simMachines is user friendly, easy to understand, and fast to learn. We are passionate about bringing exceptional performance, innovation, and explainability to every client application. We train your teams on how to use our technology and applications, or we can provide broader support for your application deployment.
Technology Evaluations
simMachines offers several types of Technology Evaluations or Proof-of-Concept (POC) projects for clients to evaluate functionality, insights, prediction accuracy, and performance lift. Pricing and duration of each project are dependent upon project complexity.
Dynamic or Augmented Segmentation Projects
simMachines clusters first-party data to identify the specific weighted factors that differentiate machine driven audience groups or other data being analyzed, like social media or campaigns.
Requirements: A single file containing customer names and all available information, with a unique identifier number for each record that is relevant to the analysis being performed. The broadest set of data is desired to uncover new insights. Augmented projects will leverage 3rd party data sources, to enrich these insights with additional demographic, behavioral, or attitudinal data.
Deliverables: A cloud instance with interactive visualizations of the resulting clusters, CSV file identifying the cluster each record belongs in, project notes, and findings / recommendations.
Estimated Delivery: 1 – 2 weeks
Standard Custom Project
simMachines produces a solution for a standard application (e.g. customer churn, fraud detection, hiring recommendation, etc.), using first-party data from a single file and a custom tuned distance function.
Requirements: A single file containing all data required for the project. Ground truth data will be required for performance reporting at the end of the project.
Deliverables: A cloud instance capable of providing predictions through our web-based GUI, visualizations to provide insight into the data, project notes, and findings / recommendations. The POC can easily be leveraged into a deployable solution.
Estimated Delivery: 4 – 6 weeks
Advanced Custom Projects
simMachines produces a solution for a standard or non-standard application(s) (e.g. customer churn, fraud detection, hiring recommendation, etc.), using first-party or third-party data from one or more file(s) and a custom tuned distance function.
Requirements: One or more file(s) containing all data required for the project. Ground truth data will be required for performance reporting at the end of the project.
Deliverables: A cloud instance capable of providing predictions through our web-based GUI, visualizations to provide insight into the data, project notes, and findings / recommendations. The POC can easily be leveraged into a deployable solution.
Estimated Delivery: 6+ weeks
Technology
simMachines offers flexible technology options for deploying our software and applications:
- Cloud-based, client provided or on premise SaaS solutions with recommendation, classification, regression, and search, based on non-metric and metric distance functions.
- Universal distance function that automatically and dynamically weights variables by relevance for each individual prediction
- Succinct similarity data structures for maximum scalability
- Access through our web platform, or integration into your existing workflow with our APIs
Training
We offer extensive training and support to learn how to deploy our technology. After training, our Client Success organization provides ongoing support and guidance to ensure that all software upgrades, questions, and performance objectives are managed successfully. Training typically involves live projects, with real data sets to set up in the software, and configuring the appropriate engines to run experiments on. Over the course of several days users will work within the software to learn how to deploy solutions for specific use cases, quickly and efficiently. User manuals and tutorials are provided.
Pricing
simMachines prices marketing, fraud and compliance applications on a per prediction basis, unless the application is for analysis only, then pricing is based on a monthly fee. Multi-year contracts that take into account annual volumes and monthly minimums are typical following technology evaluation projects. Standard components of our pricing model include an initial set-up fee and either minimal, or full-service support options.
Application Deployment
There are four main phases for application deployment for a use case: 1) Data pre-processing, 2) Data sculpting, 3) Distance function optimization and generation of predictions, 4) Reporting & results.
The following timelines are typical in project deployment but can vary in duration as a result of the data pre-processing required or the complexity of custom projects.
Deployment Overview
simMachines client success organization ensures the success of each application deployment by providing ongoing access to data scientists and general support resources to answer questions, review and monitor performance, ensure prediction volumes are being tracked and reported accurately, and to make proactive recommendations regarding approaches and data sources that can improve performance based on application.
1. Customer Data
Customer transfers data in a manner agreed upon by the technical teams of both companies. Data scientists examine data in detail to understand relationships to facilitate sculpting.
2. Data Sculpting
Data is normalized and feature engineering is conducted so that the distance function will be more effective. Data is transformed in a semantics-preserving way.
3. Distance Function Optimization
Distance function most suitable for data is identified or produced. Our proprietary Dense Nearest Neighbor distance function generator is well suited for most tasks, otherwise we will identify another metric or, if necessary, will design one manually. In either case, an optimization step is required.
4. Report Generation
A report is exposed to the customer. Data scientists review the results with the customer and perform calibrations and updates. The report is ready for consumption.
Pre-sculpted Data Differences
Data consistency between projects dramatically reduces delivery time. No sculpting requirement and analyzing new customers/products requires only minor changes to distance function. Possible to complete projects in hours.
