How do email analytics platforms classify email client usage, handle web proxies, and differentiate mobile from desktop views?
Matthew Whittaker
Co-founder & CTO, Suped
Published 20 Jun 2025
Updated 19 Aug 2025
8 min read
Understanding how recipients engage with your emails is crucial for effective email marketing. Email analytics platforms strive to provide a clear picture of this engagement, but the landscape is complex. From identifying specific email clients to differentiating between mobile and desktop views, challenges arise, particularly with the widespread use of web proxies and privacy features.
The primary method for collecting this data involves tracking pixels embedded in emails. When an email is opened, these pixels request an image from a server, transmitting data about the client environment. This data, often in the form of a user agent string, is what analytics platforms attempt to interpret. However, the accuracy of this interpretation can be significantly impacted by how email service providers (ESPs) and mailbox providers (MBPs) handle these requests.
It's a constant effort to refine these classification methods, especially as new privacy measures and caching behaviors emerge. My goal is always to get the most precise understanding possible of audience interaction, even amidst these technical hurdles. This insight helps to tailor campaigns, optimize design, and ultimately boost deliverability.
User agent strings and their limitations
At the core of email client classification lies the user agent string. This is a small piece of text sent by the client (e.g., a web browser, an email application) with every request to a server. It typically contains information about the application, operating system, and sometimes the device type. Email analytics platforms parse these strings to identify whether an email was opened in Gmail, Outlook, Apple Mail, or another client. Here is a basic example of what a user agent string might look like:
Example user agent stringplain
Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36
While parsing these strings gives us a baseline, the data is not always perfect. User agent strings can be inconsistent, deliberately generic, or even absent due to various factors. For instance, some clients or security software might strip down or alter the user agent string, making precise identification difficult. This means that an open attributed to a generic Mozilla user agent might actually be from an iPhone using Apple Mail, rather than a desktop browser.
My team and I have observed that different versions of the same email client can also present unique user agent patterns, leading to further classification challenges. For example, distinguishing between Outlook 2013 and Outlook 2013 120 DPI often relies on subtle or non-existent differences in their reported user agent strings, making granular distinction nearly impossible from the pixel alone. This can impact the accuracy of client market share data, as we discuss in our article How accurate is email client market share data.
Understanding web proxies and image caching
A significant factor complicating email client classification is the use of web proxies by major mailbox providers. Providers like Gmail, Yahoo Mail (including AOL and Verizon Media Group), and more recently Microsoft Outlook (with MicrosoftDefender for Office 365) employ image caching proxies. These proxies pre-fetch images when an email arrives in the inbox, regardless of whether a human recipient has actually opened the email.
This pre-fetching results in what we call artificial opens or machine opens. When an email is opened by a proxy, the user agent string often reflects the proxy server itself, not the end-user's actual email client or device. This means a significant portion of your opens might be attributed to a generic "web proxy" category in your analytics, making it challenging to get granular data on true human engagement. For a deeper dive into these, check out why automated scripts and crawlers open emails.
The impact of web proxies is substantial, as they obscure the true email client and device type for a large segment of your audience. This is particularly true for Gmail and Yahoo users, where image caching is ubiquitous. While some platforms occasionally override the caching and provide a full user agent string (e.g., Edge 12 for Gmail webmail), these instances are generally anomalies. This means that if you're looking at your email client data, a large Web Proxy category often includes a mix of mobile, desktop, and webmail users.
Distinguishing mobile versus desktop engagement
Differentiating between mobile and desktop views is crucial for optimizing email design and understanding audience behavior. Email marketers heavily rely on this data to ensure their responsive designs are performing as expected and to segment their audience for targeted campaigns. However, as with client classification, web proxies complicate this distinction, as they can mask the true device type.
Despite the challenges posed by proxies and privacy features, there are still ways to infer mobile versus desktop engagement. For example, some analytics providers can sometimes derive device type from specific, uncached user agent strings that might slip through the proxy, such as an Edge 12 browser on a mobile device accessing Gmail webmail. These rare instances can offer glimpses into actual device usage behind the proxy. Understanding these nuances is key to accurately measuring email trends.
For more precise insights, some advanced techniques involve employing multiple tracking pixels that load based on CSS media queries (breakpoints). By triggering different pixels for different screen sizes, marketers can infer whether an email was opened on a mobile device or a desktop, even if the initial open request came from a proxy. iOS devices, for example, have very specific breakpoints that can be leveraged for this purpose. While this method isn't foolproof, it provides a more granular view than relying solely on the primary tracking pixel's user agent.
Traditional open tracking
Relies on a single tracking pixel. The user agent string from the open request is used to identify the email client and device. This method is straightforward but highly susceptible to proxy interference and privacy features.
Advanced device inference
Employs multiple tracking pixels, each configured to load based on specific CSS media queries. For example, a pixel might load only if the screen width is less than 480px, indicating a mobile device. This provides a more accurate, albeit still imperfect, picture of device usage behind proxies.
Strategies for accurate email analytics
The introduction of privacy features like Apple Mail Privacy Protection (MPP) has further altered the landscape of email analytics. Apple now routes all images through a proxy server, regardless of the user's actual open action. This means that email open rates, which were already skewed by other proxies, are now even less reliable as a direct measure of recipient engagement. It's becoming increasingly difficult to distinguish human opens from proxy opens.
Given these complexities, I advocate for shifting focus beyond simple open rates. While opens still provide some indication of inbox placement, they no longer reliably convey engagement. Instead, I focus on clicks and conversions as more robust indicators of true recipient interaction. These actions typically occur after the proxy has pre-fetched the email, meaning they reflect genuine interest. Monitoring how internet service providers track engagement also provides valuable context.
To effectively navigate these challenges, I recommend leveraging analytics platforms that not only provide aggregated data but also allow access to raw user agent strings or offer advanced parsing capabilities. This granular data, when combined with a sophisticated understanding of proxy behavior and privacy features, can help uncover more meaningful insights into your audience. It's a continuous process of adaptation to ensure email deliverability remains strong, as highlighted in our latest email deliverability report.
Views from the trenches
Best practices
Actively monitor user agent strings, especially for anomalies that might indicate un-proxied opens or specific client versions.
Utilize CSS media queries with multiple tracking pixels to infer device types (mobile vs. desktop) more accurately, particularly for iOS users.
Prioritize clicks and conversions over open rates as primary engagement metrics, given the impact of image caching proxies and privacy features.
Segment your audience based on inferred device usage to tailor content and design, enhancing the user experience.
Regularly review industry reports and updates on email client behavior and privacy changes to adapt your tracking strategies.
Common pitfalls
Over-relying on basic open rate data as an accurate measure of engagement, without accounting for proxy pre-fetching.
Misinterpreting 'Web Proxy' data as a distinct client type, rather than a mix of various clients and devices.
Failing to adapt email design and content strategy based on the true mobile-vs-desktop usage, leading to poor user experience.
Ignoring the impact of privacy features like Apple Mail Privacy Protection on the accuracy of your email analytics.
Not leveraging subscriber-level data from your analytics platform to gain deeper insights into individual client usage.
Expert tips
Explore advanced methods like loading different pixels at specific breakpoints via CSS to differentiate device types, especially for iOS.
Access raw user agent string data from your analytics platform to manually parse and uncover more granular client details.
Understand that even with advanced techniques, a perfectly clear picture of client usage behind proxies may not always be possible.
Consider a shift in focus towards post-open metrics such as click-through rates and conversion rates as more reliable indicators of engagement.
Stay informed about updates from major mailbox providers regarding their image caching and privacy practices.
Marketer view
Marketer from Email Geeks says they are trying to figure out the exact percentage of customers viewing emails on mobile, but the 'Web Proxy' category seems to include a mixed portion of these.
2019-07-02 - Email Geeks
Expert view
Expert from Email Geeks says that web proxy data includes subscribers from mobile, webmail, and other platforms mixed together. Gmail and Yahoo cache images for all their users, so opens attributed to them outside of the web proxy category are those that have not been cached.
2019-07-02 - Email Geeks
Refining your analytics approach
Navigating the complexities of email client classification, web proxies, and mobile versus desktop differentiation requires a nuanced approach. While user agent strings provide a starting point, the pervasive use of image caching by major mailbox providers and new privacy features means that simple open rates are no longer the most reliable metric for understanding true engagement or client usage.
Gmail, 
Yahoo Mail (including 
AOL and 
Verizon Media Group), and more recently 
Microsoft Outlook (with 
MicrosoftDefender for Office 365) employ image caching proxies. These proxies pre-fetch images when an email arrives in the inbox, regardless of whether a human recipient has actually opened the email.
Apple now routes all images through a proxy server, regardless of the user's actual open action. This means that email open rates, which were already skewed by other proxies, are now even less reliable as a direct measure of recipient engagement. It's becoming increasingly difficult to distinguish human opens from proxy opens.
