Mobilewalla Glossary

Affinity Score

A scored measure of how strongly a user or household aligns with a particular product, service, lifestyle, or behavioral segment. It is built by analysing engagement patterns, app usage clusters, purchasing proxies, and digital context signals.

Related Terms:
Engagement Score, Behavioral Signals, Audience Intelligence, Consumer Features

Alternative Data

Alternative data refers to non-traditional information sources, such as device metadata, app behavior, and location derived attributes, that provide insights into a borrower’s financial habits and digital activity. Unlike credit bureau data, it captures real-world behavioral patterns that can predict repayment capacity and intent. 

In LendBetter: We use alternative data to help lenders score applicants who lack meaningful borrowing history, improving credit access while managing risk.

Related Terms:
Thin-file Borrowers, New-to-Credit, Predictive Features, Data Signals

API Integration

The process of connecting different software systems to enable seamless data exchange. In the context of alternative data, APIs allow lenders, marketers, and telco operators to access real-time intelligence without managing complex infrastructure. Mobilewalla’s APIs deliver structured consumer insights, risk and fraud scores, and market intelligence directly into existing decision systems, empowering digital lenders to enhance credit underwriting workflows, and telcos to make smarter network and acquisition decisions.

Related terms: Real-time Scoring, Data Pipelines, SDK, AI-driven Scoring

Audience Intelligence

A data-driven understanding of consumer behavior, affinities, and intent derived from large-scale device and app signals. Audience intelligence helps businesses identify, segment, and engage high-value users based on how they interact digitally. Through MarketFlow, Mobilewalla provides broadband and telecom companies with granular visibility into subscriber behavior, competitive overlap, and regional dynamics, helping optimise marketing, acquisition, and retention strategies.

Related terms: Behavioral Analytics, Predictive Modeling, Alternative Data, Affinity Score

Behaviour Analytics

Behaviour analytics involves processing of signals from various user actions such as device usage, location footprint, app usage activity, to inform market segmentation and audience understanding. 

In Market Flow: These insights are combined with network data to build richer consumer intelligence for marketing and growth planning.

Related Terms:
Data Signals, Engagement Score, Event Analytics

Behaviour Cohorts

Clusters of users who exhibit similar digital behavior patterns over times such as location routines, app usage depth, or content affinity.

Use in LendBetter:
Improves risk model lift by capturing behavioral stability, volatility, and intent.

Related Terms:
Behavioral Analytics, Predictive Features, Affinity Index

Consumer Features

Consumer features are data-derived attributes that describe an individual’s behavior, preferences, or engagement patterns. These are the building blocks for predictive analytics and segmentation models.

In the context of Mobilewalla’s Data Intelligence:
Consumer features are generated from large-scale, privacy-compliant device data signals, such as app usage, mobility patterns, or transaction indicators. They enable businesses to understand users at a granular level and power models for credit scoring, personalization, and audience targeting.

Related Terms:
Predictive Features, Data Signals, Feature Mart

Credit Underwriting

Credit underwriting is the process through which lenders evaluate the risk associated with offering a loan or credit facility to a consumer or business. It involves assessing an applicant’s ability and willingness to repay based on available data and predictive models.

In LendBetter:
Adds behavioral and contextual features that reveal repayment likelihood and fraud risk.

Related Terms: Alternative Data, AI-driven scoring, Real-time Scoring, Predictive Modelling

Crop Circles

When plotted across time, geography, or signal density distributions, stable datasets form smooth, predictable shapes, while deviations or distortions indicate ingestion problems, signal loss, sampling bias, or SDK/RTB drift.
Crop-circle are clusters of unnaturally dense signals in specific geos. They can lead to misleading heat maps and inaccurate density estimates, distorting customer market analysis.

Mobilewalla solves for crop circles by applying spatial smoothening, detecting over-concentrations, and de-weighting non-residential geos. 

Related Terms:
Signal Integrity, Feature Stability, Imputation Logic

Data Enrichment

Data enrichment enhances existing datasets by adding new variables or context that improve accuracy and analytical power. 

In LendBetter: Lenders can enrich their applicant data with high-quality external signals, from device behavior to location derived insights, to build a more complete risk view. 

In Market Flow: Operators can enrich their first-party customer data with our insights to improve campaign targeting, capacity planning, and network rollouts.

Data Signals

Raw or lightly processed indicators drawn from location events, device metadata, app interactions, or network usage patterns.

In LendBetter:
Form the foundational input layer for fraud and risk features.

In Market Flow:
Enable insights on household broadband behavior and shifts.

Related Terms:
Behavioral Analytics, Time Series Device Connection, Device Metadata

Device Connection Classification

Device connection classification identifies how a device accesses the internet, via fixed broadband, fixed wireless, or cellular networks,  and which carrier provides it. 
In Market Flow: This classification underpins our ability to determine market share, flow share, and technology adoption patterns accurately. 

Related Terms:
Fixed Broadband, FWA, Subscriber Movement Analysis

Device Graph

A structured mapping of devices, households, and usage contexts that allows identity resolution and householding without personal data.

In LendBetter:
Highlights potential fraud risk (multiple devices per identity, device reuse patterns).

In Market Flow:
Supports accurate household broadband penetration estimates.

Related Terms:
Householding, Entity Resolution, Behaviour Analytics

Device Metadata

Technical and behavioural descriptors associated with a device, for example, OS version, app footprint, movement patterns, usage volatility, etc.

Related Terms:
Behavioral Analytics, Device Graph, Data Signals

Engagement Score

A derived measure of how consistently and actively a user interacts with apps, services, or networks. This score can help digital businesses identify cohorts suited for cross-selling.

Related Terms:
Affinity Index, Behavioral Signals, Audience Intelligence

Event Analytics

The process of capturing and analysing user actions or “events”, such as app installs, network changes, or ad interactions, to understand behavioural patterns and intent. In alternative data and telco intelligence, event analytics helps uncover how consumers engage with services across channels and timeframes. 
Mobilewalla leverages anonymized event-level data to build rich behavioral features that power credit risk models, audience segmentation, and market flow intelligence, enabling clients to make faster, more informed decisions. 

Related terms: Behavioural Analytics, Consumer Features, Predictive Modelling, Data Enrichment 

Feature Engineering

Feature engineering is the process of transforming raw, often unstructured data into analytically meaningful variables, called features, that can be used by machine learning models. It bridges the gap between raw signal ingestion and predictive intelligence by identifying, creating, and refining attributes that best represent underlying behaviours or phenomena. Effective feature engineering determines not only model accuracy but also its fairness, stability, and interpretability over time.

Why It Matters:
Feature engineering sits at the core of Mobilewalla’s differentiation: the ability to move beyond raw data access toward model-grade intelligence. It directly impacts:

• Predictive Power – features derived from real behavioral patterns improve model lift and reduce noise.

• Operational Efficiency – reusable engineered features can be served instantly via the Feature Mart and our Real-time API.

Related Terms:
Predictive Features, Feature Mart, Data Signals, Imputation Logic

Feature Mart

Mobilewalla Feature Mart is a repository of predictive features for various machine learning applications. It finds applications in various digital-first industries, including but not limited to Fintech, Telcos, Online Commerce, QSR and Retail etc.

Use in LendBetter:
Provides lenders with API access to Mobilewalla’s curated feature catalog for immediate integration into risk and fraud scoring workflows.

Related Terms:
Predictive Features, Consumer Features, Real-time API, Data Signals

Feature Stability

Feature stability measures how consistently a feature behaves across time, geographies, device cohorts, and data sources. A stable feature retains its predictive meaning even as inputs fluctuate, its distribution, variance, and correlation with outcomes do not change unpredictably.

Why It Matters:
In the alternative data ecosystem, stability is a marker of feature quality. Mobilewalla’s large-scale multi-source datasets produce millions of signals daily; stability ensures these signals translate into dependable, production-grade intelligence.

Related Terms:
Feature Engineering, Time Series Device Connection, Data Enrichment

Fixed Broadband (FBB)

Fixed Broadband refers to high-speed internet services delivered through physical infrastructure such as fiber, DSL, or cable lines. FBB represents a stable and high-capacity connectivity medium, forming the baseline against which other access technologies like FWA are compared.

Use in Market Flow:
Used to measure fixed-line coverage and market competition at neighborhood or regional scales.

Related Terms:
Fixed Wireless Access (FWA), Device Connection Classification

Fixed Wireless Access (FWA)

Fixed Wireless Access delivers broadband internet through wireless technologies (typically 4G/5G) to fixed locations. It provides an alternative to wired broadband in regions with limited fiber penetration and is a leading driver of connectivity growth in emerging markets.

Use in Market Flow:
Helps identify adoption rates, overlap zones, and competitive dynamics between wireless and wired access.

Related Terms:
Fixed Broadband, Device Connection Classification, Market Share

Flow Share

Flow Share quantifies the net subscriber movement between carriers over a given period, offering a dynamic perspective on market share evolution. It reflects not only who holds market share but how customer loyalty and churn are shifting.

Use in Market Flow:
Measures subscriber inflows and outflows across carriers to assess promotional impact and churn patterns.

Related Terms:
Market Share, Subscriber Movement Analysis, Mover Indicator

Fraud Detection Signals

Fraud detection signals are derived data features designed to identify irregular, high-risk, or potentially synthetic behavioral patterns. These may include device reuse anomalies, inconsistent location histories, or correlated device clusters suggesting identity fabrication.

Use in LendBetter:
Helps detect fraudulent loan applications by recognizing behavioral inconsistencies or improbable device patterns.

Related Terms:
Risk Scoring, Alternative Data

Geographical Roll-Up

Geographical roll-up is the process of aggregating data from micro-level geographies (e.g., households or census blocks) to broader administrative levels like postal codes, cities, or states. It provides a scalable view for both tactical execution and strategic decision-making, preserving flexibility in analysis granularity.

Use in Market Flow:
Enables visualisation of market performance at any spatial level — from neighbourhood level to national strategy.

Related Terms:
Granular Market Insights, Market Share

Granular Market Insights

Granular market insights are intelligence outputs derived from highly detailed data, often at household or block level. They help Telcos detect variations invisible in aggregated datasets, empowering precision strategy at the local level.

Use in Market Flow:
Supports network planning, acquisition targeting, and retention campaigns based on true micro-market behavior.

Related Terms:
Geographical Roll-Up, Audience Segmentation, Market Share, Flow Share

Household Penetration

Household penetration measures the proportion of homes within a defined geography that have active broadband or fixed-wireless connections. It is a key indicator of service maturity and market opportunity, reflecting both infrastructure reach and consumer adoption. By tracking changes in penetration, operators can identify underserved zones and forecast network demand.

Use in Market Flow:
Helps telecom operators pinpoint expansion areas, quantify addressable opportunity, and prioritize rollout planning at sub-regional levels.

Related Terms:
Householding, Market Share, Granular Market Insights

Householding

Householding is the process of linking multiple devices, identifiers, or users to the same residential entity. This grouping enables more accurate analysis of broadband adoption, network overlap, and household-level churn, while maintaining full privacy compliance.

Use in Market Flow:
Provides the foundation for household-based penetration metrics, churn analysis, and competitive mapping.

Related Terms:
Market Share, Flow Share, Household Penetration

Identity Graph

An identity graph maps connections between different identifiers — such as MAIDs, cookies, device signatures, and IP clusters — that correspond to the same anonymized user or household. It enables privacy-safe linkage across datasets without exposing personally identifiable information (PII).

Use in Mobilewalla Context:
Supports accurate cross-device recognition and enrichment by unifying SDK and RTB signals under a single anonymized device identity.

Related Terms:
MAID, Privacy-by-Design, 

Imputation Logic

A statistical framework for estimating missing or incomplete data points using observed correlations and contextual signals. Proper imputation safeguards feature completeness and reduces model bias.

Use in Data Products:
Applied to normalize and standardize feature vectors derived from heterogeneous SDK and RTB datasets.

Related Terms:
Data Sanity, Feature Engineering

Knowledge Graph

A knowledge graph organises interconnected entities, such as households, carriers, and geographies, into a structured network that reveals relationships and hierarchies. It supports faster querying and pattern recognition across complex datasets.

Use in Market Flow:
Allows users to navigate from national-level carrier trends to neighborhood-level subscriber patterns, improving competitive analysis and visualization.

Related Terms:
Geographical Roll-Up, Granular Market Insights

Location Derived Attributes

Location-derived attributes are contextual indicators inferred from mobility data, such as residential stability, commuting distance, or visitation behavior. These attributes enrich both financial and telco datasets by connecting user behaviour to physical context, a crucial dimension for risk and market understanding.

Use in LendBetter:
Reveal lifestyle stability and income proxies for better credit evaluation.
Use in Market Flow:
Assist in mapping catchment areas and regional demand pockets.

Related Terms:
Behavioral Signals, Data Signals, Granular Market Insights

Market Share

Market share represents the proportion of total subscribers or households served by a carrier within a defined market. It is a static snapshot that quantifies competitive position but, when paired with flow-share dynamics, reveals both market power and momentum.

Use in Market Flow:
Enables benchmarking, performance tracking, and competitive diagnostics at geographic and temporal granularity.

Related Terms:
Flow Share, Household Penetration, Granular Market Insights

Mover Indicator

A mover indicator detects whether a household or subscriber has recently relocated or changed broadband connections. Derived from temporal and location signals, it identifies transition events that often precede new subscriptions or churn.

Use in Market Flow:
Supports proactive outreach to newly relocated households likely to seek fresh connectivity options.

Related Terms:
Subscriber Movement Analysis, Flow Share, Householding

New-To-Credit

New-to-credit (NTC) consumers are individuals applying for their first formal credit product. They typically lack a bureau score but may demonstrate strong behavioral reliability through digital footprints. Advanced data modeling bridges this informational gap by evaluating digital stability and transaction regularity.

LendBetter empowers lenders to confidently underwrite first-time borrowers using device- and app-derived behavioral proxies.

Related Terms:
Thin-file Borrowers, Alternative Data, Risk Scoring

Predictive Features

Predictive features are engineered variables that amplify a model’s ability to forecast an outcome. They result from transforming raw, high-frequency data, behavioural, spatial, transactional, into normalised, stable, and interpretable signals suitable for AI pipelines. Predictive features are continuously evaluated for lift, stability, and drift to ensure sustained model performance.

LendBetter includes a library of predictive features that feed directly into risk, intent, and fraud models, boosting precision and enabling adaptive credit scoring.

Related Terms:
Feature Mart, Consumer Features, Data Signals, Risk Scoring

Real-time API

A real-time API delivers data or computed insights instantly to connected systems via programmatic calls. In Mobilewalla's advanced data products, these APIs act as conduits for live decisioning, powering underwriting, targeting, or fraud screening at the moment of interaction.

LendBetter's real-time API streams model-ready features and risk scores directly into credit decision engines, enabling instant approvals and adaptive pricing.
Market Flow's real-time API provides live access to carrier-level market metrics for automated dashboards and campaign execution.

Related Terms:
Feature Mart, Underwriting Automation, Consumer Features

Risk Scoring

Risk scoring quantifies the probability of default, churn, or fraudulent activity by synthesizing behavioral, transactional, and contextual data into a numerical index. It is the backbone of modern decision science, translating complex multi-signal behavior into actionable risk probabilities that drive automation and fairness in decisions.

LendBetter powers risk scoring for new-to-credit and thin file customer, strengthening underwriting accuracy and extend credit safely to these high risk segments.

Related Terms:
Predictive Features, Behavioral Signals, Credit Underwriting

SMB Carrier Identification

SMB carrier identification determines which broadband or wireless provider an individual small or medium-sized business uses. It relies on device clustering, network fingerprinting, and location intelligence to infer the connectivity profile of business premises.
This insight is crucial for commercial sales teams aiming to build precise prospect lists and understand competitor penetration among SMBs.

Market Flow helps broadband providers target high-value SMB accounts of competitors and track shifts in business broadband adoption.

Related Terms:
Market Share, Granular Market Insights, Device Connection Classification

Signal Integrity

Signal integrity measures the reliability, completeness, and consistency of raw data streams, such as SDK events, RTB impressions, or location pings, that feed enrichment and feature engineering. It ensures that variation in engineered features reflects genuine behavioral change rather than noise, missing data, or ingestion anomalies.
Mobilewalla's data solutions exhibit high signal integrity that underpins trustworthy predictive modeling and safeguards against error propagation throughout the feature pipeline.

Related Terms:
Feature Stability, Data Signals, Imputation Logic

SMS Parsing

SMS parsing is the automated extraction and interpretation of structured information from non-personal text messages stored on a device, such as banking alerts, bill payments, salary credits, wallet transactions, or utility notifications. 

LendBetter's SDK performs on-device SMS parsing to derive financial and behavioural features used for risk scoring, fraud detection, and affordability analysis. It allows lenders to access high-signal intelligence during underwriting while maintaining full privacy compliance.

Related Terms:
Predictive Features, Feature Engineering

Subscriber Movement Analytics

Subscriber Movement Analysis tracks how users migrate across broadband or wireless carriers over time. It blends device connection patterns, location behavior, and network usage to reconstruct the customer journey from one provider to another.
This longitudinal perspective helps operators understand churn triggers and competitive pressures at a granular level.

Mobilewalla's Market Flow identifies neighborhoods with high churn, reveals competitor acquisition surges, and highlights where technology transitions (DSL → Fiber, Fiber → FWA) are occurring.

Related Terms:
Flow Share, Mover Indicator, Device Connection Classification,

Thin-File Applicants

Thin-file loan applicants have limited or no traditional credit history, making bureau-based risk assessment insufficient. Alternative data, including device behavior, location-derived stability metrics, app usage patterns, and SMS-derived financial signals, enables lenders to infer creditworthiness from real-world actions rather than formal account history.

LendBetter provides lenders with rich behavioral features allowing confident underwriting of thin-file and first-time borrowers.

Related Terms:
New-to-Credit, Alternative Data, Risk Scoring, Predictive Features

Time Series Device Connections

Time series device connection analysis tracks how a device’s connectivity type (e.g., fixed broadband, FWA, cellular) evolves over hours, days, or months. These longitudinal patterns reveal technology adoption trends, provider switching behavior, and stability of usage.
Time series modelling enhances market intelligence by identifying leading indicators of churn, upgrade cycles, and service dissatisfaction.

Market Flow supports detection of emerging migration trends, such as shifts from DSL to fibre or fibre to 5G FWA, at neighbourhood-level granularity.

Related Terms:
Device Connection Classification, Feature Stability, Subscriber Movement Analysis

Underwriting Automation

Underwriting automation integrates real-time data, predictive features, and decisioning logic into a seamless approval workflow. It replaces manual evaluation with automated scoring and rule-based outcomes, reducing turnaround time while increasing accuracy and operational efficiency.
Modern underwriting automation relies heavily on high-frequency behavioural features that provide deeper insight than traditional application data alone.

LendBetter combines Mobilewalla’s real-time API features with lender decision engines to automate approvals, detect fraud earlier, and optimize risk-adjusted pricing.

Related Terms:
Real-time API, Risk Scoring, Credit Underwriting Intelligence, Predictive Features

View-ability Window

A view-ability window is the defined time horizon, such as 30, 60, or 90 days, over which behavioural signals are aggregated and analysed. Different windows reveal different behavioral patterns: short windows capture recency and volatility, while longer windows capture stability and lifecycle trends.
Optimizing viewability windows is crucial for achieving the right balance between predictive power and model responsiveness.

Use in Market Flow:
Determines the temporal sensitivity of churn and upgrade prediction models.

Related Terms:
Time Series Device Connection, Feature Stability, Predictive Features