Skip to main content

Case study · A six-store retail group · AI Chatbots

Each ticket checked against the order and the live stock, then drafted for one click.

Result

700 → 280 tickets a week

−60% support volume handled by floor staff, before the pilot and after

Measured, incomplete
Baseline
The prior quarter’s ticket queue
Checked by
The Founder, against Shopify reports and the POS export

The weeks the 280 was counted over are not on the record.

Floor staff at six stores were answering the support inbox between customers, and the tickets had roughly doubled in a year. A pipeline now matches each ticket to its Shopify order and the live stock in the POS, and drafts a reply a person sends in one click or rewrites.

Before opening, a staff member reaches up to one carton on a wall of plain cartons, a soft violet glow spilling from it.

A business like yours?

Business
A six-store retail group
Tickets
Roughly doubled in a year
Systems
Shopify, an in-store POS, two shared inboxes

Explainer · about 22 s · silent

Read the transcript
  1. At a six-store retail group, floor staff answered around 700 support tickets a week between customers.
  2. Each ticket is matched to its order in Shopify and checked against the live stock in the POS.
  3. A reply is drafted against the matched order, and the support agent sends it in one click or rewrites it.
  4. Support volume handled by floor staff fell 60%, from around 700 tickets a week to 280, checked by the Founder against Shopify reports and the POS export. Measured, incomplete: the weeks the 280 was counted over are not on the record.

Support queue · Doubled in twelve months

Support tickets had doubled in a year, and the floor staff were the support team.

A six-store retail group was running three systems behind the counter: Shopify for online, a POS for in-store, and two shared inboxes for customer support.

The same question arrived through every channel: where is my order, can I swap a size, did the warehouse pick the right colour. Floor staff were splitting the day between serving customers and answering tickets.

The support volume had roughly doubled across the prior twelve months. The Founder had hired a part-time helper to keep up, and the helper was already at capacity.

Build · 14 days

A drafted reply for every ticket, never sent without a person.

Not a chatbot: a pipeline that reads every incoming ticket, matches it against the order in Shopify and the live stock in the POS, and drafts a reply the support agent sends in one click, or rewrites when the situation needs a human eye.

The agent stays in the loop on every reply, and the model never sends on its own. The easy 60% of the inbox, the status checks, swap requests and stock questions, is now answered in one click.

n8n joins the shared inbox to Shopify orders and POS stock, with a reconciliation step that catches stock drift between the two. Anthropic Claude drafts each reply against the matched order and stock. Nexus, Raava’s orchestrator, handles retries, queues the agent’s review and writes the chosen reply back to the inbox.

The reply pipeline took 14 days as a pilot, then five weeks of iteration with the Founder and the lead support agent.

A ticket arrives. A customer asks where their order is, through one of the shared inboxes.

  1. A customer asks where their order is, through one of the shared inboxes.

  2. The pipeline finds the Shopify order the ticket is about.

  3. The POS and Shopify counts are compared, and a gap of more than a unit goes to a person with both numbers.

  4. A draft answer is written against the matched order and stock.

  5. The support agent sends it as drafted or rewrites it; nothing goes out on its own.

Channels ShopifyInbox Shared support inboxGlue n8nReconciliation Custom POS syncModel Anthropic ClaudeCoordination Nexus, Raava’s orchestratorPrivacy Customer-PII redaction before the model call

Stock checks · Five weeks

Five weeks on the times the POS and Shopify disagreed.

The five weeks after the pilot went on stock: the times the POS said one thing and Shopify said another. A draft reply that quoted the wrong stock figure burned more trust than no reply at all.

So a reconciliation pass runs before the model sees the ticket. If the two systems disagree by more than a unit, the ticket goes to a person with both numbers shown side by side.

We had assumed the model would take most of the effort; the systems beneath it did. The first week of the pilot caught more stock-drift bugs than the prior six months of manual reconciliation, and the Founder kept that step running long after the support pipeline went live.

The pass holds about 12 reconciliation rules.

After the pilot · Against the prior quarter

What the floor staff got back.

The pipeline now answers the easy part of the inbox in one click, leaving the tickets that need a human eye.

Numbers verified by the Founder against Shopify reports and the POS export. Anonymised under our standard case-study disclosure: vertical and outcome are real; retailer name, store locations, and POS vendor are not.

700 → 280 tickets a week

−60% support volume handled by floor staff, before the pilot and after

Measured, incomplete
Baseline
The prior quarter’s ticket queue
Checked by
The Founder, against Shopify reports and the POS export

The weeks the 280 was counted over are not on the record.

4 person-days a week

back to the floor across the operations team

Measured, incomplete

The page states the four days but not how they were counted: no baseline, window or checker is on the record.

$40K AUD a year

in support cost saved

Projection

Estimated from the part-time helper hours the Founder no longer needed to schedule; not a measured saving.

+80%

faster catalogue updates across the six stores

Measured, incomplete
Baseline
The manual reconciliation cycle that ran every Friday
Checked by
The Founder, against Shopify reports and the POS export

No time frame on the record, and it counts catalogue updates, not the ticket queue.

“My team is on the floor again. The inbox is still there, and we still read it, but it is not eating the day. That is the bit I did not know we could buy back.”

Founder · 6-store retail group · QLD (anonymised at the client’s request)

Curious whether your store has the same gap?

A free 45-minute audit. We look at your support volume, your POS, and where the floor team is losing the day. You get a written follow-up within 48 hours, whether we’d be a fit or not.