MOVING FORWARD: NEW METHODS FOR CUSTOMER … › wp-content › uploads › ...MOVING FORWARD: NEW...
Transcript of MOVING FORWARD: NEW METHODS FOR CUSTOMER … › wp-content › uploads › ...MOVING FORWARD: NEW...
MOVING FORWARD: NEW METHODS FOR CUSTOMER ENGAGEMENT
Anne E. Dougherty Katherine Randazzo
Amanda Dwelley
We require more of end users as energy and climate goals
increase
The energy landscape is changing dramatically
To rise to these challenges, we need more sophisticated research techniques
Increasingly, we need to move to a more one-to-one understanding of end-users
Learnings from other fields can help get us there
Why Look Further?
2
Consumer Product Marketing: Latent Class Discrete Choice Analysis (LCDC)
Political Science: Micro-Targeting
Human Development: Latent Growth Curve Analysis
New Approaches for New Insight
3
Consumer Product Marketing: LCDC
4
Consumer Product Marketing: LCDC
5
Latent Class Discrete Choice Modeling generates customer segments by identifying product attribute preferences through trade-off analysis
Consumer Product Marketing: LCDC
6
LCDC: A-Line Segments
7
No Spend, No LED
On-the-Go Shoppers (14%)
Tech Lovers (25%)
Frugal DIY-ers (31%)
Conventional Shoppers (30%)
No Spend, LED
Spend, No LED
Spend, LED
Willing to Spend
Sample Size: 252
LED Interest
LCDC: Reflector Segments
8
Deal-Driven (13%)
Energy Enthusiasts ( 49 %)
Value-Seekers (14%)
No Spend, LED
Spend, No LED
Spend, LED
No Spend, No LED
Willing to Spend
Sample Size: 224
LED Interest
Product Experimenters (24%)
LCDC: Short-term Targets
9
On-the-Go Shoppers (14%)
Deal-Driven (13%)
Tech Lovers (25%)
Frugal DIY-ers (31%)
Energy Enthusiasts ( 49 %)
Value - Seekers (14%)
Conventional Shoppers (30%)
No Spend, LED
Spend, No LED
Spend, LED
No Spend, No LED
Willing to Spend
Reflector groups A - line groups
LED Interest
Product Experimenters (24%)
It’s Best to Use it When: You have a new product or offering to take to market You need clear direction for design and merchandising You need to prioritize marketing efforts in competitive
channels to target customers
Latent Class Discrete Choice Modeling
10
Political Science: Micro-Targeting
11
Political Science: Micro-Targeting
12
Traditional segmentation defines and divide a large homogenous population into identifiable groups based on similar characteristics
Micro-targeting identifies individual household propensities to act
Political Science: Micro-Targeting
13
Political Science: Micro-Targeting
14
The Obama 2012 campaign was the first in history to successfully leverage Facebook as a tool for mobilizing voter turnout on a mass scale
Applying Micro-Targeting to Program Offers
15
1. Develop Core Database
2. Surveys to Obtain APS-specific Behavioral & Psychographic Data
3. Intermediate Segment Variables
4. Populate Customer Database with New Variables
5. Develop Program-specific Propensity Models
Applying Micro-Targeting to Program Offers
16
1% 4%
5.7%
It’s Best to Use it When: You have access to multiple data sources Transactional or behavioral data is available You have teams willing and able to collaborate and to share
knowledge
Micro-targeting
17
LGCA uses cross-sectional, time-series analysis Identifies consumption curves and how they differ Predicts and describes the shape of consumption curves Identifies which factors mediate consumption
growth (or decay)
Human Development: Latent Growth Curve Analysis
18
Human Development: Latent Growth Curve Analysis
19
25th Percentile
50th Percentile
75th Percentile 95th Percentile
4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
Cogn
ition
Age (Years)
Human Development: Latent Growth Curve Analysis
20 Intelligence volume 40, issue 1, January ti February 2012 "Is age kinder to the initially more able?: Yes, and no“ J. Gowa, Wendy Johnsona, Gita Mishrab, HALCyon Study Teamb, Marcus Richardsb, Diana Kuhb, Ian J. Dearya
Human Development: Latent Growth Curve Analysis
21
1 2 3 4 5 6 7 8 9 10 11 12
Usag
e
Month Mean Growth Smith Jones Taylor Anderson
Human Development: Latent Growth Curve Analysis
22
Control Group
Treatment A
Treatment B
Treatment C
-1 -2 -3 -4 -5 -6 -7 -8 -9 -10 -11 -12 0 1 2 3 4 5 6 7 8 9 10 11 12
Usag
e
Month
It’s Best to Use it When: You are working with a heterogeneous population You expect wide variation in response to stimulus You have to identify customer groups for targeting
Latent Growth Curve Analysis
23
Market Segmentation 24
Other disciplines offer field-tested methods Greater granularity provides a finer point for targeting Expand programs by aligning offerings with: Customer needs and wants Customer-specific messaging
Closing Points
25
QUESTIONS? Anne E. Dougherty
Director of Social and Behavioral Research [email protected]
REFINING AN AKA-B MODEL FOR GREATER BEHAVIOR CHANGE
Katherine V. Randazzo, Ph.D., Opinion Dynamics Jane S. Peters, Ph.D., Research Into Action
Caroline Chen, Statwizards Brian Smith, PG&E; Andrew Fessel, PG&E
Introduction
Refining the AKA-B Model 2
This presentation is about going beyond the low-hanging fruit in promoting energy-efficient behavior For many years the dominant “model” of how to get customers to save
energy was the PTEM (Physical-Technical-Economic Model) Incent people to purchase technologies and the technologies will save
the energy This was enough for a while, but we have less low-hanging fruit now
I’m going to talk about using a conceptual model of change to help us think about how to go beyond where we have been How do we go beyond installing lighting and certain one-time purchases? How do we get to the human element and to durable changes in the use of
the energy-efficient products? How do we avoid take-back?
California’s IOUs have long used a simple model of behavior change to think about influencing customers to change their energy efficiency behavior AKA or AKA-B Awareness, Knowledge, Attitudes, Behavior Shows up in most program logic models and statewide program planning metrics
Background
3 Refining the AKA-B Model
The CA IOUs wanted an assessment of the currency & usefulness of AKA AKA toward what? Programs? Energy efficiency? A technology? Attitudes are not enough to change behavior
Subsequently, jointly we developed a revised model and generated generic questionnaire items to track movement of customers based on that model and describe participants and non-participants at any point in time to see if
the programs were reaching the right customers
Revised Model (synthesized from a variety of literatures)
4 Refining the AKA-B Model
Stage Model—not causal Can be used for different motivational
domains Environmental Financial Comfort
Addresses durable change-not one-time or immediate decisions
Can be used for general or specific objects of awareness etc Program Technology/appliance, etc
(these aren’t actual interview questions)
Mapping Revised Model to Original AKA-B
5 Refining the AKA-B Model
Awareness and Knowledge Collapsed Attitude expanded into
Concern Personal Responsibility
Intention Added Behavior Expanded into
Behavior Change Maintenance
ak A B
Some Uses of the Model
6 Refining the AKA-B Model
Describes where customers are in the process of changing behavior in a sustained way
Knowing this helps with messaging and targeting If target customers are already far along, then information will not be
effective If customer is not knowledgeable, then information may be most effective
Provides new insight into the decision drivers for participants and nonparticipants, possibly a better estimate of true readiness to adopt technology and change habits on their own
Can be combined with segmentation to even better describe customers/participants/non-participants
Provides guidance on what to measure and at what point
Causal Model with Intervention Points
7 Refining the AKA-B Model
What We Did Next
8 Refining the AKA-B Model
Developed questionnaire items to measure each construct Reliability and validity good—following two pre tests testing nearly 100 items
Some insights from our initial uses of the model: It is essential to distinguish between influencing someone to make an energy-
efficient purchase (or behavior) on one occasion versus influencing sustained changes (we knew this before but testing the model reminded us)
It is easier to predict/influence specific attitudes and behaviors than general ones
But general attitudes may be important as the context in which to influence specific behaviors at any given moment Triggers Making new behaviors convenient
It is a lot to ask for a specific utility/third party program to change customers’ environmental awareness or convince them to take personal responsibility for it!
Refined Model (representing insights from prior slide)
9 Refining the AKA-B Model
But programs can build on a foundation of environmental awareness/concern that was built by mass media campaigns
/
Awareness/ Concern Personal Responsibility Intention Behavior
Change Maintenance
Awareness/Knowledge Intention
Program Makes it Convenient to Change
Habit
Program Information
Behavior Change
Program Messaging
Provides Decision Heuristics
Program Provides Trigger
Specific
Overcoming Barriers: Examples
General Knowledge
Refining the AKA-B Model 10
Examples: General akAB Messaging
Making the connection between energy and the environment (AK)
Refining the AKA-B Model 11
Examples: General akAB Messaging
Concern, Personal Responsibility Message
Examples: Specific Programs
Refining the AKA-B Model 12
Back to Some Uses of the Model
13 Refining the AKA-B Model
If we can identify customers are the general Intention stage, we can focus efforts/messaging to such things as: Awareness and knowledge of program Convenience Decision aids Triggers
If they are at the general Awareness/Knowledge stage Convenience isn’t enough Information about the program isn’t enough
If have some concern about the environment but have low self-efficacy about doing anything about it, apply different kinds of messaging, e.g.: Value of collective action Examples of the power of the collective: if ½ Californians replace 5
incandescent bulbs with CFLs, it is like taking 200,000 cars off the road
Back to Some Uses of the Model (cont)
14 Refining the AKA-B Model
The model can be used as the basis for experimentation Identify groups of customers at different stages Randomly assign them to different messaging conditions Assign to different program interventions Then test the relative efficacy of different interventions on different groups
(at different stages) This will help you fine tune messages but also test whether you are thinking
correctly about your customers and what will influence them
Summary
Refining the AKA-B Model 15
If there is already a foundation at the general level, programs can provide such things as: Information about the program and its benefits, Triggers to take long-intended action, Ways to make taking program-recommended actions convenient, Help for customers to change habits
Mass media programs by utilities, commissions, etc. can work at the general level so that specific programs can be more effective at a level appropriate to them These efforts would clearly be more effective if coordinated
We have said that the model can be used for general or specific behavior and attitude changes, but this version of the model shows how these levels can be used together for greater effectiveness
However, that approach starts with using a conceptual model such as this one for directing both types of actions
SAVING WASTE: ENERGY USE AND WASTE ANALYSIS BY END-USE
Bill Norton Opinion Dynamics Corporation
November 13, 2012
Presentation Overview
Statement of research objectives
Review of traditional approach
Overview of usage and waste analysis
Application of results/implications for future research
2
Research Objectives
Identify gaps in program offerings by providing a more complete assessment of usage at the end-use level
Disaggregate electricity usage by end-use and segment
Develop energy use profiles by end-use and segment that quantify: Base case usage Excess energy use due to inefficient technologies Energy “waste” due to customer behaviors Efficient case usage
3
Research Objectives (cont.)
4
Plug Load %
HVAC %
Cooling %
Other %
Segment Usage by End Use (% kWh)
Ref. %
Lighting %
HVAC End Use Energy Profile (%kWh)
Efficient Case Use,
X %
Technology Waste,
Y %
Behavioral Waste,
Z %
Disaggregation of segment usage by end use
Disaggregation of end use specific usage into efficient case and waste components
Standard Approach
A traditional potential study quantifies available energy savings from DSM by segments
Forecasts are developed by: Understanding baseline energy usage and market
conditions Modeling/forecasting market response to DSM programs
5
Where Behavior Gets Lost
Traditional potential study produces results too blunt for strategic program planning and program gap analysis: Rely heavily on secondary data reflecting regional/national trends Results do not adequately reflect potential associated with behavior
change
Behavior is addressed: Embedded within engineering algorithms (e.g. Hours of Use) Reflected in assumptions regarding efficient measure adoption
However… Baseline assumptions used - no adjustment for “efficient” behaviors Engineering assumptions fail to capture all behavioral influences
6
Behavioral Component
Energy use is defined by the interaction between end users and technology – accurately assessing behavioral component is critical to quantifying savings potential EXAMPLE: ELECTRIC SAVINGS FOR HIGH SEER CAC
ΔkWH = (FLHcool * BtuH * (1/SEERbase - 1/SEERee))/1000
7
End user behavior is embedded here and represents assumed set points at given
outside air temps
A New Approach
Enhanced primary data collection to inform understanding of: Baseline equipment saturation and penetration Baseline building and equipment characteristics Customer equipment use and occupancy patterns
Determination of efficient case behaviors for end uses
Enhanced engineering assessment to more accurately reflect behavioral component of energy use and waste
8
Primary Data Collection: C&I
9
Telephone Survey: 1,600 completes Penetration/saturation 3 end uses; Behavioral/operational practices
Onsite Audits: 345 completed visits Penetration/saturation all end uses; Equipment technical specifications; Behavioral/operational practices
Monitoring: 70 audited sites Lighting, HVAC (limited ref); Occupancy
• Nested sample design 311 onsite audits • Audits completed at sites not nested in phone survey sample included collection of
segment & operational information during recruitment calls
Primary Data Collection: Residential
10
Mail Survey: 4,414 completes; Penetration/saturation; Behavioral/operational practices
On-Site Audits: 297 completes; Penetration/saturation all end uses; Equipment technical specifications; Behavioral/operational practices
Monitoring: 140 completes; Current logging on all circuits; Lighting / occupancy logging; Temperature and humidity
• Originally designed as fully nested sample • Audits completed at sites not nested in phone survey sample included collection of
segment & operational information during recruitment calls
Waste Definitions
11
Waste Type Examples
Equipment type is not high efficiency - incandescent instead of CFL; - standard instead of Energy Star; - regular furnace fan (not ECM); - regular showerheads - no faucet aerators
Equipment is left on or in standby modewhen not in useProgrammable thermostat not aligned withoccupancy hours Water temperate too highThermostat set points too low in summer Furnace fans always on, rather than auto
Maintenance HVAC tuned up regularly
Technological
Equipment Characteristics
Behavioral
Hours of Use
Performance/Temperature Settings
End Use Usage & Waste Definition
12
End Use Energy Consumption
kWh Technology Waste
kWh Behavioral
Waste
Hours
Wat
ts
kWh Minimum Usage
Actual Hours x (kWinstalled – kWeff)
Actual Run Time
Efficient Run Time
kW of installed equipment
kW of efficient equipment
(Hoursactual – Hourseff) x kWeff
Determine actual usage based on primary data: kWhactual = kW/Tontype X Tonsuser X EFLHuser
where: KW/Tontype = Power draw per ton of cooling, a function of SEER Tonsuser = User System capacity in tons EFLHuser = Equivalent full load hours
Analytic Example: Residential CAC
13
EFLHuser determined by CDD profile and equipment “design day”. Model user CDD for different time of day periods and occupancy conditions based on
customer behaviors (set points).
Analytic Example: Residential CAC
Sample Presentation Name 14
Time Set PointActual
CDD6am-9am 67 85.3
9am-12pm 67 201.312pm-4pm 67 368.54pm-7pm 67 235.6
7pm-10pm 67 131.210pm-6am 67 154.9
Total 1176.8
Time Set PointActual
CDD6am-9am 78.5 16.7
9am-12pm 82 41.212pm-4pm 82 117.64pm-7pm 76.5 135.8
7pm-10pm 76.5 59.110pm-6am 78.5 30.0
Total 400.4
Home A
Home B
EFLH = 320.3
EFLH = 941.5
Analytic Example: Residential CAC
Determine technology waste associated with three categories of potential efficiency upgrades that affect CAC usage: CAC unit efficiency Building shell Duct sealing
For each category waste calculated as: kWhwaste = kWhactual - kWhefficient
15
Analytic Example: Residential CAC
16
TimeOccu-pancy
Rec. Set Pt. Opt. CDD
6am-9am Home 78 19.29am-12pm Away 85 23.412pm-4pm Away 85 75.54pm-7pm Home 78 115.3
7pm-10pm Home 78 46.310pm-6am Asleep 82 12.1
Total 291.8
TimeOccu-pancy
Rec. Set Pt. Opt. CDD
6am-9am Home 78 19.29am-12pm Home 78 87.212pm-4pm Home 78 190.94pm-7pm Home 78 115.3
7pm-10pm Home 78 46.310pm-6am Asleep 82 12.1
Total 471.0
Home A Home B
http://www.energystar.gov/ia/partners/publications/pubdocs/HeatingCoolingGuide%20FINAL_9-4-09.pdf
EFLHopt behave= 376.8 EFLHopt behave= 233.4
Getting to Behavioral Waste
17
End Use Energy Consumption
kWh Technology Waste
kWh Behavioral
Waste kWh Minimum Usage
Efficient Technology Threshold
FLHuser FLHopt behav
Why This Matters
Benefits of the study: Improved understanding of current end use energy consumption – particularly behavioral drivers Measurement of the behavior savings potential by end-use and segment Enhanced primary data provides basis for other analyses and ability to address other research questions - stimulate new research objectives
18
Implications for Program Planning
Assess efficacy of technology and behavioral program options to optimize DSM investment
Identify and prioritize among “opportunity pockets” Customer segments or end uses where energy savings can be realized through behavioral program elements or messaging
Assess benefits of replacing widgets or attempting to change how widgets are used, or both
19
STRATEGIES FOR INCREASING ENERGY SMART ACTION IN
COMMERCIAL ORGANIZATIONS
Hannah Arnold
Equipment costs Availability of capital Budget constraints Lengthy internal decision-making processes The corporate approval process Lack of internal support for projects Inability to gain consensus among decision-makers Program awareness
STAFFING
Barriers to Energy Savings Actions
2 2012 BECC Conference
Staffing Grant Approach
2012 BECC Conference 3
RFP and Application
Staffing and Management
Changes
Complete EE Projects
Incentivizing Internal Resources
Pay Overtime to
current employees
Complete project
paperwork
Submit incentive
applications
2012 BECC Conference 4
Staffing Grant
Approval
Completed Motor and
Grocery Projects
Funding External Resources
Hire External
Consultants
Present findings to
internal stakeholders
Secure funding
2012 BECC Conference 5
Staffing Grant
Approval
Implemented Pump
Optimization
Overcoming Participation Barriers
6 2012 BECC Conference
Equipment Costs
Internal Decision-Making
Availability of Capital
Corporate Approval
Staff Resource Constraints
Thank you!
Hannah Arnold, Project Manager [email protected]
510-444-5050 Ext. 118
7 2012 BECC Conference
PUSH ME – PULL YOU
Positioning Emerging Technologies to Move Markets
Olivia Patterson
Mary Sutter
Jenn Mitchell-Jackson
Dimensions of Emerging
Technology Program Design
Dimension 1: Identify Placement in Technology
Development Continuum
Clearly defining a program’s
market position helps to:
1. Define a program’s role vis-à-
vis other market players;
2. Provide a framework for
tactics (such as development
assessment and introduction
support);
3. Guide selection of end-uses
and projects.
CA technologies:
- Advanced HVAC
- Smart Appliances and
Plug Loads
- Advanced Lighting
- Integrated Building
Designs and Operations
California ContextWhy Emerging Technologies?
Stages of Utility Intervention for Emerging Technologies
Dimension 2: Choose from a Suite of Tactics
Dimension 3: Select and Prioritize
End-Uses
Dimension 4: Collaborate to Identify Strategic
Market Position
Utility Emerging Technology Programs…
PUSH supply side
market actors to
provide innovations
and conduct targeted
R&D
PULL California end
users into adopting
technologies,
approaches &
practices
- Identify where the targeted technology is in
its development (in Stage 1 to 5) and
whether pushes or pulls are needed
- Utility Emerging Technology Programs are
well placed for Stages 2 to 4
Top-rated push tactics
in CA:
- Improving available
data on existing
systems’ performance
- Testing products in
advance of Codes and
Standards
development
Top-rated pull tactics
in CA:
- Verifying equipment
performance and
energy savings
- Increasing visibility of
energy efficient
technologies
LEGAL NOTICE
This poster was prepared as an account of work sponsored by the California Public Utilities Commission. It does not
necessarily represent the views of the Commission or any of its employees except to the extent, if any, that it has
formally been approved by the Commission at a public meeting. For information regarding any such action,
communicate directly with the Commission at 505 Van Ness Avenue, San Francisco, California 94102. Neither the
Commission nor the State of California, nor any officer, employee, or any of its contractors or subcontractors makes
any warrant, express or implied, or assumes any legal liability whatsoever for the contents of this document.
Policy Background Emerging
Technology
Program Goals
- All residential new
construction will be zero
net energy (ZNE) by
2020
- All commercial new
construction will be ZNE
by 2030
- Transform HVAC
market to ensure
optimal equipment
performance for CA
climate
Goal 1: Increase
adoption of energy
efficiency
measures
Goal 2: Increase
supply of energy
efficiency
technologies
Goal 3: Support CA
Strategic Plan and
related solutions
including zero net
energy
CA Long-Term
Strategic Plan lists
four “Big Bold”
Energy Efficiency
Strategies through
2030:
- All eligible low-income
customers have
opportunities to
participate in
efficiency programs by
2020Stage 1
Research and
Development
Stage 2
Development
Support
Stage 3
Assessment
Support
Stage 4
Deployment
Support
Stage 5
Energy Efficiency
Program Support
Bring
entrepreneurs
and investors
together (do not
need to conduct
own R&D)
Support for
specifications,
Codes and
Standards
Prove technical
validity in multiple
settings and for
integrated suite
of measures
Assess market
barriers, enhance
visibility, and
increase market
traction
Incentives for
selling and
distributing
technology; also
support through
outreach,
education, and
training programs
More Pushes More Pulls
Validate savings from new
technologies
Identify new energy efficiency
technologies
Provide knowledge about how
customers interact with new
technologies
Demonstrate to the
marketplace what works
As energy efficiency goals increase, Emerging Technology Programs help: