perfect price - silicon valley pricing webinar
Post on 13-Apr-2017
87 Views
Preview:
TRANSCRIPT
alex@perfectprice.io chris@svpricing.com
Silicon Valley Pricing + Perfect Price
Copyright 2017 Perfect Price, Inc. All rights reserved.
Introductions
Alexander Shartsis CEO, Founder Perfect Price
Chris Herbert Founder
Silicon Valley Pricing
PERFECTPRICE
Agenda
Nobody is selling anything here
PERFECTPRICE
Logistics
Ending promptly at 10:55 Ask questions in GoToWebinar chat How to ask more questions 1:1 afterwards
PERFECTPRICE
Feedback
Survey at the end via email Topics for future webinars Improvements to webinar format Takes only 2 minutes!
PERFECTPRICE
Case Studies: Pricing Today
Pricing yesterday
PERFECTPRICE
Pricing today
1993: Charge against GM earnings
$744,000,000
1995: Sold by General Motors
$1,200,000,000
PERFECTPRICE
Enterprise locations in NYC (just the first 20 of 60+)
Data from everywhere
Dynamic pricing everywhere
Just the USA price
How can humans keep up?
“There is a gap between where we are and where we should be Every minute we are behind, we lose money.”
- CEO, Fortune 500 Company
PERFECTPRICE
Perfect Price solution
Behaviors (JS, API)
Hadoop Cluster (AWS)
Raw Data (API)
Raw Data Aggregate Data Intelligence Insight Decisions Action
Demand Model
Using Artificial Intelligence
+ Machine Learning Price Testing
Strategic Decisions
Cost/Profit Model
Outside Data
Strategic ChangesVisualizations
Dynamic Pricing
Customer Perfect Price
SAP, Oracle, etc.
Forecast
Predictive Analytics
Dynamic Pricing
3 steps to know what you’re losing
Clean up data Model demand Quantify losses
Collect required data from across your organization
Cluster and segment Build demand model as if you were repricing
Compare new prices to current Quantify losses at all levels
Step 1: Clean up data
Myth: We have all the data
Fact: Right data doesn’t exist
Lots of data, none of it is in a usable format
Missing data in a $300mm retail company
64%
36%
Source: Baby Center 2015 US Mobile Moms Report
0%% of moms who
check prices on mobilemobile app data
captured by this company
3mmMonthly App Users
Data collection timelines
Simple
Complex
50 days 100 days 150 days 200 days
Perfect Price Experience
Return on effort?
How long should data cleanup take?
It depends on number of channels and availability of data Most companies can get a sample together in <2 weeks Raw data is usually ready in ~4 weeks, sometimes less Processed data can be blocked by engineering roadmap
Step 2: Model demand
Myth: Fact: Not every hypothesis turns out to be right The costly part is knowing the approach
Excel + 1 analyst + 1 weekend = The Answer
The way Perfect Price models demand
Global data(Weather, &c)
Industry data(Aggregator)
Your data(Views, prices,
conversions, &c)
Machine learning engine
Pre-processing Clustering Demand
predictionRevenue
mgmt
you.com
Your product$75.00
Buy now
Explaining machine learning to laypeople
Scrutinize the model
weekday
weekend
Scrutinize the model
ideal prices (vertical lines)
dots = raw data
Quantify losses
You’ll know you’re ready when…
You know exactly how much you’re losing You know exactly where you’re losing it You feel confident in making a change Budget is thrown at you to fix it ASAP
Comprehensive ≠ all knowing
ClusteringProduct, customer, and location Enables accuracy with limited data
Relatively static
Long modelPrices for every product, time combination Powers dynamic pricing, forecasts
Burst modelRerun every ~5 min Captures the unpredictable Enables long model to be simple Captures max value
What Clean Data Looks Like
The Same
Product
Totals/Avg.
Combining internal and external data
Powerful search
Quick filtering
At-a-glance competitive environment
Focus on the important
Make the important accessible
Share data with stakeholders
Testing and Experiments
The Perfect Price testing cycle
Get results Monitor
Analyze
Plan Create
Run
“We just want to test it”
Difference in differences made simple Output where stakeholders or end users need it Enabling industry professionals like Optimizely in A/B testing
Easily track, access and collaborate
Centralized control of price testing
Google centralizes all testing in 1 team, to prevent bad data Perfect Price is 1 test dashboard for the entire company Be warned whenever trying to add a product that is in a test
Key study results, easy to read and share
Enabling the Business
The future: predictive analytics
Summary & Conclusions
3 step process for starting out Powerful, dedicated system for pricing only Clean, centralized data for fast decision making and experiments Simple, powerful testing framework enables more testing Dual model approach for real time pricing applications
alex@perfectprice.io chris@svpricing.com
Q&A
top related