Many businesses spend heavily on digital marketing to generate leads for their products or services. Yet a growing share of leads can turn out to be of little or no value. The digital landscape is one – to an increasing degree – where automated submissions, fabricated identities, and organised fraud now operate. Many of these bogus entities are constructed specifically to pass basic validation checks and enter sales pipelines, masquerading as genuine prospects. The result is a drain on marketing budgets and sales capacity, and where each cent of marketing spend yields less value.
Lead fraud refers to submissions that appear valid but do not represent real customers and can take several forms. Automated scripts generate large numbers of entries, disposable email services allow false identities to pass superficial checks, and fraud networks recycle or duplicate lead data across campaigns. While automated software is behind many such dubious operations, in some cases groups employ human workers to submit forms manually using fabricated details.
Traditional lead verification methods have not kept pace with these developments. Email validation confirms whether an address can receive messages, not whether the person behind it exists. Phone verification can be bypassed through virtual numbers and SMS is easily hackable. IP tracking provides only limited insight, as virtual private networks can obscure location.
The problem stems from many validation tools operating on static data. The answer may lie in observing how users behave during digital interactions.
Real-time user analysis
Behavioural biometrics examine patterns in how a user types, moves a cursor, scrolls a page, or completes a form. These signal human motor habits that are difficult to reproduce artificially. While fraud techniques have advanced, replicating the variability of human behaviour remains challenging.
Systems can measure typing speed, the interval between keystrokes and typing corrections. Human input, as might be expected, tends to vary, with hesitations and occasional errors. Automated scripts by default will produce uniform timing and machine speed completion. Even when designed to mimic human input, they struggle to reproduce natural irregularity over longer sequences.
Mouse movement analysis offers another source of verification. Movements show small adjustments, pauses and changes in direction, while automation will appear precise and linear. Therefore, immediate clicks and consistent mouse travel paths will indicate non-human interaction.
Form completion behaviour provides longer context windows. Systems can track how long a user spends on each field and the overall time taken to submit a form. Genuine users tend to consider and revise their input, while algorithms complete complex forms in fractions of a second. Page scrolling and navigation around an online property give similar indicators that can help determine whether a user is human or silicon.
A key feature of behavioural biometrics is an ability to address automated bots and human-staffed fraud operations. Bots use the same identifiers that are exhibited by browsers and devices, but their interaction patterns remain detectable. Human fraud farms will present more complex challenges: those so employed are real people, yet behaviour may still show repetition. Similar typing speeds, consistent interaction flows and high submission volumes can reveal coordinated activity. The presence of human-powered fraudulent outfits will inevitably muddy the waters, and potentially lead to false positives and inaccurate red-flagging.
In practice, behavioural biometrics is rarely used in isolation, being one of many tools used to discover anomalous behaviour. Advanced systems combine behavioural analysis with more traditional checks running on the basis of email, IP and geolocation, and device fingerprinting. Each tool adds its partial information to produce a probabilistic score that helps organisations better assess lead quality. Outcomes will vary by sector and the nature and mix of tools.
Outcomes that marketers want
Removing fraudulent leads reduces wasted advertising spend. Sales teams will spend more time on viable prospects, thus improving conversion rates, and overall, detection of organised fraud operations will help prevent repeated losses. Part of the balancing act is the need to reduce false positives. Overly aggressive filtering may exclude legitimate users whose behaviour falls outside expected patterns, and those entities disregarded by software toolsets have all the incentive they need to seek a competitor’s product or platform.
Adoption of real-time biometric verification has been most visible in sectors where lead quality affects financial risk directly. Financial services screening loan and credit applications are a prime example of current use, and industries with high customer acquisition costs – home services and solar panel installation, for example – have shown interest. In e-commerce and software services, the focus is often on protecting promotional campaigns and onboarding processes.
It’s inevitable that fraud techniques will evolve, and the nature of the metrics gathered by static verification methods means the indicators used by software can be spoofed. Behavioural analysis introduces a dynamic element by focusing on how interactions occur, but requires thought about ways organisations can calibrate their judgements over time.
There is clearly no single solution that will eliminate lead fraud. Behavioural biometrics have to balance accuracy, the cost of the platform, the user experience, and the potential of a lead lost in error. It adds a layer of scrutiny difficult for fraud systems to replicate consistently (or, more accurately, to replicate inconsistently-enough). For organisations that rely on lead generation, the question is how to integrate behavioural biometric signals without introducing another source of error.
(Image source: “Studying Flame Behavior in Microgravity with a Solid “High-Five”” by NASA’s Marshall Space Flight Center is licensed under CC BY-NC 2.0. To view a copy of this license, visit https://creativecommons.org/licenses/by-nc/2.0)
Find out more about the Digital Marketing World Forum series and register here.
