Compare your site with competitors

What gets compared

The rows depend on your kind of site.

For a shop: free shipping and its minimum amount, shipping cost, delivery time, an order-by time or delivery date, pickup, payment methods and pay later, express checkout, return period, free returns, warranty, ways to get help, opening hours, review score, product reviews, and what product listings and product pages show.

Shops also get rows for the cart page: selling points in the cart, a free-shipping progress line, the shipping cost and delivery time shown in the cart, express checkout buttons, payment icons, a review score or trust mark, help or contact, product suggestions and savings shown.

For a lead generation site (services, B2B, SaaS): whether the headline says what you offer and for whom, the main button in the first screen, prices, a guarantee or free trial, booking a call online, client logos and testimonials, case studies, your team and credentials, and how quickly people can reach you.

Varly picks the kind of site the same way as in Page review: orders, a shop platform or a cart mean a shop. You can change it next to Compare under This site is a.

Suggestions from Oran

Ask Oran for suggestions works in two steps. Oran first describes your site from your homepage and your busiest pages: what you offer, to whom, how you sell and in which country and language. Above the suggestions you see who Oran looked for, for example "Oran looked for Dutch webshops selling bathroom furniture". Then Oran names up to five direct online competitors it knows: sites that offer the same thing to the same customers in the same country and language. If Oran knows fewer, you get fewer; it doesn't make any up. Marketplaces and big general stores are left out, and so are suppliers, tools and partners.

Varly then checks that every suggestion is a real, working site; domains that are for sale or don't answer are left out. A site that only blocks bots still counts, because it exists. Nothing else is visited until you add a suggestion. This takes about 10 to 20 seconds.

They do better, about the same, you do better

Next to every competitor value you see how it compares with your site, always in words:

• They do better (red): the competitor does this better. This is a chance for you and where to look first.
• About the same
• You do better (green): you are ahead here.
• Not found: Varly did not find it on the pages it opened.
• Not checked: Varly did not open the cart on this site (see below), so the cart rows have no value and no verdict.

On a phone you compare your site with one competitor at a time. Pick the competitor above the table; the verdict sits next to the name of each row.

Varly does these comparisons itself, with fixed rules: a lower free-shipping amount, lower shipping cost or faster delivery is better, a longer return period is better, more payment methods or contact options is better. A higher review score only counts as better with at least 100 reviews.

Detected on the page or reviewed by Oran

A small mark before every value shows where it comes from. The legend above the table explains both, and you can point at a mark to see its name:

• Orange dot, Detected on the page: Varly's fixed checks found it in the text, links or structured data of the page.
• Oran, Reviewed by Oran: Oran, Varly's AI analyst, pointed to the text on the page that shows it. Varly then reads numbers such as amounts and days from that text itself, so the comparison never depends on what the AI says. Where Oran judges something himself, for example how clear a headline is, the cell says Reviewed by Oran in full.

Click a cell to see the exact text, a screenshot and where it came from. A cell without proof always says Not found.

Your own selling points

Below the table, Varly checks how your own selling points come across in five parts. Each selling point gets one row with five marks, one per part (on a phone just the first letter). A check means strong, a dash okay, a cross weak and an empty circle not measured. Open a row to read why.

• Clarity: is it stated plainly? (Oran)
• Unique: do competitors offer the same or better? (from the table)
• Relevance: do your visitors care? (from your survey answers, once you have at least 20)
• Visibility: do visitors see it where they decide, for example in the first screen or near the buy button? (from your pages and your scroll data)
• Enticing: is the offer attractive? (Oran)

Ideas from competitors

Oran also points out things on competitor sites you could test, such as a delivery date in the header or a guarantee next to the main button. Each one has a screenshot and a Create idea button.

A comparison is not a measurement. Ideas from it show as a Signal. They only count as Confirmed when your own data agrees, for example survey answers about the same topic.

How Varly visits other sites

Varly identifies itself honestly as VarlyReview and follows each site's robots.txt and opt-outs from text and data mining. If a site asks bots not to visit, or blocks the visit, it is skipped and the table says so. Varly does not try again in another way.

Per site Varly opens at most 12 public pages, one page per second, on desktop and mobile. It never logs in, types, searches, fills in forms or opens the checkout. Read more on the VarlyReview bot page.

Varly visits each competitor at most once a day. If it read a competitor's public pages for any comparison in the last 24 hours, it uses those pages instead of visiting again, and the table shows the date they were read. A site that asked bots to stay away is not tried again within that time either. Your own site is always read fresh.

How often you can compare

You can compare each site once every 24 hours. Until then, the Compare again button shows when you can start the next comparison, in your own time zone. An organization can start up to 10 comparisons in 24 hours across all its sites. A comparison that stopped because of a problem on Varly's side does not count.

The cart page

On a shop, Varly adds one product to a cart to see the cart page, on your own site and on each competitor:

1. It opens the product page it already visited, in a new browser session.
2. It picks the first available option (for example a size) and leaves the quantity at one. It never types anything.
3. It clicks the add-to-cart button, then opens the cart and takes a screenshot on desktop and mobile.
4. It throws the session away, so nothing stays in a cart.

It never goes to checkout: requests to checkout or payment pages are stopped, also when a site tries to redirect there. While it is in the cart, it blocks known analytics and ad tracking requests, so its visit does not count as a shopper in anyone's add-to-cart numbers. The cart counts toward the 12 pages.

Varly skips the cart when robots.txt asks bots not to open it, or when the add-to-cart button sends to a path robots.txt does not allow. Shopify, for example, does this by default. Above the table you then see Cart skipped with the reason, and the cart rows say Not checked. Varly also skips the cart when there is no add-to-cart button it can use, when adding needs something typed (such as engraving text), when the cart sits inside the checkout, or when the site asks visitors to log in first.

Privacy

Your comparisons are private to your site. Nobody outside your team sees your table, your selling points, your ideas or anything from your own site. The only thing Varly reuses between customers is what it read on a competitor's public pages in the last 24 hours, so it does not visit the same site twice.

Varly keeps your last two comparisons. It keeps the screenshots of the findings and deletes the full page screenshots once a comparison is done. Comparisons older than 180 days are deleted too, except your most recent one.

If you plan to publish a comparison, for example in an ad, rules for comparative advertising apply. Check with a legal advisor first.

What it costs

A comparison needs your own AI key, with a model that can read images. Without one, Varly does not start a comparison: Oran fills in most of the table, judges your selling points and finds the ideas, so the table would be mostly empty. Add a key under Organization settings → AI key. Oran's part costs about $1 for a comparison with four competitors on Claude Opus 5.5, and about half that on Claude Sonnet 5.5. Varly shows the estimate before you start and the actual cost after the comparison, and it also appears under Organization settings → AI usage.

A comparison takes 10 to 20 minutes, because Varly visits each site slowly. Comparing once a quarter is usually enough.