Don’t Get Showroomed- Are you frustrated with showrooming?
True AI from vapor: Retail ROI with deep learning for retail
1. TRUE AI FROM VAPOR:
REAL ROI WITH DEEP
LEARNING FOR RETAIL
2. 89%
OF MARKETERS
SAY THEY ARE
PERSONALIZING
EXPERIENCES
AND MESSAGES
ONLY 5%
OF CONSUMERS SAY MESSAGE
OFFERS ARE USUALLY
WELL-TIMED WITH THEIR NEEDS
Source: Forrester report Evolve Now To Personalization 2.0: Individualization; Forrester Data Consumer Technographics Global Online Benchmark Survey (Part 2), 2018
3. CUSTOMER MARKETING IS INCREDIBLY HARD
YOU HAVE
MILLIONS/BILLIONS
OF DATA POINTS
RELEVANCY
PERSONALIZATION
CUSTOMER EXPERIENCE
DATA-ACTION
GAP
6. IT’S HARD TO BUILD INTELLIGENT CAMPAIGNS
WHEN YOUR MARTECH STACK REQUIRES YOU
TO MICRO-MANAGE SEGMENTS & RULES
7. Xbox Fifa 19’
$59.99
HOLIDAY GIFTS
PURCHAS
E
HISTORY
CUSTOMER
SEGMENTS
VALUE
SIMILAR
PRODUCT
CATEGOR
Y
BROWSIN
G
HISTORY
PRODUCT
PAGE
VISITORS
TARGETING STILL HAS ROOM FOR IMPROVEMENT
16. DEEP LEARNING CAN BRIDGE THE DATA-ACTION
GAP
YOU HAVE
MILLIONS/BILLIONS
OF DATA POINTS
RELEVANCY
PERSONALIZATION
CUSTOMER EXPERIENCE
DATA-ACTION
GAP
17. Xbox Fifa 19’
$59.99
HOLIDAY GIFTS
PURCHAS
E
HISTORY
CUSTOMER
SEGMENTS
VALUE
SIMILAR
PRODUCT
CATEGOR
Y
BROWSIN
G
HISTORY
PRODUCT
PAGE
VISITORS
THE OLD WAY OF TARGETING
18. HOW TRADITIONAL TARGETING SEES PAT
WILL PAT BUY “FIFA 19” IN NOVEMBER?
Gender = "F"
Lives in the city
RFM Segment = "High-
value customer"
Never purchased in
this category
-0.5%
-4%
+1%
+3%
-2%
Email Address =
"Patricia.Jones@gmail.com"
Deep learning finds in your data that people with the email
structure “firstname.lastname@gmail.com” are more likely
to buy high-techor gaming items now.
Video games are not very popular in Hamilton, Montana
right now.
A Hoverboard. Not the same category, but activates "kid
stuff" latent feature in the Deep Learning.
The theme was “Holidaysshopping guide”. Deep Learning
detects that this link was aboutsomething involving the
"Christmas" and "gaming" latent feature.
Latent sociographics of high-end laptop users show they’re
slightly more likely to buy video games for their kids or
grandkids
All attributes are calibrated and interpreted using patterns unique to your own business
Zipcode = "59835"
Bought SKU “TS9085”
5 months ago
Clicked link #9 in your
newsletter 2 weeks ago
Browsed via Safari on latest
MacBook Air
+ HUNDREDS of other
"tiny clues"
+0.3%
-0.5%
+3.2%
+0.8%
+1.2%
+2.4%
Conclusion: VERY LIKELY TO BUY.
INCLUDE IN THE CAMPAIGN
Age = "61"
ATTRIBUTE IMPACT ATTRIBUTE IMPACTINTERPRETATION
Conclusion: UNLIKELY TO BUY.
DO NOT INCLUDE IN THE
HOW DEEP LEARNING SEES PAT
All attributes are calibrated and interpreted using patterns unique to your own business
All attributes are calibrated and interpreted using patterns unique to your own business
All attributes are calibrated and interpreted using patterns unique to your own business
All attributes are calibrated and interpreted using patterns unique to your own business
All attributes arecalibrated and interpretedusingpatternsuniquetoyourownbusiness
All attributes arecalibrated and interpretedusingpatternsuniquetoyourownbusiness
A chess game. Irrelevant.
Bought SKU “MC8790”
2 months ago
0%
21. If your solution requires you to input complex rules
in order to work, it’s not true AI.
22. If the UI gets more complex as you drill further into
your marketing strategy, it’s not true AI.
23. If your AI provider asks you to format your data or apply your
own logic to your data before integrating, it’s not true AI.
24. “AI-INSIDE” IS ABOUT ADDING 'YET-ANOTHER-
FEATURE’
“AI-FIRST” IS ABOUT RETHINKING THE ENTIRE UX
PRE-ENGINE ENGINE-
INSIDE
ENGINE-FIRST
25. SEAMLESS
TINYCLUES IN A FEW WORDS
SAAS
BUILT FOR
MARKETER
S
PROVEN
IMPACT
ACROSS ALL
CHANNELS
UP IN
2 WEEKS
ANONYMIZED
DATA
DEEP
LEARNING
26. BUILD THE BEST MARKETING PLAN FROM YOUR IDEAS & OBJECTIVES
“We need to be very strong
during the summer sales
period. Who’s the best
audience for our Facebook
Ads on Weber BBQs?”
“What’s the best audience
for push notifications on our
summer collection, in order
to drive revenue online and
in-store?”
“We have 22 campaigns
running next week; but no
one should get more than 3
messages, and we want to
distribute the messages
based on highest potential
to buy”
”Who should receive the
Samsung TV offer?
We need to target at least
1 million customers.”
“Our range of handbags is
niche AND strategic. We
want a very small target of
super buyers”
“Does it make sense to send
the following 2 campaigns on
the same day: ‘Dishwashers’
and ‘Recipe books’?”
27. RETAILERS LOVE TINYCLUES
-80%AVERAGE TIME
TO CREATE
CAMPAIGNS
-19% +51%AVERAGE
DECREASE
IN UNSUBS
AVERAGE INCREASE
IN ENGAGEMENT
+79%AVERAGE CAMPAIGN
REVENUE UPLIFT
28. LACOSTE BOOSTING CAMPAIGN REVENUE
FASHION / RETAIL
"We have more than doubled
our campaign revenue while
finally being able to sell our
strategic products ."
+151%
CAMPAIGN
REVENUE
MORE CAMPAIGNS
LOWER UNSUBSCRIBE
29. HOLLAND AND BARRETT DRIVING MESSAGE RELEVANCY & CAMPAIGN
REVENUE
RETAIL
-23%
SENT EMAILS
+19%
OPEN RATE
+27%
CAMPAIGN
REVENUE
"Since we have started using
Tinyclues, we have increased
our campaign revenue by 27%
while sending 23% less emails."
30. FNAC DARTY OPTIMIZING OMNICHANNEL CAMPAIGNS
RETAIL
"Tinyclues is at the very heart of our
CRM, enabling my team to target and
optimize more than 1500 campaigns
per year, driving millions of revenue per
year. The solution has a great impact
on our strategy and revenue”
+30%
CAMPAIGN REVENUE
+$8M
ADDITIONAL REVENUE PER YEAR
ON DIRECT MAIL CAMPAIGN
31. A LARGE AMERICAN JEWELRY RETAILER DRIVING CAMPAIGN
REVENUE
RETAIL
+$1M
IN STORE
CAMPAIGN REVENUE
32. • Evaluate AI solutions based on their UX as well as
their impact
• If you need to normalize data to use a solution, it’s
not a true AI solution.
• AI will not replace marketers. It will enable them to
do what they do best: focus on content and
strategy.
Actionable Takeaways from this session:
This is an example of an email from fictional company, Games R Us, promoting Fifa 19’, to target this campaign, a retailer might use purchase history…
So if you do all of this today, your marketing stack probably looks a bit like this: [show the slide with the messy, stressful-looking switchboard]
Every time you add more functionality, you add more complexity.
Every time you add more products to your stack, you increase the load on your team.
These products promise “simplicity” but they add complexity in order to give you that “simplicity.”
And let’s be clear: if you are segmenting your customers, you are using a switchboard. If you are operating from a messy excel sheet or its sleek technology equivalent, you are using a switchboard. If you are doing anything but inputting your strategic marketing goals and getting an output of actionable data/information, you are using a switchboard.
The future of marketing needs deep learning if it is to truly simplify the process for marketers, clarify customer intent without annoying/fatiguing them, and drastically improve performance / revenue / engagement.
What is Deep Learning? [Brief statement defining deep learning]
What are the Advantages? Deep Learning can leverage all of your marketing data so you can inform all of your marketing decisions.
You as a marketer have millions of data points and you’re trying to send relevant messages, with personalized content and trying to improve customer experience but there’s a clear disconnect between what you want to achieve, what you’re actually doing and what is perceived by your customers which makes customer marketing incredibly complex
This is an example of an email from Macy’s promoting Fifa 19’, to target this campaign, a retailer might use purchase history…
Now let’s see the difference between deep learning and data science in determining who is going to receive that Fifa 19’ email, we have here Pat…
So this is what we're fixing. We're using AI to create a completely new experience for marketers where they can simply, in a calendar view, say on Monday I have a BBQ campaign but have also three different campaigns. Some of them going on emails, some of them going on Facebook, and it will instantly, in a matter of minutes, create the optimal audiences and the best campaign plan based on the true data points that you have.
So you can at the same time send relevant messages that drives revenue. Finally, with Tinyclues, you can be Customer Centric & Business Centric.
What is Deep Learning? [Brief statement defining deep learning]
What are the Advantages? Deep Learning can leverage all of your marketing data so you can inform all of your marketing decisions.
This is adding complexity to your stack, your team, and your strategy. Does it pay off in revenue, customer satisfaction and engagement?
We’re here to share with you some tips we’ve learned along the way from working with retailers to target and plan their campaigns with deep learning
Let’s go into a very detailed example of what we are doing with Lacoste. Many people know Lacoste through their polos. I think they are probably the biggest company in terms of selling polos, I think they sell 2 items every second. The thing is that it’s not a sustainable business strategy for them to sell only polos. So Lacoste has created a very diversified list of products and it’s a big challenge for them to drive revenue on these new products categories, because if you start to sell handbags, and you have a database of millions of people who came in to buy polos, it’s basically impossible without destroying your brand to spam everyone, all the polo buyers, with a newsletter about handbags. What becomes possible with Tinyclues is to find the right audience for these handbags campaigns. In terms of retargeting, you could try to sell the handbags to the people who have interacted with handbags before but the thing is, it’s a tiny fraction of the database, so you have basically no reach. What is possible with Tinyclues is to find the lookalikes of the people who recently bought Lacoste handbags and to find the right 10% of the database that is going to be interested by this handbag even though they never knew that Lacoste had handbags. And this kind of campaign performs more than twice better than the generic campaigns with their prior approach of targeting their messages. So they have been very happy with the impact that they saw, first on the French market, and we recently signed with them to go globally on their remaining countries. They are a very global company. I think they serve more than 100 countries.
Let’s go into a very detailed example of what we are doing with Lacoste. Many people know Lacoste through their polos. I think they are probably the biggest company in terms of selling polos, I think they sell 2 items every second. The thing is that it’s not a sustainable business strategy for them to sell only polos. So Lacoste has created a very diversified list of products and it’s a big challenge for them to drive revenue on these new products categories, because if you start to sell handbags, and you have a database of millions of people who came in to buy polos, it’s basically impossible without destroying your brand to spam everyone, all the polo buyers, with a newsletter about handbags. What becomes possible with Tinyclues is to find the right audience for these handbags campaigns. In terms of retargeting, you could try to sell the handbags to the people who have interacted with handbags before but the thing is, it’s a tiny fraction of the database, so you have basically no reach. What is possible with Tinyclues is to find the lookalikes of the people who recently bought Lacoste handbags and to find the right 10% of the database that is going to be interested by this handbag even though they never knew that Lacoste had handbags. And this kind of campaign performs more than twice better than the generic campaigns with their prior approach of targeting their messages. So they have been very happy with the impact that they saw, first on the French market, and we recently signed with them to go globally on their remaining countries. They are a very global company. I think they serve more than 100 countries.
Let’s go into a very detailed example of what we are doing with Lacoste. Many people know Lacoste through their polos. I think they are probably the biggest company in terms of selling polos, I think they sell 2 items every second. The thing is that it’s not a sustainable business strategy for them to sell only polos. So Lacoste has created a very diversified list of products and it’s a big challenge for them to drive revenue on these new products categories, because if you start to sell handbags, and you have a database of millions of people who came in to buy polos, it’s basically impossible without destroying your brand to spam everyone, all the polo buyers, with a newsletter about handbags. What becomes possible with Tinyclues is to find the right audience for these handbags campaigns. In terms of retargeting, you could try to sell the handbags to the people who have interacted with handbags before but the thing is, it’s a tiny fraction of the database, so you have basically no reach. What is possible with Tinyclues is to find the lookalikes of the people who recently bought Lacoste handbags and to find the right 10% of the database that is going to be interested by this handbag even though they never knew that Lacoste had handbags. And this kind of campaign performs more than twice better than the generic campaigns with their prior approach of targeting their messages. So they have been very happy with the impact that they saw, first on the French market, and we recently signed with them to go globally on their remaining countries. They are a very global company. I think they serve more than 100 countries.
Let’s go into a very detailed example of what we are doing with Lacoste. Many people know Lacoste through their polos. I think they are probably the biggest company in terms of selling polos, I think they sell 2 items every second. The thing is that it’s not a sustainable business strategy for them to sell only polos. So Lacoste has created a very diversified list of products and it’s a big challenge for them to drive revenue on these new products categories, because if you start to sell handbags, and you have a database of millions of people who came in to buy polos, it’s basically impossible without destroying your brand to spam everyone, all the polo buyers, with a newsletter about handbags. What becomes possible with Tinyclues is to find the right audience for these handbags campaigns. In terms of retargeting, you could try to sell the handbags to the people who have interacted with handbags before but the thing is, it’s a tiny fraction of the database, so you have basically no reach. What is possible with Tinyclues is to find the lookalikes of the people who recently bought Lacoste handbags and to find the right 10% of the database that is going to be interested by this handbag even though they never knew that Lacoste had handbags. And this kind of campaign performs more than twice better than the generic campaigns with their prior approach of targeting their messages. So they have been very happy with the impact that they saw, first on the French market, and we recently signed with them to go globally on their remaining countries. They are a very global company. I think they serve more than 100 countries.