Adtech Data for Intelligence: What It Is, the Signal Types, and How It Comes Together
Public sector teams rely on commercially available location signals to understand how people and populations move in the real world. From supporting missions to researching responses to natural disasters, no other data set matches their granularity. Much of that location information comes out of the digital advertising ecosystem, and analysts often group it under a single shorthand: "adtech data." It's a convenient label for the device and location signals that the advertising world generates as it operates at global scale.
The adtech ecosystem is layered and includes a variety of signals. It spans several distinct forms of location data that originate at different points in the advertising process, from real-time ad auctions to signals captured directly inside applications, each with its own reach, precision, and analytic value. Understanding those distinctions, and how the pieces fit together, helps teams understand how the data might fit into their workflows and current stack.
This guide breaks down what adtech data is, focuses on the two sources that matter most for intelligence work, and explains why bringing them together produces a more complete picture than either provides alone.
What adtech data actually is
Put simply, adtech data is a byproduct of digital advertising. Every time an ad is served, a chain of automated processes runs in the background to decide which ad to show and to whom, and many of those processes generate or pass along a location signal.
The mechanics are easiest to see in an everyday shopping experience. Say you spend an evening researching a new SUV, comparing makes and models across a few sites. Within minutes, ads for that manufacturer start following you around the web. Keep looking and they get more specific, until you're seeing ads for a dealership a few miles away, promoting the exact model you searched and this weekend's pricing. That last step is the part worth noticing. The manufacturer's ad is based on your interest, but the local dealer's ad is only possible because a location attribute traveled alongside the ad request. The system matched what you wanted with roughly where you were and connected you to a nearby advertiser in milliseconds. That location attribute, thrown off as a routine part of serving the ad, is the raw material of adtech data.
This happens across the full advertising ecosystem: mobile apps, mobile and desktop web, and connected TV. Multiply it across billions of ad-supported sessions a day, across an international footprint, and the advertising ecosystem becomes one of the largest sources of device location information in the world.
Not every part of that ecosystem contributes location equally, though, and the distinction matters for intelligence work. Mobile devices are where precise, GPS-grade location originates, because that's where positioning hardware and location permissions come together. Desktop and connected TV environments generally carry broader, network-derived locations. For teams that need accurate positioning, smartphone data is the most desirable, which is why the highest-value adtech location signals come from phones and tablets rather than browsers or televisions.
These are device-level signals. For the mobile data at the heart of intelligence work, a single record typically contains three core elements: a location (a set of coordinates), a timestamp marking when the signal was generated, and a device identifier known as a mobile ad ID, or MAID. The location tells you where, the timestamp tells you when, and the MAID is the field that lets otherwise-separate signals be associated with the same device.
For intelligence purposes, two sources dominate, and they sit at opposite ends of a tradeoff between reach and precision: bidstream data and SDK-derived data. The sections that follow break down how each works, what it's suited for, and how a shared identifier lets them be combined into something more useful than either alone.
The signal types intelligence teams should know
Bidstream data
Bidstream data is generated by real-time bidding, the automated auction behind the car-shopping example above. In the instant an app, website, or connected TV device has ad space to fill, it broadcasts a request, a bid request, to a marketplace of potential advertisers, and that request often includes a location. Because nearly the entire ad-supported ecosystem participates in this process, bidstream is enormous in both volume and geographic reach, spanning mobile, web, and connected TV.
That breadth is its defining strength. Bidstream can surface activity across a vast number of devices and a wide international footprint. The tradeoff is variability, and precision differs sharply from one signal to the next. Some mobile bid requests carry true GPS-grade coordinates that the device passed into the auction, while many others carry only cached, tower-derived, or IP-derived location that is far less precise; web and connected TV requests are similar, often placing the device at a neighborhood, zip code, or city level. The auction ecosystem also produces noise and signals of inconsistent quality. Bidstream is powerful for coverage, but it rewards teams that know how to separate the precise, reliable signals from the rest.
That same scale makes quality hard to police. With billions of bid requests moving through a sprawling chain of exchanges and intermediaries, invalid and fraudulent traffic is difficult to filter, and synthetic location signals, points that no real device ever produced, travel alongside the genuine ones. The incentive is financial: inventory tagged with desirable location commands higher prices, so fabricated or placeholder coordinates get inserted to inflate its value. These enter mainly at the bid-request stage, where location, app, and publisher fields can be spoofed, and they range from crude data-center bots to more sophisticated fraud built to mimic real users. For analysts, the result is that authentic, precise signals sit beside cached, imprecise, and entirely fabricated ones, and separating them is the real work.
SDK-derived data
SDK-derived data comes straight from the device. A software development kit, or SDK, is a piece of code that an app developer builds into an application. When a user installs that app and grants it location permissions, the SDK can read position directly from the phone's onboard positioning systems, the same GPS, Wi-Fi, and Bluetooth signals the device uses to place itself on a map. The location is collected at the source, as determined by the device itself, rather than inferred downstream.
That direct line to the hardware is what makes SDK data precise. When an app is actively using location services and the device has a clean GPS fix, the resulting coordinate can be accurate to within a few meters, close enough to tell one building from the one beside it. SDK signals also tend to arrive structured and well-labeled, frequently carrying an accuracy estimate for each reading, so an analyst can see not just where a device was but how confident that measurement is. Precision does vary with conditions, since dense urban cores, indoor spaces, and weak reception all degrade a GPS fix, but at its best SDK data is the most accurate location the advertising ecosystem produces.
That said, not all applications require precise location to operate. As an example, consider your favorite hardware store's mobile application. To determine which store you're at, the app likely doesn't require precise positioning, but rather a broad idea of your position to know which zip code you're in versus another. So the precision of SDK data varies, but as a general rule it has more granularity than other adtech signal types.
But the cost of that precision is reach. SDK data exists only for apps that have integrated a location SDK and only for users who have granted those apps permission to collect location. That makes its footprint narrower and more selective.
The MAID: the identifier that connects them
A mobile ad ID, or MAID, is a pseudonymous but persistent device identifier issued by the phone's operating system, Apple's IDFA on iOS and Google's Advertising ID on Android. It names the device rather than the person: it carries no name, number, or account, but it stays attached to the same device across apps and over time until the user resets or clears it. That combination, anonymous on its face yet stable across sessions, is what makes it useful for measurement and, by extension, for analysis.
The MAID is not a third source alongside bidstream and SDK. It is a field within both, the identifier each signal carries regardless of how the location was captured. Because the same MAID appears in both, signals from the two sources can be attributed to a single device. A broad, lower-precision bidstream signal and a sparse, high-precision SDK signal that share a MAID resolve to the same device, which is what lets two separate streams be assembled into one continuous record rather than read in isolation.
IP-derived location and metadata
Two supporting elements round out the picture. IP-derived location, the lower-resolution network-based signal that dominates desktop and connected TV environments, uses a device's network connection to estimate position. It's less precise than GPS-grade SDK data, but it adds useful context, particularly around connectivity. Alongside it, each signal typically carries metadata, including timestamps, signal frequency, and indicators of dwell versus movement, that gives the raw coordinates analytic meaning. A location is a dot on a map; a location with timing and frequency context starts to describe behavior.
Bringing the data together: the richer picture
No single signal type tells the whole story. Bidstream delivers breadth but variable precision. SDK delivers precision but narrower reach. IP-derived location adds context but lacks fidelity. Used on their own, each leaves a gap the others could fill.
Bringing the two primary sources together is what closes those gaps. Because the mobile ad ID is common to both bidstream and SDK signals, the broad coverage of one source and the high precision of the other can be combined rather than chosen between. The result is a more complete view of movement activity: wider coverage and higher confidence in the same picture, instead of a tradeoff between them.
Combining sources also changes what's possible in terms of timing. Different sources update on different rhythms, and joining them can yield more current visibility into movement activity than any single retrospective feed. This is the principle behind newer low-latency capabilities such as Venntel's Location Stream, built to deliver timely operational awareness rather than purely after-the-fact analysis. The shift from "what happened" to "what's happening now" is one of the more meaningful developments in how adtech-derived data is being applied.
In practice, this richer picture supports use cases oriented around areas and activity rather than any single point of interest. Teams use combined location signals for situational awareness across a geographic region, for monitoring movement around critical infrastructure and ports of entry, and for detecting shifts in activity that may indicate an emerging event. The common thread is understanding patterns of movement at scale, the kind of context that a single, partial data source can't reliably provide.
From combined signals to trustworthy intelligence
Bringing sources together is necessary, but it isn't sufficient. Raw adtech data, from any source, carries real quality challenges. Some signals are spoofed. Some describe physically implausible movement. Some are synthetic, generated by the commercial ecosystem itself rather than reflecting genuine activity. Combine two noisy sources without addressing this, and you get a larger pile of noise, not better intelligence.
This is why refinement is a distinct and essential step. Deduplication, forensic evaluation, and quality indicators are what turn combined signals into something analysts can actually rely on, separating authentic activity from artifacts before it ever reaches a workflow. It's the difference between raw location data and location intelligence, a distinction we explore in depth in Location Intelligence vs. Raw Location Data. The market has matured past the point where volume alone is the differentiator; today, the question sophisticated teams ask is how the data is validated and made meaningful.
The takeaway
For intelligence and OSINT professionals, the practical implication is straightforward: understand the sources, understand how they fit together, and prioritize quality at every step. Reach out to us here to see how Venntel brings these signals together for mission-critical operations.
