Chapter 03
Opportunities
Twelve opportunities, each sized from the evidence and checked against the tools you already have. Values are net of the review and exception work each change introduces.
Value against effort
How we decide
- Do first3
Clear value, proven approach, ready to start.
- Validate first4
Promising. One question to answer before building.
- Defer3
Not now. Revisit when the conditions change.
- Not recommended2
Little gain, or the risk belongs with people.
- Value
- Net hours returned each month, at $38 an hour fully loaded, plus costs avoided. Shown for conservative, expected and upside cases.
- Effort
- Build complexity, systems to connect and how much the work itself has to change.
- Confidence
- How strong the evidence is: measured, sampled or estimated.
All 12 opportunities
O1Order-status answers with live order and 3PL dataCustomer experience · Let the helpdesk answer “where’s my order?” from real order, carrier and warehouse data.Value / yr$86K$108K$127KHours / mo150 h180 h205 hEffortLowDo first
Today
3,720 order-status tickets a month, 38% of volume. Agents copy details from Shopify, the carrier site and the ShipBob portal. First response reaches 14 hours at peak.
Proposed
Switch on Gorgias’ AI agent for order-status questions, grounded in Shopify orders and carrier tracking. Add a small integration so it can see ShipBob exceptions, and hand anything unusual to an agent with the facts attached.
Where people stay in control
Refunds, address changes, damaged items and VIP customers always go to an agent. Your CX lead approves the answer rules before launch and reviews a weekly sample.
Approach
- Configure existing
- Connect systems
Live in week 5Confidence: High
The value, worked through
- 150 h180 h205 h a month × 12 × $38
- $68K$82K$93K
- Seasonal temp cover not needed
- $18K$26K$34K
- Annual value
- $86K$108K$127K
Assumptions
- Order status stays near 38% of tickets outside peak (sampled June–August).
- 4.5 minutes average handle time, from Gorgias reporting.
- 60–80% fully resolved without an agent, from our ticket sample and helpdesk benchmarks.
- About 10% of the time saved goes to review and exceptions.
Data it needs
- Shopify orders and fulfillments
- Carrier tracking events
- ShipBob exception feed
- Gorgias macros and tags
Risks to manage
- Two regional carriers lack scan events
- Wrong delivery promises if data is stale
- Customers who prefer a person
How we’ll measure it
- First response time
- Resolved without an agent
- Repeat contact within 7 days
- Satisfaction on automated answers
O2Supplier files to review-ready product recordsCatalog · Read supplier sheets, fill your product schema and flag gaps before review instead of after.Value / yr$50K$82K$125KHours / mo65 h82 h98 hEffortMediumDo first
Today
2,600 new SKUs a year arrive in more than 30 supplier formats. Each takes about 42 minutes to key, check and publish; launches wait 19 days on average for product data.
Proposed
An extraction workflow that reads supplier files, maps attributes to your listing standard, drafts titles and descriptions in your voice and flags missing care, sizing or material details at intake.
Where people stay in control
Merchandisers review every record before publication. Material, sustainability and safety claims require a source document.
Approach
- Custom build
- Connect systems
First category in week 9Confidence: Medium
The value, worked through
- 65 h82 h98 h a month × 12 × $38
- $30K$37K$45K
- Margin from launching sooner
- $20K$45K$80K
- Annual value
- $50K$82K$125K
Assumptions
- About 2,600 new SKUs a year across spring, fall and drops.
- 42 minutes per SKU, from shadowing three people on 24 SKUs.
- 45–65% of preparation time saved after review.
- Launch benefit: margin on sales pulled forward about 13 days, applied to 30% of new SKUs.
Data it needs
- Two seasons of supplier files
- Airtable product base
- Shopify product schema
- Listing standard and style guide
Risks to manage
- Supplier formats change without notice
- Invented attributes, prevented by source-required fields
- Inconsistent brand voice
How we’ll measure it
- Time per SKU
- Corrections in review
- Attribute completeness
- Days from receipt to live
O3Daily 3PL exception monitor and triageFulfillment · Find stuck orders before customers do, with the fix already drafted.Value / yr$36K$50K$65KHours / mo45 h58 h70 hEffortMediumDo first
Today
About 1,140 orders a month hit a 3PL exception. Ops finds them by exporting reports twice a day; 62% surface only when a customer writes in.
Proposed
A daily monitor that joins ShipBob, Shopify and carrier events, classifies each exception, drafts the fix and the customer update, and posts a prioritized queue to Ops every morning.
Where people stay in control
Ops approves reships, refunds and anything over $150. Customer messages use approved templates.
Approach
- Connect systems
- Custom build
Live in week 7Confidence: High
The value, worked through
- 45 h58 h70 h a month × 12 × $38
- $21K$26K$32K
- Goodwill credits and reships avoided
- $15K$24K$33K
- Annual value
- $36K$50K$65K
Assumptions
- Ops spends about 14 hours a week finding and chasing exceptions; CX about 7.
- Half of today’s credits are avoidable when the problem is caught early.
- Also reduces order-status tickets for O1; that effect isn’t counted twice.
Data it needs
- ShipBob orders and exceptions API
- Shopify fulfillments
- Carrier events
- Exception tracker history
Risks to manage
- 3PL API rate limits
- Needs the shared admin login replaced first
How we’ll measure it
- Exceptions found before the customer
- Time to resolve
- Credits issued
- Order-status tickets
O4Automated weekly trading reportFinance · Reconciled numbers and drafted commentary by Monday morning.Value / yr$16K$20K$24KHours / mo36 h44 h52 hEffortMediumValidate
Why validate first. Two definitions of net revenue are in use. Automating now would automate the disagreement.
To validate: Finance and merchandising agree one definition of net revenue and of returns timing.
Today
The Monday report takes two days of an analyst’s time and reaches leadership on Tuesday afternoon.
Proposed
Agree metric definitions, automate the data pull and reconciliation, and draft commentary on the biggest movements with links to the underlying numbers.
Where people stay in control
The FP&A analyst edits the commentary; the Controller signs off before circulation.
Approach
- Process change
- Custom build
4 weeks after definitionsConfidence: Medium
The value, worked through
- 36 h44 h52 h a month × 12 × $38
- $16K$20K$24K
- Annual value
- $16K$20K$24K
Assumptions
- Two days a week of preparation today.
- 80–90% of assembly automatable once definitions are fixed.
Data it needs
- Shopify, NetSuite and Loop exports
- Current report workbook
Risks to manage
- Definitions drift again without an owner
How we’ll measure it
- Preparation time
- Delivery time
- Questions answered in the meeting
O5Invoice and PO matchingFinance · Match invoices to POs and receipts automatically; route mismatches with context.Value / yr$17K$25K$34KHours / mo28 h36 h44 hEffortLowValidate
Why validate first. This is a solved problem. Buying beats building, if a product fits your NetSuite setup.
To validate: Confirm two shortlisted products handle your multi-currency suppliers.
Today
About 520 supplier invoices a month are matched by hand. 14% have price or quantity mismatches that take a day or more.
Proposed
Evaluate two AP automation products that integrate with NetSuite. Configure matching rules and route mismatches to the buyer with the PO and receipt attached.
Where people stay in control
AP approves every payment; buyers resolve mismatches.
Approach
- Buy a product
- Configure existing
6 weeks after selectionConfidence: Medium
The value, worked through
- 28 h36 h44 h a month × 12 × $38
- $13K$16K$20K
- Early-payment discounts captured
- $4K$9K$14K
- Annual value
- $17K$25K$34K
Assumptions
- 520 invoices a month at about 6 minutes each, plus mismatch work.
- Product cost excluded; compared in the evaluation.
Data it needs
- NetSuite POs, receipts and bills
- Supplier terms
Risks to manage
- Implementation effort varies by vendor
How we’ll measure it
- Invoices matched without touch
- Days to pay
- Discounts captured
O6Return-reason analysis for product and sourcingCustomer experience · Turn free-text return reasons into a weekly fit and quality brief.Value / yr$14K$35K$67KHours / mo8 h12 h16 hEffortLowValidate
Why validate first. The value depends on whether the free text holds a usable signal. A short pilot will tell.
To validate: A two-week tagging pilot on one season’s returns.
Today
Return reasons are free text in Loop. Nobody reads them systematically, so fit and quality issues reach product teams months late.
Proposed
Classify return reasons weekly, link them to SKUs and factories, and send a short brief to product and sourcing.
Where people stay in control
Product decides what to change; the brief only surfaces patterns.
Approach
- Custom build
Pilot in 2 weeksConfidence: Low
The value, worked through
- 8 h12 h16 h a month × 12 × $38
- $3.6K$5.5K$7.3K
- Returns avoided by earlier fixes
- $10K$30K$60K
- Annual value
- $14K$35K$67K
Assumptions
- Return rate of 14%; a 0.5–2 point reduction on affected styles.
Data it needs
- Loop Returns export
- SKU and factory mapping
Risks to manage
- Free-text reasons are too vague to act on
How we’ll measure it
- Issues found
- Return rate on fixed SKUs
O7Wholesale PO intakeFulfillment · Read retailer POs and create draft sales orders for confirmation.Value / yr$10K$14K$17KHours / mo22 h30 h38 hEffortMediumValidate
Why validate first. Modest volume today. Worth it if wholesale grows as planned; the single-person risk needs fixing either way.
To validate: Confirm the 2027 wholesale forecast. Document the macro now regardless.
Today
310 POs a month arrive as PDFs and emails. One coordinator keys them into NetSuite with a personal macro.
Proposed
Extract line items from incoming POs, check them against price lists and stock, and create draft sales orders for the coordinator to confirm.
Where people stay in control
The coordinator confirms every order and handles retailer exceptions.
Approach
- Custom build
- Connect systems
6 weeksConfidence: Medium
The value, worked through
- 22 h30 h38 h a month × 12 × $38
- $10K$14K$17K
- Annual value
- $10K$14K$17K
Assumptions
- 18 minutes per PO today; 60–75% saved.
Data it needs
- Sample POs from top 10 retailers
- Price lists
- NetSuite items and stock
Risks to manage
- Retailer formats vary widely
How we’ll measure it
- Minutes per PO
- Corrections
- Orders entered same day
O8Demand forecasting modelPlanning · Forecast demand by style and size to cut stockouts and overstock.Value / yr$4.6K$69K$164KHours / mo10 h20 h30 hEffortHighDefer
Why defer. Eighteen months of reliable history and stockout-distorted data make a custom model unreliable today. Use NetSuite’s planning features and clean SKU data first; revisit next year.
Today
Buying uses spreadsheets and last year’s sales; frequent stockouts distort the history.
Proposed
A custom forecasting model trained on sales, traffic and seasonality.
Where people stay in control
Buyers own every purchase decision.
Approach
- Custom build
6+ monthsConfidence: Low
The value, worked through
- 10 h20 h30 h a month × 12 × $38
- $4.6K$9.1K$14K
- Lower carrying cost and fewer stockouts
- $0$60K$150K
- Annual value
- $4.6K$69K$164K
Assumptions
- Value is potential, not expected, until data quality improves.
Data it needs
- Sales history
- Stockout log
- Traffic
Risks to manage
- Unreliable history
- High cost to build and maintain
How we’ll measure it
- Forecast accuracy
- Stockout rate
O9AI-generated lifestyle imageryCreative · Generate lifestyle scenes for new products without a shoot.Value / yr$2.7K$13K$26KHours / mo6 h10 h14 hEffortMediumDefer
Why defer. Small time saving and real brand risk. Your creative standards for generated imagery aren’t set yet; revisit once they are.
Today
Two lifestyle shoots a season cover hero products only.
Proposed
Generate supporting lifestyle images from product photography.
Where people stay in control
Creative approves every image.
Approach
- Buy a product
—Confidence: Low
The value, worked through
- 6 h10 h14 h a month × 12 × $38
- $2.7K$4.6K$6.4K
- Shoot costs reduced
- $0$8K$20K
- Annual value
- $2.7K$13K$26K
Assumptions
- Limited to supporting images, never product accuracy shots.
Data it needs
- Product photography
Risks to manage
- Brand and authenticity
- Product misrepresentation
How we’ll measure it
- Approval rate
- Conversion on affected pages
O10Personalized email contentMarketing · Generate subject lines and content variants for segments.Value / yr$1.8K$7.7K$14KHours / mo4 h6 h8 hEffortLowDefer
Why defer. Not a bottleneck, and Klaviyo already includes it. Your marketing team can trial it without us.
Today
Marketing writes three to four campaigns a week with light segmentation.
Proposed
Trial Klaviyo’s built-in AI features for subject lines and variants.
Where people stay in control
Marketing approves all sends.
Approach
- Configure existing
—Confidence: Medium
The value, worked through
- 4 h6 h8 h a month × 12 × $38
- $1.8K$2.7K$3.6K
- Incremental email revenue
- $0$5K$10K
- Annual value
- $1.8K$7.7K$14K
Assumptions
- Uses existing Klaviyo plan features.
Data it needs
- Klaviyo account
Risks to manage
- Off-brand copy
How we’ll measure it
- Open and click rates
O11Fully automated refundsCustomer experience · Approve and issue refunds without an agent.Value / yr$4.6K$6.4K$8.2KHours / mo10 h14 h18 hEffortMediumDon’t do
Why not recommended. Refunds are a small share of tickets but carry most of the fraud and policy risk. Keep them with agents; O1 and O3 give agents the context to decide faster.
Today
Refunds are 4% of tickets and are approved by agents within policy limits.
Proposed
Automatic approval of refunds that meet policy rules.
Where people stay in control
Agents approve all refunds today.
Approach
- Custom build
—Confidence: High
The value, worked through
- 10 h14 h18 h a month × 12 × $38
- $4.6K$6.4K$8.2K
- Annual value
- $4.6K$6.4K$8.2K
Assumptions
- Hours shown for comparison only.
Data it needs
- Refund history
Risks to manage
- Fraud
- Policy abuse
How we’ll measure it
- —
O12Custom chatbot to replace the helpdeskCustomer experience · Build a bespoke support assistant instead of using Gorgias.Value / yr$0Hours / mo0 hEffortHighDon’t do
Why not recommended. Gorgias’ own AI agent covers the need (O1). A custom build adds cost and maintenance with no clear gain.
Today
Gorgias handles every channel and already includes an AI agent.
Proposed
A custom assistant integrated with Shopify and ShipBob.
Where people stay in control
—
Approach
- Custom build
—Confidence: High
The value, worked through
- 0 h a month × 12 × $38
- $0
- Annual value
- $0
Assumptions
- No gain over O1.
Data it needs
- —
Risks to manage
- Ongoing maintenance
- Duplicated tooling
How we’ll measure it
- —