"No customer issues in this window"Widen the date range, or select Left out to see whether tickets were left out.
"None of these customer issues are sorted into an issue yet"Wait. Commander is still sorting them into issues.
Still sorting tickets or Sorting tickets: N of M doneWait. The findings wait too, and the issues show placeholder rows until most tickets are sorted.
Comparison unavailable: earlier data still syncingWait until the earlier period is imported. Current numbers are right.
A link such as 3 sections not available yetSelect it. The list says what each section needs: more data, a connection or an answer set. The section appears by itself once it has enough.
"Could not load your support data."Select Try again.

Demo Mode ​

In Demo Mode the tab shows a sample Gorgias account with a Sample data badge. The preview, Hide this preview and the Top findings row do not appear.

Commander is built by Omniconvert.