Inspire 2013 - Paychex +Alteryx = Efficiency Redefined


Published on

Over the past several years, with the growth of big data, the need for high-level analytics at Paychex has grown exponentially. For quite some time, hard work, long hours and expertise in basic database management software was enough to get by and meet deadlines. However, in the current environment with growing information demands, the pre-existing technology environment wasn't cutting it. Enter Alteryx. Alteryx has redefined Risk Analytic's capabilities and bandwidth within the organization. After just a few months of usage, we are recognizing a 32x improvement (at the minimum) in processing speeds alone. This has, and will continue, to open doors for new projects, larger data sets and cutting edge predictive modeling

  • Be the first to comment

  • Be the first to like this

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide

Inspire 2013 - Paychex +Alteryx = Efficiency Redefined

  1. 1. Paychex + Alteryx = Efficiency RedefinedPaychex, Inc.Michael BaroneWednesday, March 6th
  2. 2. Paychex, Inc. Profile Provider of comprehensive payroll, human resource, and benefits outsourcing solutions for small to medium sized businesses Approximately 564,000 clients & more than 100 offices across the U.S. & Germany 9 million client employees provided with Payroll and HR services Fiscal 2012 – revenue of over $2 billion and net income of $0.5 billion
  3. 3. Accolades
  4. 4. Risk Analytics – Who are we? Risk Management Frank Fiorille, Director Risk Analytics Erika McBride, Manager Paychex Peer Predictive Process Program Analytics MIS & Portfolio (P4) 3 Modelers/SAS 4 Analysts 1 Analyst Programmers
  5. 5. Risk Analytics – What we do…
  6. 6. Model Profiles TIM (Territory Identification & SARAH Mapping) In Development Ranks the inherent (Sales Rep SAM sales opportunities Attrition Client Referral Model (Sales within a zip code Heuristic) PEG Anticipation Identifies our Model) Sales outside sales reps Prospect Model (Price Elasticity Gauge) OLIVER Strategy with a propensityPAM Clustering model that Predicts the Models (Objective to turn over for Product Sequencing gauges the price number of(Paychex Attrition Model) Lifetime controllable sensitivity of each sales units per Indicator of reasonsClient retention model client the Value zip code for Discount Recommendationsgeared towards the coming of Expectedcontrollable losses fiscal year Revenue) PST1000 Operations MARCO Predicts PAUL and Risk (Paychex Smart Models lifetime (Model for Accounts Time) (PEO Algorithm Utilizing value for Receivable revenue Identifies clients Leads) Collections Credit Risk and most likely to Identifies clients most likely Optimization ) retention purchase our Time Identifies to cross over to our Identifies which clients most propensity Clock solution Professional Employer delinquent accounts likely to have a PATRICK Organization product Up-Sell are most likely to payment (Predictive Cross-Sell repay their balance returned Models STEPH Algorithm within 90 days within the next Targeting (Sales Tool 90 days Retirement Evaluating Workers Investment Client Potential Health Compensation Knowledge) and Benefit Identifies clients Clients) Identifies clients most likely to most likely to Identifies clients purchase our purchase our 401(k) most likely to Workers product purchase one of Compensation our Health and product Benefit products
  7. 7. Monthly Data Prep and Processing Overview 9 Alteryx 55 data files 55 data files databases 18 Text Text databases 14 Excel Excel Alteryx Access HTML modules Oracle HTML Netezza 3 GB new data 10 GB new data B storage storage A each raw file overwritten 475 E F Alteryx F each raw file can be saved 475 variables variables O T R E E R 51 33 hours Saves 7 GB per month new data storage minutes processing and time 32 hours per month processing time
  8. 8. Specific Example (PAUL Model) – Before Alteryx Database 1 1 GB 5 linked files 12 queries 3 temp tables ODBC 3 final tables Database 3 1 hour processing time Database 1 GB 4 linked tables 1 ODBC connection 48 queries Database 2 1 GB 16 temp tables 3 linked files 1 final table 4 queries 2 hours processing time 1 temp tables 1 final table 1 hour processing time
  9. 9. Specific Example (PAUL Model) – After Alteryx
  10. 10. Another ExampleWanted to track and trend detailed client historical firmographicinformation over several years, outside of the Enterprise DataWarehouse (no IT involvement)Each month required its own database due to size limitations(currently we are at 60+ databases, each holding about 500,000records)When querying the data, we’d have to link these databases together,and running queries took hours, if not daysMaking a change to the process could take up to a weekWith an Alteryx database (.yxdb) we can store all data in one placeQuerying the data takes minutes
  11. 11. Other benefits! Top of the line Support Team ( – quick response time, knowledgeable and makes you feel like you’re their only client Very user-friendly interface Superb ‘Alteryx Help’ files - actually explains and walks you through things instead of just repeating definitions like most help files!
  12. 12. What’s next… Enhance/expand data mining Convert internal data warehouse process to Alteryx Explore ‘Big Data’ opportunities Continue driving Analytics further into Paychex Explore analytical services for our clients
  13. 13. Thank You!