Skip to content

Reliability

Attribution data is only useful when it can be trusted for business decisions.

This page explains how accuracy is tested, how the script is kept available, and how browser-side processing supports compliance.

Why reliability matters for attribution decisions

Attribution figures influence budget and channel decisions, so the checks behind them matter as much as the figures themselves. The approach described here separates three concerns: whether data is accurate across real-world setups, whether the collection script is reachable, and whether personal data handling creates compliance risk.

Accurate data: tested across devices, systems and browsers

Accuracy is checked across multiple devices, operating systems and browsers rather than on a single reference setup. Device testing covers Apple, Samsung, Google Pixel, OnePlus, Xiaomi and Motorola hardware. Operating-system testing covers Mac OS X, Windows, iOS and Android. Browser testing covers Chrome, Firefox, Safari, Edge and Opera. Use this as a pre-purchase checklist: ask whether a tool's test matrix covers the devices and browsers your own visitors actually use, and whether results are compared across that matrix instead of reported from one environment.

Always available through a distributed network

The script is served through Amazon CloudFront, with hundreds of copies held at data centres around the world. The practical distinction is between a single hosted file, which is a single point of failure, and a distributed network, where a visitor can be served from a nearby copy. When evaluating availability, check for a content delivery network, geographic distribution of copies, and what happens if one location is unavailable.

Fully compliant by design: browser-side processing

Attribution processing can be designed to run inside the visitor's browser, so that no visitor-level data is transmitted to the tool's servers. That architecture is often presented as a way to reduce GDPR and CCPA exposure, because the data flow itself is narrower. It is still worth checking for any attribution tool: identify where processing takes place, what information leaves the browser, and what is retained server-side. Whether such a design is sufficient for a particular organisation depends on its own legal review and data-processing arrangements, so the compliance position should be confirmed against the applicable regime before relying on it.

Battle-tested scale and getting started

Scale is one signal of maturity, but it should be weighed alongside accuracy testing and availability. A useful evaluation question is how many sites a tool runs on, how much visitor traffic it handles each month, and whether those figures can be independently substantiated. Commercial terms deserve the same scrutiny: check whether there is a free entry point, how long any trial lasts, and what the paid tiers cost per month, then compare those terms with your own budget and contract requirements. Confirm current pricing and trial conditions directly with the provider before committing.

Reliability checklist for evaluating an attribution tool
Check areaWhat is coveredWhat to verify
DevicesApple, Samsung, Google Pixel, OnePlus, Xiaomi, MotorolaCoverage of your visitors' actual devices
Operating systemsMac OS X, Windows, iOS, AndroidCoverage of your visitors' actual systems
BrowsersChrome, Firefox, Safari, Edge, OperaCoverage of your visitors' actual browsers
AvailabilityCloudFront network, hundreds of copies worldwideDistributed delivery and fallback behaviour
ComplianceBrowser-side processing; no user data to serversData-flow and legal position for your setup

Test matrix as described in captured evidence; confirm current coverage with owner before publication.

Frequently asked questions

How is attribution accuracy checked across different setups?

Accuracy checks cover several device brands, four operating systems and five browsers, so a result seen on one setup can be compared with others before it is relied on for decisions.

How is the script kept available?

The script is distributed through a global content delivery network with copies held in data centres in many locations, which is intended to keep it reachable when a visitor loads a page.

Does use affect GDPR or CCPA compliance?

Processing happens in the visitor's browser and user data is not sent to the tool's servers, which is presented as avoiding an impact on GDPR or CCPA compliance from this data flow.