How it works
What it reads, how often, and where it stops
The mechanics behind every alert: how setup works, which sources Orlea reads, how often on each plan, and the limits worth knowing before you rely on it.
14-day free trial · No card needed · Enterprise? Request a demo
Step by step
From their website to your team.
Setup
- Your website
- You enter your company's website. Orlea reads it and suggests the companies your buyers compare you with, each with its reasons.
- Your competitors
- Pick from the suggestions or add any company by its website. Your plan sets how many you can track.
- Pages chosen for you
- For each competitor we pick the pages worth watching (pricing, product, offers and terms) and find its feeds, job board, apps and filings. You can add, remove or replace any of them.
- The last 90 days
- Copies of those pages from about 90 days ago, kept by the Internet Archive, are compared with today, so the feed starts with a recap. Changes found this way are dated as a window, not a day.
From change to alert
- Compared with the last read
- Each page's readable text is compared with its previous version. Rotating banners, dates, cookie notices and site-wide navigation are filtered out first.
- Confirmed by a second read
- A material change is read a second time before it counts, so A/B tests and passing glitches rarely reach you. Screenshots of the page before and after are kept as evidence when they can be taken.
- Explained and ranked
- Claude, Anthropic's AI model, writes what changed, why it matters and what to do, and ranks it high, medium or low. The exact before and after are shown apart from that interpretation.
- Delivered
- High priority by email right away, the rest in a daily or weekly digest and a weekly brief. Slack and webhooks on Pro, Team and Enterprise; HubSpot on Team and Enterprise.
What it reads
- Web pages
- Pricing, product, offers, loyalty and delivery terms, changelogs and other pages you choose.
- News and releases
- Blogs, changelogs and newsrooms through their feeds, press releases on Mynewsdesk, Cision News and the Nordic wires, and public GitHub releases.
- Press coverage
- Articles from GDELT, a global news index, and stories shared on Hacker News, matched to the company by name and website.
- Job boards
- Public boards on Greenhouse, Lever and Ashby: roles opened and closed, by team and place.
- Apps
- iPhone, iPad and Mac apps on the US App Store: versions, release notes, price and rating.
- Official filings
- SEC EDGAR for US-listed companies and Companies House for UK-registered ones, including director changes.
- Product prices
- For shops on Pro, Team and Enterprise: competitors' product pages, found in their sitemaps and read once a day from the structured data they publish for search engines, matched to your own product feed.
Frequency
How often each source is read.
Pages adapt: a page that changes is read at your plan's fastest rate, a quiet one backs off to once a day. A page is read once however many workspaces track it.
| Starter | Pro | Team | Enterprise | |
|---|---|---|---|---|
| Competitors | 5 | 15 | 40 | 200 |
| Pages per competitor | 4 | 8 | 12 | 25 |
| Pages that change | Every 12 hours | Every 3 hours | Hourly | Hourly |
| Quiet pages | Daily | Daily | Daily | Daily |
| Press coverage | Every 12 hours | Every 3 hours | Hourly | Hourly |
| Feeds, job boards, apps, filings | Daily | Daily | Daily | Daily |
The 14-day free trial reads changing pages every 3 hours. Everything each plan includes is on the pricing page.
Limits
What it can't see.
Better to know these now than to find them in a meeting. When something can't be read, Orlea says so.
- Nothing behind a login or a paywall, and no social media.
- A page that a site's robots.txt closes to us is not read. It shows as blocked, never as unchanged.
- Some sites turn every automated visitor away. We then read the page through a fetch service, still only where robots.txt allows; if that fails too, the page shows as unreadable.
- The 90-day recap needs the Internet Archive to have kept a copy of the page. Pages it didn't keep start with no history.
- Filings come from the US and UK registers only, apps from Apple's US App Store only, and job boards from Greenhouse, Lever and Ashby only.
- Press coverage comes from an open news index and can miss articles, especially from smaller outlets.
- Product prices are read only from structured data in the page: a price drawn by JavaScript alone can't be read, and prices in another currency, from-prices and prices excluding VAT are listed but not compared.
- AI write-ups can be wrong. The facts are shown apart from the interpretation, uncertain intent is called likely, and any member of your team can mark a write-up inaccurate.
How our crawler behaves, and how a site can opt out: OrleaBot. How we protect your workspace: Security.
Be the first to know. Every time.
Pick your first competitors in a few clicks and see their last 90 days within minutes.