E-commerceCase study3 min read
E-commerce: the pre-purchase question that decides the cart
Shoppers abandon over one unanswered question about sizing, stock or shipping. Nobody was awake to answer it.
The business in this scenario: A furniture retailer shipping from a single warehouse, with a growing trade side
In short
One unanswered question about sizing, stock or shipping ends a session. The store put an assistant on its product pages that answers from live catalogue records and published policies, serves international shoppers in their own language, and turns the high-intent conversations — bulk, wholesale, custom — into scored leads instead of closed tabs.
The question is small. The cart it costs is not
Will it fit through the door. Is it available in walnut. When does it actually arrive. Does the 120 come with the same bracket as the 90.
None of these are hard questions. All of them are blocking ones — the shopper will not complete the purchase without an answer, and the answer is not on the product page, because no product page can anticipate every version of every question.
So the shopper does one of two things. They hunt through a shipping policy and a sizing guide and a returns page, or they close the tab. Traffic to a store like this peaks on weekend evenings. Support hours do not. A contact form promising a reply within one business day is, from the shopper's point of view, slower than a competitor's product page.
And the best enquiries look exactly like the worst ones
The store has a growing trade side — office fit-outs, hospitality, interior designers buying six of something. Those enquiries are worth many times an ordinary order.
In the analytics they are indistinguishable from browsing. Somebody asking about bulk pricing and somebody idly scrolling generate the same events, and both end up, at best, in the same generic contact form. The highest-value conversations in the business are being handled by the lowest-value piece of infrastructure on the website.
The highest-value conversations in the business are handled by the lowest-value piece of infrastructure on the website.
Setup
The catalogue goes in first: price, SKU, variant and stock level per product, each marked In stock, Low stock or Out of stock. Because the assistant answers from those records, stock accuracy in the assistant is the same problem as stock accuracy in the store — one place to maintain, not two.
Then the policies the store already publishes: shipping, returns, sizing guides, care instructions. Existing pages and PDFs, no rewriting.
One line of code. On Shopify the install assistant handles the placement and verifies the widget is live.
What a person does
- Let it learn the storeIt crawls your site and takes your catalogue — price, SKU, variant and stock level per product.
- Add the policies shoppers ask aboutShipping, returns, sizing guides, care instructions. The pages and PDFs you already publish.
- Paste one line of codeOr let the install assistant handle Shopify, WordPress, Wix or Webflow and verify it is live.
What the system does from then on, unattended
- Answers price, SKU, variant and stock from your own product records, and says In stock, Low stock or Out of stock.
- Answers shipping, returns and sizing from your published policies rather than from guesswork.
- Serves international shoppers in 30+ languages, right-to-left included.
- Turns a high-intent conversation — bulk, wholesale, custom, pre-order — into a scored lead in the pipeline rather than a lost session.
- Ties the conversation to the campaign that brought the visitor, with cost per lead and ROAS per campaign.
Half past nine on a Saturday night
Watch what happens at the third exchange, when an ordinary stock question turns into something else entirely.
- Visitor: Do you have the walnut shelving in the 120cm width?
- Assistant: The 120cm walnut is low stock — $340, SKU SHV-120-WAL. The 90cm and 150cm are both in stock if the width is flexible.
- Visitor: I need six of the 120cm for an office fit-out.
- Assistant: Six is beyond what is on the shelf, so that is a wholesale enquiry — I will pass it to the team with the SKU and quantity. What is the best email for the quote?
- Six units
- Specific SKU
- B2B fit-out
- Email captured
A session that would have ended at the stock counter became a wholesale lead.
The session that became a lead
On the old site, that conversation ends at the stock counter. The 120 is low stock, the shopper wants six, the page cannot help, and the tab closes. Nobody in the business ever learns that a six-unit office fit-out enquiry came and went on a Saturday night.
Instead, the assistant recognised the shape of the request, said the honest thing about stock, and routed it as a wholesale enquiry with the SKU, the quantity and a captured email — scored Hot, in the pipeline, with the transcript attached and the originating campaign tagged.
The retail answer and the trade capture came out of the same conversation, which is the part worth noticing. Nobody had to build a separate wholesale funnel. The assistant simply stopped treating a high-intent question as a support ticket.
The shape of the change
Behaviour, checkable in the app.
- A sizing question at 9:30pm closes the tab.It is answered from live product data, and the cart survives.
- Bulk enquiries leak into a generic contact form.They arrive tagged Hot, with SKU and quantity, in the pipeline.
- International shoppers get English or nothing.30+ languages, right-to-left included.
- Ad spend is judged on platform-reported conversions.Cost per lead and ROAS against revenue you recorded as won.
What to watch once it is running
The last line is the sleeper. The questions shoppers ask most often are the fastest, cheapest signal you will ever get about what your product pages are failing to say.
- Average price, price range, variant counts and in-stock rate, generated from your catalogue.
- How many conversations become scored leads rather than closed tabs.
- Cost per lead and ROAS per campaign, with influenced attribution across the path.
- Which questions come up most — the fastest signal about what your product pages are missing.
- Line of code to install
- 1Line of code to install
- Guided setup, start to live
- ~5 minGuided setup, start to live
- Languages, right-to-left included
- 30+Languages, right-to-left included
- Answering, including 2am
- 24/7Answering, including 2am
Product facts, checkable in the app. Not outcome claims.
Questions this industry always asks
- Will it say something is in stock when it is not?
- It answers from the product records you maintain in ConGreeto, including the Low stock and Out of stock statuses. Keep those current and the assistant is current.
- Does it replace support?
- It handles the pre-purchase questions that decide a cart. Anything it has not been taught becomes a scored lead with a transcript, so a person picks it up with the context already written.
Keep reading
- Your industryNot on the list? Then name the record and the rest followsConGreeto was not built for five industries. It is a model you define — name your record, name your fields, and the assistant, pipeline and dashboards build themselves around it.
- Real estateReal estate: the 11pm property enquiry, answered at 11pmBuyers browse listings at night and message three agencies at once. The one that answers first gets the viewing.
- AutomotiveAutomotive: the test drive booked while the showroom was shutBuyers shortlist cars at midnight. The dealership that answers the odometer question is the one they turn up to.
Your visitors are already asking these questions.
One line of code, about five minutes of setup, and something is there to answer them.