Suppli Web¶

Connect your website so Suppli reads your business data from it. You verify the site once; from then on Suppli can read your policy and information pages into Business Information (which the Website Chatbot and Suppli DM answer from), find your products, watch the site, and count its visitors. It lives in the sidebar between CRM and Ads.

Install it¶

Open the Marketplace, find Suppli Web and press Enable. Then open Suppli Web and press Connect site.

Connect and verify by DNS¶

Enter your website address (for example yourbrand.com).
Suppli shows a TXT record: type TXT, host your domain, value suppli-site-verification=…. Copy each field into your domain's DNS provider.
Press Verify. DNS changes can take up to an hour to appear; Suppli tells you whether the record is missing, has the wrong value, or timed out.

Can't edit DNS? Put the meta tag from Settings inside the <head> of your homepage instead and press Verify. Some stores answer automated visitors with a block on the homepage (Shopify does); the DNS record always works.

Nothing is read, watched or counted until the site is verified. Suppli checks the record again every week; if it disappears, everything pauses and you get an Inbox alert. Your site needs https; Suppli fetches nothing over plain http.

What each fetch does¶

Press Fetch on the Suppli Web page:

Fetch policy and business pages — Suppli finds your pages through your sitemap (or by following links from the homepage), keeps the shipping, returns and refunds, warranty, privacy, terms, payment, FAQ, about and contact pages, and writes one Business Information entry per page: a short paragraph of at most about 120 words, written only from what the page says, with the page's address and the date it was read. Product, collection, cart, account, search and blog pages are skipped. Up to 100 pages per fetch.
Fetch products — every product and variant on the site. The first time, each one lands in Unmapped SKUs with just its SKU and name, under a supplier named after your site. Map the ones you sell; from the next fetch on, Suppli fills in the title, description, barcode, brand, website price, photo (copied to Suppli, with the original address kept), weight and dimensions where published, options, categories and tags, and keeps every other field it finds on the product's record. Up to 2,000 products from a store's product listing, or 300 product pages when Suppli has to read pages one by one.
Fetch one page — a single page on your site, a product page or a policy page, nothing else.

Suppli reads structured data first and uses AI last: a Shopify store's product listing, a WooCommerce store's product API, or the product data a page carries for search engines. Only a page with none of those is read by a model. Each product records which method found it.

Your website is never a stock source and never a cost source. Nothing Suppli reads from it changes inventory, the missing-SKU rule, supplier cost or the Repricer. A product missing from a fetch is not out of stock. The website price is stored as information only.

A site whose pages need JavaScript to render, or that blocks automated visitors, is reported as such on each page rather than skipped silently.

How the chatbot uses the entries¶

The Website Chatbot and Suppli DM answer policy questions from the stored From website entries, and may link the page they came from. They never fetch your site while replying. If you change your site a lot, run the pages fetch again; Suppli also re-reads your pages once a week and sends one Inbox alert listing the pages that changed, with a link to refetch. It never rewrites an entry on its own.

The manual-edit rule¶

Open Business Information. Entries from your site are labelled From website. Edit one and Suppli keeps your wording: a later fetch that finds new text shows it under the entry as a pending update you accept or keep your own. The same rule applies to product fields you edited by hand in the Catalog.

Turning on tracking¶

Suppli Web has its own tracking script, separate from the chatbot embed. There are three ways to turn it on, and Suppli Web → Settings shows the one that fits your site first:

Your Shopify store is connected as a sales channel: press Enable tracking. Suppli puts a Shopify web pixel on the store through your connection: page views, product views, add to cart and checkout, inside Shopify's consent-aware sandbox, with no theme edit. Disable tracking removes it. If the store was connected before tracking existed, Suppli asks you to reconnect with the tracking permission once. The status shown is the pixel's real state on Shopify.
WordPress or WooCommerce: press Download the plugin. The zip is pre-configured for your site; upload it once in WordPress (Plugins → Add New → Upload Plugin) and activate it. It prints the script on every front-end page and sends each WooCommerce product page's SKU as a product view. Deactivate it to stop.
Any other site: paste the script once, in your site's global header or footer; it then runs on every page. The settings card shows the exact place for WordPress (if you would rather not use the plugin), Shopify (unconnected), Wix, Squarespace, Webflow, Google Tag Manager and other sites.

The script sends page views (path, referrer, UTM parameters, device type), the Core Web Vitals your visitors actually experience (LCP, INP, CLS), and product views. A page counts as a product view when its address matches a product Suppli has found on your site, or when the page carries standard product data (schema.org); Shopify and the WordPress plugin send it natively.

Cookieless. The script sets no cookie and Suppli stores no IP address. A visitor is a daily-rotating hash that cannot be tied to a person after the day, so in most cases no consent banner is needed. You remain responsible for your own privacy notice.

Why not through DNS? DNS can prove you own a domain, but it cannot put anything on a page: the only DNS-based route would send all of your store's traffic through Suppli, which would make Suppli an outage point for your whole store. Suppli does not do that.

Metrics¶

The top strip shows LCP P75, INP P75 and Cumulative Layout Shift with their change against the previous period and a Good / Needs improvement / Poor badge, then visitors, page views and uptime, for 7, 30 or 90 days. Until the script is installed, the speed numbers and the picture of your homepage come from Google PageSpeed Insights, refreshed daily; once the script is sending, they are your visitors' own 75th percentiles.

Below the pages you'll find the visitor dashboard: sessions (a visitor's run of activity with no gap over 30 minutes), visitors, page views and uptime, each with a sparkline and its change, plus live visitors (the last 5 minutes), top pages, top referrers, UTM campaigns and product views. Every number is counted by the database. Sessions and live visitors also join the strip on Home once the script has sent data.

Suppli checks your homepage every 5 minutes and your SSL certificate daily. You get one Inbox alert when the site is down (two failed checks in a row), one when it is back, and one when the certificate expires within 14 days. Alerts arrive in the Suppli Web section of Suppli Inbox, with the usual Slack and email routing.

Costs¶

Suppli Web is a Pro app. On the free plan you get 1 business-page fetch and 1 product fetch a month; a single-page fetch is always allowed. Monitoring, verification, structured product fetches and visitor tracking cost nothing. AI work (page summaries, and product extraction where a page carries no structured data) is charged at its real measured cost from your Suppli AI credit, then your own key, and every fetch's cost is listed on the Suppli Web page (Recent fetches) and on your billing page. What was taken from your credit also appears as the Suppli Web row of This month's usage on API Keys & Credit, exactly, to the millionth of a dollar: a short page read by gpt-4o-mini costs well under a cent (about $0.00006 for a 300-token page), and the bill shows that figure, never $0.00.

Turning it off¶

Turn Suppli Web off from the Marketplace and monitoring, fetching and tracking stop; the script's data is refused. Your Business Information entries stay, still labelled From website. Turn it back on and everything resumes from the stored site; you only verify again if the DNS record is gone.