5. “The ability to use algorithms to
automatically determine the likelihood that
something will happen in the future, for
instance likelihood to make a purchase.”
Predictive marketers are 1.8
times more likely to consistently
exceed shared organizational
goals (Forbes, 2018)
6. Innovation
Trigger
Peak of
Inflated
Expectations
Source: Gartner
Trough of
Disillusionment
Slope of
Enlightenment
Plateau
of Productivity
Time
Expectations
Mobile BI
Social Analytics
Mobile App Analytics
Predictive Analytics
Real-time Analytics
Conversational
Chatbots
Artificial
General
Intelligence Key
Less than 2 years
2 to 5 years
5 to 10 years
More than 10 years
89% of marketers have predictive analytics on
their roadmap for 2016.
(Forbes, 2016)
93% of consumer-facing businesses are unable
to use predictive analytics (SAP, 2018)
State of Disillusion… Does the market even know?
11. - Welcome Program
- Registered Not Purchased
- Abandon Category
- Abandon Product
- Abandon Search
- Abandon Basket
- Abandon Checkout
- Back in Stock
- Replenishment Campaign
- Post - purchase campaign
- First to second purchase
- VIP Series
- Anniversary
- Birthday
- Multi – purchase Campaign
- Lapsing and Lapsed
12. 12
Prospect Conversion
Campaign
Put an automated series
in place to target
customers they have
signed up
Use the typical time lag
between signing
up to the brand and
making a first purchase
12
Talking to your customers:
Why purchase with our brand?
Personalise with content they
are interested in
Channels
13. Limitations
Not sure which
customers are
going to purchase
Therefore not sure
of what content to
send them
Inefficient
channel
spend
Using Predictive Technology
Vary content based
on a customer with a
high or low propensity
Improve spend
efficiency and
resources
Send them content
that is relevant
to them
14. High Likelihood to
Convert In Next 30 Days
Low Likelihood to
Convert In Next 30 Days
USP and brand Automated channels
with lower spend and
inspire purchase
Offer and
incentive driven
More channels
to convert
1414
15. Converting Prospect Customers
High Likelihood to
Convert In Next 30 Days
Low Likelihood to
Convert In Next 30 Days
Create audience
exclusion for
paid social
(Google Match)
and search
adverts (Custom
Audience)
Get Inspired this fall.
Inspiration content
with heavily focus
on customer
testimonials,
delivery and returns
Time sensitive offer email
supported with count down
clock
Personalised Site Banner:
Promote offer or reinforce
delivery or returns option to
increase likelihood of purchase
Social extract: Facebook or
Instagram advert. New customer
offer this weekend only
Follow up email
campaign with
closing sale
message. SMS
used to support
non openers of
first or final
campaign.
Interactive email
with social scrape of
latest Instagram
content to promote
purchase from
trending content
Customers Scored By Model
Final email with
new ranges just
for you –
Recommend
content
1515
17. 17
Acquire More Then We Churn Create Brand LoyaltyGrow Customer Value
How Do You Measure Success?
.
• What is the uplift in VIP customers?
• Control test
• Have re-reduced our overall
customer churn?
• Have we reduced the cost to
acquire a customer?
• Have we created increased lifetime
value across our database?
Hi I’m Roisin form RedEye
We help brands increase their customer LTV with AI driven database technologyJust to be clear I am not sales person - I am not going to sell you on what predictiveAlso I am not an analysts - I am not going to talk to weather the models are built on R or Python
What I do is help brands intrepid their data and what want to do is talk through the common questions I get asked
How to use predictive
When to use predictive
And how to measure success
Be Practical – Not what it is
I am not a sales person I work with brands everyday to implement the right solution for them objectives.
How to use Predictive
When to prioritise it
How to Measure success
Lets bring it back you and what you are trying to achieve?
In my experience regardless of industry sector it boils down to:
- acquire the right customer, and ensure we retain more then we churn
- give rich customer experience – to grow value - and we wan to build a loyal customer base
We all know that key driver for success is data
But for once I am not talking about building that SVC and what are the richest dataset
We are focusing on the accessibility and actionability of that data
Key areas we focus on to optimise the customer experience is - Automation - Segmentation - drive 18 time more revenue - Personalisation - out sell competitors by 30%
These have been the bread and butter for years and evolved over time.
Now we are throwing predictive analytics into the mix!!The questions I get asked is:How do I use this?A common theme I find when it come to planning to use predictive we try and overly complicate already complex solution.
This is not surprising for years we have had to hand pull data form multiple sources, overly an analytics tool, try and drive insights form this and drive this into a creative marketing plan
My recommendation is“Let the product do what it is designed to do”
We need to stop trying to control everything.I will hold my hands up I can be an control freak – I was on honeymoon just last week and I was checking work email every chance I could.I am sure you can imagine how well that was down.
We need to trust the algorithms, they are taking into consideration100’s of thousand customer touch points to determine the what the customer will do next
Not sure did anyone get the chance to see our CCO’s presentation this morning?Basically we are now at the point where we can confidently sat these work
I hold my hand ups I was shocked by the results - we are reducing customer churn by 54%, increasing LTV by 34%
I don’t know about you guys but I am bit of geek I love reading about our technology and data what happening.
I am sure everyone has seen the Top Trends in the Gartner Hype cycle – about emerging technology
Its we don’t start trusting the data and making this real risk in another 2 years we will be conflicting on what we had planned to do.
There are lots of different predictive models in the market . . .
But we have looked at the foundation of the marketing strategy
improve performance at each stage of the customer lifecycle with the overall goal of improving customer value!
Our approach is not about using this as a separate identity but use presitive to drive what you are currently to maximise the value of your marketing mix.
This is a typical customer lifecycle, were we are drive customer through the different stages.Some brands have a top 11 campaigns live, some have over 400 targeting customer as they move through stage.
Let not under estimate the value this can bring.
Brands see 25-30 online being generating form these automation comms
But there is still revenue opportunity that we can start to take advantage of.
We have 6 Models5 – enable to improve and drive actions1 – allows us to calculate LTV and predictive what your LTV will be in 12 months for frocasting.Don’t worry I am only looking at 1 today.Driving your prospects to make their first purchase.
Why
We spend so much time and effort acquiring these visitors
To end up with 44% of the overall database who have never purchased
There is so much opportunity for you as a brand.
What are you currently doing to convert prospects
Running analysis to see the lag time
Then applying this into an automated series
You should be able to talk to them based on what they are interested
Typically we see brands use 6 channels at this stage
We are already doing a lot to convert these customer what are out limitations.WE don’t know who is likely to purchaseThere not spending right channels
Not commutating with the relevant message
By apply predictive this should reduce this risk within your targeting approach.
What happens:
- algorithms look at 1000’s of variable - Scoring prospects - Automatically builds threshold - based on your seasonal trends, customers buying habits - Define the time frame you have to convert - create a persona of those customer so you know what to say to them.
So we have our segmentWe have the information about the customerWe know what your objective isNow we just use it.At this point I would always let your prospects go through your welcome and conversion campaigns then use predictive to drive additional revenue. High - BAU
- they are going to convert use that segment as an exclusion to manage spend in your paid activity. - Can we inspire to spend more by scraping through Instagram content on key trendsLow - your are probability asking yourself why bother - if we can convert an additional 10% what would this mean to your brand.
- use your key marketing message creating urgency, if appreciate try incentivicing these people instead of everyone on the database, maximise your channel reach.
Most important thing I would say is TEST.Measure these against a control cell.When we don’t this for footasylm we seen increase against the control of 27%
Renton rate
LTV
Control
If I can leave you with points:It does not need to be complicated – let the Modelling do what it is designed to doTest what works – model give you what you need to miked decisions – perfect blend between technology and human interationsUse LTV are Key KPI Measure against control cells.