G3: connected operations across a venue, shop and online store.
Supplier changes, customer questions, staff schedules and follow-up all belong to the same business. Here is how custom systems help G3 coordinate that work.
G3 is owned and operated by Canvo founder George Ding. This is a case study from our own business.
Six connected areas of work.
Choose an area to follow its inputs, decisions and finished work. Each chapter shows where people still take responsibility.
Actual G3 capabilities, with separate review and execution boundaries—not one unattended chain.
Supplier changes become shop updates.
A morning catalogue pipeline and regular availability checks connect supplier information, local stock and the online shop. Product additions, promotions and fulfilment information become maintained operating data.
- Supplier catalogue
- Match products & variants
- Check content & pricing
- Update eligible products
- Refresh stock & ETA
Different stock sources, different promises
Stock on hand, supplier availability and preorder cases are treated separately. The shop can explain how the customer will receive the item using the available source information.
What runs, and what needs attention
- Morning catalogue run
- The inspected schedule runs at 5:30 am. It identifies additions and changes, updates commerce records, applies or reverts promotions, and checks publication readiness.
- Regular availability refresh
- Stock and promotion checks run every 30 minutes. A separate hourly import brings online orders into inventory; walk-in sales follow the staff stock-entry process. These sources do not all update at the same instant.
- Incomplete products stay visible
- Missing prices, thin descriptions or unfinished media work can prevent publication or remain for another pass. A completed catalogue update does not mean every enrichment step is complete.
Useful measures: time spent on each supplier batch, source freshness, publication exceptions and stock discrepancies.
One familiar front for customer questions.
Peter brings product information, venue knowledge and relevant live availability into the conversation. When a request needs a person, staff receive a retained ticket and the conversation context.
- Customer question
- Relevant business information
- Answer or capture contact
- Staff handover
- Reply & record resolution
A partnership enquiry needs a reachable person
A specialist request can become a ticket with a summary and contact details. Staff can take over the thread and reply through the customer’s channel.
Retail · hospitality · accommodation · venues · service businesses
Explore a related workflow →Answers, actions and the handover boundary
- Connected customer information
- The customer layer supports the website and social channels. Product answers use returned price, specification and availability information; relevant booking questions can use current availability.
- A staff-controlled finish
- Staff can reply, take over and resolve a thread. Contact capture helps the team follow up with a web visitor who has left the page.
- Authority stays explicit
- Peter does not itself confirm bookings or complete refunds. Specialist requests and uncertain answers can go to the team. We have not published an independently measured resolution percentage.
Useful measures: correctly resolved enquiries, contact capture, handover completion and staff handling time.
Guidance that leads to a recorded next step.
The staff system connects operating knowledge, shift checklists, photographs, clock events and issue records. Staff can get relevant guidance and turn a reported fault into an issue that stays visible.
- Question or checklist
- Relevant guidance
- Staff action & evidence
- Issue when needed
- Follow-through
An equipment fault should survive the chat
A reported fault can receive troubleshooting guidance and become an issue-board item. Staff retain the record, add evidence and record when it is resolved.
Multi-site teams · facilities · hospitality · warehouses · field service
Explore a related workflow →How knowledge and shift work connect
- Shared operating knowledge
- The staff assistant draws on maintained facts, relevant case cards and operating instructions. Unknown questions remain a reason to seek a manager’s decision.
- Work leaves a record
- Checklists support photographs; clock events and issues are recorded. Staff can close or reopen an issue as work progresses.
- Evidence of use
- Dated operating records show checklist activity, photographs, clock events and issue resolution. These establish use during the recorded period, rather than a measured saving or independent proof of a physical repair.
Useful measures: repeat questions, missing handovers, issue age and time to a confirmed finish.
Prepare the roster. Keep the manager’s decision.
G3 Hub prepares a roster from staff availability, preferences and configured staffing rules. Difficult slots and conflicting shift sequences are considered, with unfilled coverage left visible for review.
- Availability & preferences
- Staffing rules
- Generated draft
- Manager review & publish
- Notify staff
A gap is a decision, not a hidden assignment
If no eligible person can cover a slot, the draft shows the gap. The manager reviews coverage and changes before publishing the schedule.
Retail teams · hospitality · venues · multi-site operations
Explore a related workflow →What generation and publication each do
- Preparation follows explicit rules
- Recurring preferences can be overridden for a particular week. The generator considers desired shift counts, unavailability and selected morning/night conflicts.
- Publication is a separate action
- A manager reviews the draft and publishes it. The publication step records shifts and attempts email or SMS notifications according to staff preferences, with delivery status recorded.
- Rules match this operation
- This case demonstrates G3’s configured roster rules. A different business needs its own coverage, skills, employment requirements and acceptance checks defined before rollout.
Useful measures: preparation time, manager edits, unfilled slots and successful notification delivery.
Customer behaviour can trigger the next contact.
A deployed follow-up process checks visit history, identifies eligible first-time customers who have not returned, queues contacts and invokes sending. Return visits then feed the reporting loop.
- Visit history
- Eligibility & return checks
- Contact queue
- Controlled send
- Return reporting
A customer who came back should leave the queue
The process excludes returners and checks suppression, unsubscribe and duplicate-contact rules before sending. Follow-up responds to the customer’s state.
Venues · memberships · appointment businesses · repeat-purchase retail
Explore a related workflow →Segmentation, sending and measuring the result
- Behaviour provides the starting point
- Customer analysis includes recency, frequency and spending, alongside changes in visit patterns. The inspected sending path follows a specific first-visit return rule.
- Execution has contact controls
- Eligible contacts enter a queue. Suppression, unsubscribe and duplicate-prevention checks are part of the sending path.
- Report observed results honestly
- Return visits and associated spending provide operational feedback. They are not automatically extra revenue caused by the campaign; causal uplift needs a suitable comparison.
Useful measures: eligible contacts, successful sends, return rate and incremental contribution with a comparison group.
See what tomorrow may bring—and how today is tracking.
G3’s forecasting systems generate day-ahead predictions and intraday estimates. They give the operator another input for planning and for comparing expectations with actual performance.
- Operating history
- Day-ahead forecast
- Intraday estimate
- Actual performance
- Planning decision
A forecast is an estimate to check
As the day develops, the operator can compare the estimate with recorded performance. Forecast quality and business decisions need their own measures.
Hospitality · retail · venues · businesses with variable daily demand
Explore a related workflow →Where prediction ends and action begins
- Separate capabilities, connected decisions
- Forecasting, customer segmentation and behaviour-triggered follow-up each have a role. The verified sending workflow above is triggered by customer behaviour.
- No unverified automatic loop
- This case does not claim that every revenue forecast automatically creates or sends a campaign. That connection needs separate execution evidence.
- Measure against a useful baseline
- Compare predictions with actuals over a defined period, including quiet days and unusual events. Then assess whether the decisions made from them improve the operation.
Useful measures: forecast error, bias, decision lead time and the observed result of each action.
Operating evidence, with clear limits.
The system descriptions were checked against source and selected operating evidence on 29 September 2026. They describe what is built and the paths inspected; they do not imply every step runs without people.
What was checked and what we would measure next
- Evidence of implementation
- Catalogue and roster code was compared with deployed copies. Scheduled-job information and selected aggregate logs were inspected. Peter also has a dated end-to-end social-channel check; staff use is supported by dated operating records.
- Results still need a baseline
- This case publishes no measured saving, universal automation rate or causal revenue uplift. Before-and-after handling time, confirmed customer outcomes and forecast accuracy need a defined period and comparable work.
Questions about the G3 case
Short answers with links to the relevant operating evidence.
Is G3 an independent Canvo customer?
G3 is founder George Ding’s own retail and venue business. It provides operating experience and implementation evidence. We identify it as founder-owned so readers can assess that relationship; it is not presented as an independent customer endorsement.
Read the evidence and limits →How are supplier stock and stock in the shop kept distinct?
The commerce process uses separate sources for local stock, supplier availability and preorder information. Regular checks update the shop’s availability information, while order imports and staff entries have their own schedules. This is not a claim that all sources update at the same instant.
Follow the commerce pipeline →Can Peter finish every customer request without a person?
Peter can answer using relevant business information and pass a retained ticket with context to staff. It does not itself confirm bookings or complete refunds. Staff can take over and resolve the thread. A measured automatic-resolution rate has not been published.
See answers and human handover →Does a revenue forecast automatically create and send campaigns?
The inspected follow-up sending path responds to a defined customer-behaviour rule. Forecasting and segmentation are separate capabilities. The inspected evidence does not establish a forecast-to-campaign sending connection or a measured revenue increase caused by these systems.
Explore follow-up and its checks →Can these systems be adapted to a different business?
The patterns can inform a custom design; your sources, authority, staffing rules and acceptance checks still need their own scope. Bring one input, the result you need and an awkward exception. Start with a bounded workflow before connecting more departments or locations.
Map the approach to your business →Your industry. Your rules. A finished piece of work.
G3 shows where the capability comes from. Your system starts with your own inputs, decisions, tools and responsibilities—whether you run a distributor, service business, project team or a multi-site operation.