How we collect our competitor intelligence data
Every statistic on this site comes from monitoring data our product actually collects. Here is exactly how, what it covers, and where it falls short.
Last updated: June 2026. Data last verified 2026-07-09.
What the data is
Every number cited on this site comes from events and product records generated by Beaconmon actually monitoring competitor storefronts that our customers have added to their accounts. It is not a purchased dataset and not survey data. It is the byproduct of the product running, day after day, on the URLs our customers pointed it at.
Because it comes from live monitoring rather than a one-time crawl, the dataset grows continuously and reflects what is actually changing on competitor sites right now, not a static snapshot from months ago.
How it's collected
Each tracked competitor is assigned one of three data tiers, depending on what the competitor's site exposes and what the customer configured. Extraction uses CSS-selector targeting for page content and structured parsing for product feeds.
All collection is HTML and feed based. Beaconmon does not run a JavaScript rendering engine in its monitoring fleet, so we monitor rendered HTML and feeds, not JavaScript-heavy pages. This keeps checks fast and cheap to run at scale, at the cost of missing storefronts that render pricing entirely client-side.
Full catalog
The competitor exposes a product feed. Beaconmon ingests the whole catalog on a schedule, so every SKU is tracked for price, stock, and listing changes without you naming individual URLs.
Selected products
You point Beaconmon at specific product page URLs, up to 50 per competitor. Useful when a competitor does not expose a usable feed but you know which SKUs matter.
Limited
A lighter-touch check on a page, typically a homepage or category page, watching for changes in the HTML rather than tracking individual products. The default tier when neither of the above applies.
Current scale
As of 2026-07-09, Beaconmon tracks 641 competitors (639 distinct domains) across 4 teams, with 504,500 competitor events recorded and 180,065 competitor products tracked. The data spans roughly a 6-week window, with the bulk of volume concentrated in the most recent 30 days.
This is a snapshot, not a fixed dataset. It grows every day the product runs, and the numbers cited in individual posts are pulled fresh at the time each post is written. Where a later pull changes a finding, we say so rather than leaving the old number standing.
What the data does and doesn't represent
The competitors in this dataset are the ones our customers chose to track, not a random sample of Shopify stores. Read every statistic on this site with that in mind.
Limitations
- The dataset reflects the competitors Beaconmon customers have chosen to track, not a random sample of Shopify stores. It skews toward mid-market DTC ecommerce brands across apparel, skincare and beauty, coffee, and other verticals where our customers compete.
- We monitor rendered HTML and product feeds, not JavaScript-heavy pages. A competitor that renders pricing entirely client-side may be undercounted or missed.
- The window is roughly 6 weeks, with the bulk of volume concentrated in the most recent 30 days. Seasonal and longer-term patterns are not yet observable in this dataset.
- Significance and change-type classification are produced by an AI model. Confidence scores are stored per event, and low-confidence detections are flagged rather than presented as certain.
How AI classification works in the data
Significance scoring and event classification, for example labeling a site change as a copy update versus a layout change, are produced by an AI classifier, a Growth plan and above feature. Each classification is stored with a confidence score rather than treated as ground truth.
Low-confidence detections are flagged as such rather than silently folded into the same bucket as high-confidence ones. When a post on this site cites a classification breakdown, for instance the significance tiers or the site-change category split, it reflects the classifier's output at the time of the pull, including its uncertainty.
See the glossary for definitions of terms like data tier and significance scoring used throughout this page and the rest of the site.
Update cadence
Findings get revisited as the dataset grows. An earlier version of our noise-rate analysis, for example, put the share of sub-1% price moves at 87%. A later pull on a larger sample revised that to 90.76%, and the post was corrected rather than left standing with the old figure.
If a number on this site looks out of date, it is a candidate for a re-pull, not a permanent claim. Check /research for the current index of data-backed findings and when each was last verified.
Try Beaconmon free.
10 monitors free forever. 14-day Growth trial, no card required.
Running a Shopify store? See the Shopify-specific setup →