A joint project byRSPB and Plug the Gap
• Julie Pitt– Director at Plug the Gap– Database marketing and analysis for the charity sector
• Ruth Smyth– Supporter Insight Manager at the RSPB– Understanding support and supporters from across the
organisation
• Cath Campbell– Business Information Analyst at the RSPB– Deciphers the information and helps the planning
process
Our appeals were stuck in a rut and a couple
showed poor results. We wanted to know why.
We also wanted to grow, but weren’t sure how.
The RSPB asked Plug the Gap to collaborate on a project.The objective was to learn
more about the behaviours of donors
in cash appeals and the factors that affected those
behaviours.
• How many?4- 5 each year• What topics? Purchasing a reserve /
campaigns• Who to? Members, recently
expanded• How are they segmented? –Active
–Lapsed–Deep Lapsed
We started off with some basic concepts
Some things we knew.
Some things we knew we didn’t know.
Some things we didn’t knowwe didn’t know.
But we chucked in another one as well
Some things we thought we knew, but did we?
We pulled together lots of campaign reports, pieces
of analysis and some views and ideas of the
people that worked with appeals and data
Pulling them together and reviewing across
activitiesexpanded our understanding
We could identify what we knew and
we could identifywhere the holes were
And we knew there was a lot of stuff that
wedidn’t have any ideas
about
But most importantly we were able to see
that some of the things we thought we knew
just weren’t true
myt
hs!
We crossed referencedall of the information
from all of the different sources
This showed us which pieces of insight didn’t stand up to statistical
scrutiny,
...and which pieces of our
understanding were assumed
Armed with all of this information
we came up with a radical plan
Step 1:Create an RFV matrix
...but structure it so that it accurately reflects the
donors behaviour
Step 2:Analyse gift prompts
...but not in thetried and tested way that
Is normally seen
Step 3:Take the findings and
apply them to an appeal
...but make sure the results
are measurable
We looked closely atthe R the F and the V
We kept in mind that each variable had to accurately reflect the
behaviours of the supporters
Recency is easy:measure the time
between the last gift and another fixed point
Frequency is harder:Should this be the number of times a
supporter has given?
That’s OK but what if someone has been asked to give 10 times and donated twice are they the same as another supporter who has
been asked twice and donated twice?
There’s a subtle difference in
their behaviours:ask twice > get twice
Versus
ask ten times > get twice
The difference is their propensity to donate
which should put themin different cells in the
matrix
Value is the hardest:last, highest, average
We approached this by thinking of how we would
feel as supportersin a few different
scenarios
“I had a windfall and was able to donate a little
more. I normally give ten pounds but I was able to
give one hundred pounds.”
There are many reasons why supporters donate
an amount that falls outside of their normal
pattern of giving.
As a supporter I would feel “put upon” if my chosen
charity felt that I could now afford to give at a much
higher value than normal, just because of a single higher
value gift.
With this in mind we opted to use the mode
value: the value that best reflected the supporters’normal giving behaviour
It lets supporters knowthat we value their contribution regardless of how big or small.
We avoid making them feel that their contribution is never quite
enough by always pushing them for more.
Most supporters fell into only a few cells in the
matrix
Delving deeper it seemed that most
supporters gave ten pounds; a lot less than
previously thought
They gave ten pounds because they were
always asked for ten pounds so they gave ten
pounds...
...and on and on it went!
Which brings us very neatly to...
Were the gift prompts driving the donations or
were the donationsdriving the gift prompts?
We looked around atdifferent appeals; some
good, some bad
One appeal stood outand we needed to find
out why
Unlike other appeals the gift bands were
increased in number and decreased in value
spread
We also dissected the prompts and came up
with a way of showing the relationship between the previous gift value and
the prompts
We ended up with something that looked
like this...
This is the highest valuewithin the gift band
Gift prompt 1
Gift prompt 2
Gift prompt 3
Gift bands that were too wideconsistently failed to perform asWell as those where the band wasnarrower
Gift bands with unevenly spaced
prompts that were too far removed from
previous giving at both the top and
bottom didn’t perform as well as
those that were more evenly spaced and closer to previous
behaviour
GoodBad
The evidence seemed to suggest
that the prompts didn’t always determine the
value of the donation...
...but might determineif a supporter donated
We have no way of knowing why this might be, we can only hypothesise; prompts that are too far removed hark back to being pushy or unappreciative
of the contributionsof a supporter
When the 3 values are badly spaced it feels unnatural and a little inarticulate; if the top two are grouped it looks like we don’t really want you to
give at the lowest value
How do you bring all of these bits of information into an appeal so that it
sits together comfortably?
We pared down the information into
the key elementsthat we felt could be
applied
Firstly we worked out the modal value for each
supporter, removing the impact of large or small
gifts
Next we built a gift prompt matrix; with narrow value
bands we added 3 prompts and tested each one
using our visualisation tool to ensure it met the new spacing
criteria
We added an additional criteria;
each band should aim toincrease the normal giving
behaviourby a small amount
And we wanted to break supporters old habits and establish
a new one;the gift prompts would be held in place for the period of one years’
worth of appeals even if a supporter changed their normal
giving behaviour
This is the big question and the reason that
we’re all here tonight
I’m going to hand over to Ruth and Cath who will
talk you through the outcomes
First Appeal - report by Julie
Brainstorming sessionFurther analysis for subsequent appeals
Low Prom
pt
Medium
Prompt
High Prom
pt
Below low Low to mid Mid to high Above high
And what happened to theten pound donors?
£10£10£10
£10£12£15
66% of people moved!
Appeal1 Appeal 2
M
H
L
Low: StuckDonated to subsequent appeal at
the same prompt level as the previous appeal.
Appeal 1
M
H
L
Low to Mid: MoveDonated to subsequent appeal at the next prompt level up from the
previous appeal.
Appeal 2
Appeal 1
Appeal 2
M
H
L
Low to High: MoveDonated to subsequent appeal at two prompts up from the previous
appeal.
High to Mid: MoveDonated to subsequent
appeal at the next prompt level down from the previous
appeal.
Appeal 1
Appeal 2M
H
L
Popular gifts at five pound incrementsand use of the prompt values
Low Prompt Value £6 £12 £16 £21 £26 £32 £42 £48
Most popular donation above low prompt £10 £15 £20 £25 £30 £50 £50 £50
Most popular donation appears as other prompt? H M M M M No H No
Online donors behave differently
Appeal 1 Appeal 2 Appeal 3
Online DM Online DM Online DM
25 10 25 10 25 10
15 12 11 20 15 20
11 20 15 25 11 25
20 6 50 12 20 5
10 16 20 5 50 12
50 5 10 6 5 16
Analysis: Building from the basics
Taking a step back
Idea generation
Questioning assumption & myth busting!
Myth’s busted...
Average gift is twenty five pounds!
Most people give ten.
Land purchase appeals always do better!
Super appeals have a range of characteristics.
People paying for membership give less!
Channel is more important than fee.
Adding volume increases income!
Not necessarily. It may be better to decrease volume.
The RFV was based onreal behaviours so it helped
us to explore the implications
The project raised some big questions, which
we’ve started to answer
It also reminded us that it’s all about the relationship andnot just getting a one-off
donation