This is an example of a Weather Impact Assessment performed for a national food ordering company by WeatherAlpha, the leading provider of business weather evaluations, which turn your data into instantly actionable intelligence.
3. IMPACT OF WEATHER ON FOOD DELIVERY
COMPANY
» The client tasked WeatherAlpha with evaluating the influence of weather on
orders and revenue in Madison, WI and Philadelphia, PA
» Project Objectives:
•
•
•
Complete a statistical analysis of sampled data
Provide Client with a full understanding of which weather conditions increase or decrease online orders /
revenue in Madison, WI and Philadelphia, PA
Explore weather-based digital advertising opportunities
» Data:
•
•
•
Sample size includes dates between 01/01/2005 – 12/31/2010
Hourly order data was compared with corresponding hourly weather data
For some weather conditions, a comparison of aggregate daily order data with aggregate daily weather data
was preferred
» Assumptions:
•
•
Weather at nearest weather station is representative of the weather at each city
Client revenue was calculated using 4% of the order total*
* 4% established based on a 2009 interview with Client CEO.
WeatherAlpha | Proprietary & Confidential
3
4. IMPACT OF WEATHER ON FOOD DELIVERY
COMPANY
» To complete this task, WeatherAlpha used a
three-phased approach to conduct a
comprehensive weather impact assessment
1
Develop Baseline
Metrics
Phases
Rationale
Key
Deliverables
2
3
Conduct Order
Analysis
Define Weather
Based Marketing
Opportunities
Develop understanding of order
and sales trends to determine
seasonal, weekly, and diurnal
patterns in data
Identify weather conditions that
increase/decrease online orders
Determine if there are other
variables that work with weather
Identify when and how Client can
boost orders
None
Preliminary Observations
Weather-Driven Business Analysis
WeatherAlpha | Proprietary & Confidential
4
5. FIRST PHASE:
IDENTIFICATION OF BASELINE METRICS
» Baseline Metric 1 - Orders by Month
16
Philadelphia
Madison
4
14
3.5
12
3
10
2.5
8
2
6
1.5
4
1
2
0.5
0
0
Jan Feb Mar Apr May Jun
Jul
Aug Sep Oct Nov Dec
Jan Feb Mar Apr May Jun
Jul
Aug Sep Oct Nov Dec
KEY OBSERVATIONS
Both locations exhibit similar monthly fluctuations in order volume
Two distinct periods of greater sales, Feb-Apr and Oct-Dec, coincide with the heart of the spring and fall semesters.
Jun-Aug is relatively quiet at both locations as many students leave for the summer.
Madison shows greater annual variability, with its biggest sales month (Dec) being 151% more than its smallest sales month
(Aug). Conversely, the biggest sales month in Philadelphia (Feb) is 68% greater than its smallest sales month (Jul).
WeatherAlpha | Proprietary & Confidential
5
6. FIRST PHASE:
IDENTIFICATION OF BASELINE METRICS
» Baseline Metric 2 - Orders by Day of Week
Philadelphia
Madison
3
10
2.5
Hourly Orders
3.5
12
Hourly Orders
14
8
6
2
1.5
4
1
2
0.5
0
0
Mon
Tue
Wed
Thu
Fri
Sat
Sun
Mon
Tue
Wed
Thu
Fri
Sat
KEY OBSERVATIONS
At both locations, there is little variability in order volume based on day of the week.
Madison’s biggest sales day (Wed) is 6% greater than its smallest sales day (Sat).
Philadelphia’s biggest sales day (Sun) is 5% greater than its smallest sales day (Thurs).
WeatherAlpha | Proprietary & Confidential
6
Sun
7. FIRST PHASE:
IDENTIFICATION OF BASELINE METRICS
» Baseline Metric 3: Orders by Hour of Day
Philadelphia
Madison
40
12
10
Hourly Orders
30
25
20
15
10
8
6
4
2
0
0
12:00 AM
1:00 AM
2:00 AM
3:00 AM
4:00 AM
5:00 AM
6:00 AM
7:00 AM
8:00 AM
9:00 AM
10:00 AM
11:00 AM
12:00 PM
1:00 PM
2:00 PM
3:00 PM
4:00 PM
5:00 PM
6:00 PM
7:00 PM
8:00 PM
9:00 PM
10:00 PM
11:00 PM
5
12:00 AM
1:00 AM
2:00 AM
3:00 AM
4:00 AM
5:00 AM
6:00 AM
7:00 AM
8:00 AM
9:00 AM
10:00 AM
11:00 AM
12:00 PM
1:00 PM
2:00 PM
3:00 PM
4:00 PM
5:00 PM
6:00 PM
7:00 PM
8:00 PM
9:00 PM
10:00 PM
11:00 PM
Hourly Orders
35
KEY OBSERVATIONS
Both locations exhibit similar diurnal fluctuations in order volume
At both locations, orders peak around the dinnertime hours of 6-8 p.m.
At both locations, there is a secondary lunchtime spike in orders between 12-2 p.m.
At both locations, orders are negligible during the early morning hours of 4-9 a.m.
Madison has a larger late night customer base than Philadelphia.
WeatherAlpha | Proprietary & Confidential
7
8. SECOND PHASE:
WEATHER IMPACT REPORT CARD
»
In the second phase,
WeatherAlpha completed
analysis of 14 weather conditions
in both geographies over a 6 year
period to develop a Weather
Impact Report Card.
Weather Conditions
Daily Rainfall
Precipitation
effects
Hourly Rainfall Intensity
Daily Snowfall
Hourly Snow/Ice Intensity
Temperature
Temperature/humidity
effects
Temperature Anomaly
Dewpoint
Humidity
Wind
Cloud Cover
Other notable
weather effects
Forecast/Prior Effects
Past/Lagged Effects
Snow Cover
WeatherAlpha | Proprietary & Confidential
8
9. THIRD PHASE:
MARKETING STRATEGIES
»
In the third phase, WeatherAlpha developed weather-based marketing strategies for Client
to create new lines of revenue
Removed due to client
confidentiality
WeatherAlpha | Proprietary & Confidential
9
11. OF THE 14 WEATHER CONDITIONS WEATHERALPHA ANALYZED,
THE FOLLOWING 8 WERE SELECTED AS HAVING THE GREATEST
IMPACT ON CLIENT REVENUE
» Weather Impact Report Card
Weather Condition
Madison, WI
Philadelphia, PA
Hourly Rainfall Intensity
Hourly Snow/Ice Intensity
Temperature Anomaly
Dew Point
Wind
Cloud Cover
Forecast/Past Effects
Snow Cover
High Impact
( >25% )
Moderate Impact
( 10-25% )
Low Impact
( <10% )
Note: For nearly every weather condition analyzed, Madison was significantly more weather sensitive than Philadelphia.
WeatherAlpha | Proprietary & Confidential
11
12. CONDITION 1: HOURLY RAINFALL INTENSITY
Philadelphia
Madison
270
Hourly spend ($)
Hourly spend ($)
250
230
210
190
170
150
None
5
10
15
20
Hourly rainfall (hundreths of inches)
25
80
75
70
65
60
55
50
45
40
35
30
None
5
10
15
20
25
Hourly rainfall (hundreths of inches)
KEY OBSERVATIONS
Rainfall is associated with increased order volume and hourly spend on both locations.
In Madison, hours with rain have 25% more hourly order spend than dry hours. Increasing intensity generally magnifies this
effect, up to a 35% increase.
In a given year, rainfall in Madison results in $2000-$2500 additional revenue for Client over dry hours*.
In Philadelphia, rainfall results in a 8-20% increase in hourly spend, determined by intensity. This results in an extra yearly
revenue of $400-$500 over dry hours*.
* Revenue estimates based on sales increase, frequency of rainfall, and 4% figure cited previously
WeatherAlpha | Proprietary & Confidential
12
13. CONDITION 2: HOURLY SNOW/ICE INTENSITY
Madison
6
18
5.5
16
5
Hourly Orders
Hourly Orders
Philadelphia
14
12
10
4.5
4
3.5
3
8
2.5
6
Light
Moderate
Heavy
2
Light
Moderate
Heavy
KEY OBSERVATIONS
Frozen precipitation is associated with increased order volume and hourly spend for both locations.
In Madison, frozen precipitation results in a 25-38% increase in order volume, though the trend of increasing intensity is
inconclusive.
For Madison, falling snow/ice is estimated to account for $600-1000 in extra yearly revenue over dry hours.
In Philadelphia, frozen precipitation results in a 28-50% increase in order volume. Generally speaking, order volume
increases with greater precipitation intensity.
For Philadelphia, falling snow/ice is estimated to account for $200-300 in extra yearly revenue over dry hours.
WeatherAlpha | Proprietary & Confidential
13
14. CONDITION 3A:
TEMPERATURE ANOMALY WARM DAYS
Philadelphia
Madison
72
220
71
Daily Orders
73
230
Daily Orders
240
210
200
190
180
70
69
68
67
170
66
160
65
150
5
6
7
8
9
10
11
12
13
Degrees above normal (F)
14
15
5
6
7
8
9
10
11
12
13
14
Degrees Above Normal (F)
KEY OBSERVATIONS
Frozen precipitation is associated with increased order volume and hourly spend for both locations.
In Madison, frozen precipitation results in a 25-38% increase in order volume, though the trend of increasing intensity is
inconclusive.
For Madison, falling snow/ice is estimated to account for $600-1000 in extra yearly revenue over dry hours.
In Philadelphia, frozen precipitation results in a 28-50% increase in order volume. Generally speaking, order volume
increases with greater precipitation intensity.
For Philadelphia, falling snow/ice is estimated to account for $200-300 in extra yearly revenue over dry hours.
WeatherAlpha | Proprietary & Confidential
14
15
15. CONDITION 3B:
TEMPERATURE ANOMALY COLD DAYS
Philadelphia
Madison
84
390
82
Daily Orders
Daily Orders
360
330
80
78
76
300
74
72
270
-5
-6
-7
-8
-9
-10
-11
-12
Degrees Below Normal (F)
-13
-14
-15
-5
-6
-7
-8
-9
-10
-11
-12
-13
-14
-15
Degrees Below Normal (F)
KEY OBSERVATIONS
Abnormally cold weather is associated with greater daily order volume on both locations.
In Madison, increasing the magnitude of the cold relative to normal results in incremental increases in daily order volume.
A very cold day may have up to 20% more orders vs. a typically cold day and up to 30-40% more than a day with seasonable
temperatures.
A very cold day in Madison may see ~ $75 more revenue compared to a normal day.
In Philadelphia, cold days see greater order volume than warm days, though the trend of increasing cold anomaly is
inconclusive.
WeatherAlpha | Proprietary & Confidential
15
16. CONDITION 4: HIGH DEWPOINT
8.5
Philadelphia
Madison
2.4
7.5
2.2
Hourly Orders
Hourly Orders
8
7
6.5
6
5.5
2
1.8
1.6
1.4
1.2
5
1
40
45
50
55
60
65
45
50
55
60
65
70
Dewpoint (F)
KEY OBSERVATIONS
High dewpoints in Madison are associated with increases in hourly order volume, especially above 60 ºF. At very high
dewpoints, hourly orders are as much as 30% greater than drier hours. A very moist day may be responsible for up to a $50
increase in daily revenue.
In Philadelphia, high dewpoints cause a decrease in hourly order volume, the opposite effect of that in Madison.
WeatherAlpha | Proprietary & Confidential
16
17. CONDITION 5: WIND & RAIN
18
Madison
4
Philadelphia
16
Hourly Orders
Hourly Orders
3.5
14
12
3
2.5
10
8
5
10
15
20
2
5
10
15
20
KEY OBSERVATIONS
At both locations, increasing wind speed during wet hours results in a corresponding increase in order volume. This effect
(though not shown here) is also observed when the weather is dry.
The positive effect of rain on Madison and Philadelphia order volume are accentuated by increasing wind. For instance, a
rainy hour with strong winds in Madison will have 20-25% more revenue than a rainy hour with no wind and a 50% increase
over a dry hour.
WeatherAlpha | Proprietary & Confidential
17
18. CONDITION 6: CLOUD COVER
Philadelphia
Madison
76
290
74
270
72
Daily Orders
78
310
Daily Orders
330
250
230
210
70
68
66
190
64
170
62
150
60
1
2
3
4
5
6
7
8
Fractional Sky Cover (tenths)
9
10
1
2
3
4
5
6
7
8
9
10
Fractional Sky Cover (tenths)
KEY OBSERVATIONS
Increases in average daily cloud cover have a generally positive impact on daily order volume for both locations, though
the trend is not entirely definitive.
At both Madison and Philadelphia, the highest amount of daily orders occurs on days with nearly maximum cloud cover.
Interestingly, at both locations, days that are nearly free of cloud cover have greater daily orders than partly cloudy days
It is difficult to attribute any revenue gain or loss to cloud cover alone, as this variable is directly related to likelihood of
precipitation.
WeatherAlpha | Proprietary & Confidential
18
19. CONDITION 7: FORECAST/PAST EFFECTS
Philadelphia
120
100
90
80
70
60
50
40
30
20
10
0
100
No Future Precip
Future Precip
Hourly Spend ($)
Hourly spend ($)
Madison
80
No Past Precip
60
Light Past Precip
40
Moderate Past Precip
20
1
2
3
4
5
6
Hours prior to event
0
1
2
3
4
5
6
Hours after event
KEY OBSERVATIONS
There were increases in hourly spend prior to the onset of precipitation as well as after it had departed. Though
Philadelphia is displayed here, a similar impact was observed in Madison.
For 3-4 hours leading up to a precipitation event, there is a spike in hourly order volume. This is likely the result of decisions
based on weather forecast information.
For 3 hours after a precipitation event had ended, hourly order volume remains elevated, likely the effect of damp or snowy
ground and continued cloud cover. Moreover, greater precipitation intensity is associated with a more pronounced uptick.
This observation - of forecast and past effects on sales beyond the time frame of the event itself - is important for a few
reasons. Firstly, it extends the period of revenue gain for Client. Also, it argues for the use of forecast and past weather
triggers for marketing optimization.
WeatherAlpha | Proprietary & Confidential
19
20. CONDITION 8: SNOW COVER
Philadelphia
Madison
450
100
95
400
Daily Orders
Daily Orders
90
350
300
250
85
80
75
70
65
200
60
150
None
1" +
3" +
6" +
10" +
None
1" +
3" +
6" +
10" +
KEY OBSERVATIONS
The existence of snow cover at both locations results in increased daily order volume.
In Madison, increasing snow cover is associated with incremental increases in daily orders. Days with 6” or more of snow
see a 57% uptick in orders, while 10”+ of snow cover results in an 84% gain.
This means that in Madison, having a foot of snow on the ground will translate into an additional $200/day in revenue over
a winter day with no snow cover!
In Philadelphia, snow cover is associated with increases in daily orders, though the trend is not as pronounced as in
Madison. Here, 10” or more of snow will result in an extra $30/day of revenue.
WeatherAlpha | Proprietary & Confidential
20
22. CLIENT CAN LEVERAGE THE UPCOMING WEATHER TO CREATE
NEW REVENUE STREAMS VIA WEATHER-BASED MARKETING
CAMPAIGNS
Removed due to client
confidentiality
WeatherAlpha | Proprietary & Confidential
22
24. WEATHERALPHA TEAM
Daniel Alexander
Co-Founder, Chief Meteorologist + Data Scientist
dan@weatheralpha.com
Mobile: (203) 241-4253
Jason Chen
Co-Founder, Chief Strategy + Operations Officer
jason@weatheralpha.com
Mobile: (617) 955-0759
Our Social Responsibility.
We believe that giving back to the
community should be a real
priority.
Each year, up to 5% of
WeatherAlpha’s profit will be
donated to weather disaster and
relief funds around the world.
We value our planet and respect
Mother Nature. We don’t wish to
use Mother Nature for our sole
benefit – we wish to help our
clients better understand and
value weather, and in doing so we
will also help communities that
have been shattered by weather
events.
Brooke Cunningham
Chief Alliances Officer
brooke@weatheralpha.com
Mobile: (347) 556-9613
WeatherAlpha | Proprietary & Confidential
24