This page explains the method in plain language: what we collect, how dollar figures are estimated from your traffic and transactions, what the Conversion Rate Curve does and does not do, how results are assessed after a fix, and why pasting a spreadsheet into an AI chat is not the same thing.
Performance, Core Web Vitals and related monitoring signals stay in the picture. They are inputs to the decision, not the whole product.
What does Blue Triangle collect?
After you add the Blue Triangle tag to your site, and the native SDK to your apps where you use them, Blue Triangle continuously measures real visitor experiences on the properties you instrument. That includes performance metrics such as page load time, experience signals such as bounces, exits and errors, and the revenue, order and conversion events you configure.
On your sites, the Blue Triangle tag does not collect names, addresses, phone numbers, credit card numbers, social security numbers, pictures or social preferences, and IP addresses are pseudonymized at collection. The details are in our Privacy Policy.
Other analytics and monitoring products collect their own data too. Blue Triangle's difference is a continuous, first-party baseline on your properties, used to price friction in dollars and to follow up after fixes.
How does your data inform the revenue estimate?
Dollar figures are grounded in your measured traffic, conversion behavior and transaction context, not a generic industry average applied to your site. Blue Triangle builds your performance and conversion history over time and prices friction against that baseline.
Sensitivity to friction differs by brand, funnel step, device and season, so opportunity estimates use the experience-to-outcome patterns on your property. The baseline gets richer as more sessions are observed.
Estimates are still estimates. The goal is a customer-specific view of revenue at risk that is clear enough to prioritize, not a guarantee of future results.
What does the Conversion Rate Curve do, and what are its limits?
The Conversion Rate Curve relates how people experience a page, such as how fast it loads for them, to how they convert, and expresses the revenue opportunity if that experience improves.
Useful for: seeing how sensitive your customers are on a page or step, turning "make this faster" into a ranked dollar opportunity, and percentage-based what-if planning.
Not: a claim that speed is the only friction that matters (Quantify also surfaces other costly issues, such as broken links and out-of-stocks), a guarantee that every estimated dollar lands exactly as forecast, or a substitute for product judgment and testing.
What is the difference between measured behavior and estimated opportunity?
Measured behavior is what happened in instrumented sessions: pages, timing, errors and conversions. It describes the past and present.
Estimated revenue opportunity is the dollars at risk, or recoverable, if friction improves, based on your baseline. It is forward-looking, it is for prioritization, and it assumes conditions stay reasonably comparable.
Blue Triangle connects the two on purpose. Measurement without dollars is hard to prioritize, and dollars without continuous measurement are easy to argue with.
How are results assessed after a fix?
After you ship a change, Validate compares the week before and the week after. The Revenue Attribution report shows the measured change in revenue, sessions, average order value and conversion rate, and estimates how much of the change came from experience influences (performance and stability) versus marketing influences (sessions, average order value and intent).
Treat it as an estimate with limits. Promotions, seasonality, traffic mix and other releases in the same weeks also move revenue, so the report supports a leadership conversation rather than showing that one fix caused one dollar figure. The Help Center article on the Revenue Attribution report explains the method.
What can AI assistants access?
Supported AI assistants, including Claude, Cursor and Gemini, can connect to the Blue Triangle MCP server and ask plain-language questions of your Blue Triangle data: performance summaries, comparisons between periods, errors and third-party resource timing, plus revenue, conversion and order data where revenue metrics are enabled for the account. Connecting requires a Blue Triangle account with API access. The MCP overview lists the supported clients and data.
AI is an interface, not the product. An assistant does not replace continuous first-party capture, your customer-specific baseline, or the Quantify, Resolve, Validate loop.
Why is a spreadsheet in an AI tool not the same?
An export plus a general AI assistant can summarize rows and suggest patterns, which can help brainstorming. It still lacks what Blue Triangle is built to provide:
- Continuous instrumentation: ongoing sessions, not a frozen file.
- Your baseline: history on your own properties that gets richer over time.
- Friction priced in dollars against that baseline: across pages and issue types.
- Find, fix, check: engineering handoff and an after-release comparison.
- Live assistant access to the same system of record: not whatever was in last week's export.
The gap is not that AI cannot read numbers. The gap is continuous measurement, customer-specific dollar estimates and a closed loop from finding to follow-up, which a prompt on a file does not recreate.
.svg)