Revenue
Sizing, shipping and returns decide the basket, and none of them are on the product page.
Ecommerce objections are unusually predictable. Three topics account for most hesitation, all three are documented somewhere on the site, and all three are one click away from the moment they matter.
- Private launchOnboarding selected teams now
- Built by CreoglyphSeven client engagements behind the product
One click away is one click too many
A shopper on a product page wonders whether the size runs small. The answer is in a sizing guide, linked in the footer. Getting there means leaving the product page, which means losing the basket state and the momentum that came with it.
The same applies to delivery timing and to returns. Each has a dedicated page. Each page is well written. Each is in the wrong place, because the decision happens on the product page and the answer lives elsewhere.
The measurable effect is an exit from a page with high purchase intent, which reads in analytics as a product page problem. It is not. It is a proximity problem.
What changes
Today
- Shopper hesitates on sizing at the product page
- Opens the sizing guide in a new tab, or does not
- Loses the thread, compares elsewhere
- Basket abandoned, no record of why
- The sizing guide is judged useful because it gets traffic
With an assistant
- Shopper asks about sizing without leaving the product page
- Assistant answers from the sizing guide and cites it
- Delivery and returns answered in the same conversation
- Objections resolved at the point of decision
- The topics that keep coming up are logged for the product page
Answering an objection without leaving the page
Index the policy pages
Sizing, delivery, returns, care instructions, stock policy. Small and high value. Product descriptions are usually a worse source than the policy pages, because they are written to sell rather than to answer.
Answer in place
The widget opens over the product page. The basket persists, the context persists, and the answer arrives with the product still on screen.
Cite, because trust is the objection
On a returns question the shopper is assessing risk. An answer that names the policy page is a different object from an answer that asserts.
Refuse on stock and order status
Anything requiring a live lookup is outside what a content trained assistant can do. Guessing at stock is worse than saying you cannot check.
Where the widget should not appear
This is the ecommerce specific decision, and getting it wrong costs more than the assistant earns.
Checkout: off
Checkout has one intended action, the launcher competes with it for the same thumb on mobile, and a shopper who opens chat during payment is a shopper who has stopped paying.
Basket: allow, but disable at payment
Objections do surface there, but so does distraction. Allowing it on the basket and turning it off at the payment step is the defensible position.
Product and category: always on
That is where objections form and where the answer is furthest away.
This is a conversion decision, not a styling one, and it is worth measuring rather than assuming. The counterfactual for chat on checkout is not a lost answer, it is a completed purchase.
Where the data goes
The assistant answers from published content only. It cannot check live stock, look up an order, or see a customer account, and those should be escalation triggers rather than attempted answers. Security and data boundaries.
Questions
Signal
Is this your situation?
Access is by waitlist, demo request or public-content pilot. Describe the question your visitors keep asking and we will tell you honestly whether it fits.