00:00Sous-titrage Société Radio-Canada
00:10Hi, good afternoon.
00:12We talked a lot about premium content in the past sessions.
00:18Let's talk about premium and local data.
00:21So, of course, you all know that with the deprecation of third-party cookies,
00:31everybody is looking for some valuable solutions,
00:34and AdSend Data really offers some solid solutions with database matching
00:39and first-party data of advertisers.
00:43We will have a deep dive this session into the power of first-party data
00:50and how data can be leveraged efficiently to optimize campaign performance.
00:56We will be talking about some interesting use cases, which we worked on,
01:01and these use cases really demonstrate the value and the relevance of database matching.
01:10Let's go and have a deep dive in data.
01:13So, what can you as an advertiser, which value can you create as an advertiser
01:21when you work with your own first-party data,
01:24and how can we as AdSend Data perform some really smart data analysis?
01:31So, of course, it all starts with our first-party data.
01:35We have a big data network with data coming from big brands and partners,
01:42such as MediaHealth, PlayMedia, Telenet, Proximus, and VRT.
01:47We have a lot of qualitative and local first-party data,
01:52approximately 3 million of unique data records.
01:57And so, thanks to the AdSend Data consent,
01:59we can combine the data of all our data partners
02:03and we are able to create enriched data profiles.
02:09And thanks to the AdSend Data ID,
02:12this makes it possible to integrate data coming from different platforms,
02:18so both TV and online.
02:22So, if we talk about database matching,
02:25there are three important building blocks,
02:28insights, activation of data, and measurement.
02:32And they can be used in standalone,
02:35but they also can be used in a full closed loop.
02:40So, that's the closed loop that you were talking about.
02:43So, let's go into the details.
02:45So, it all starts with the database matching
02:48of the databases of the advertiser
02:50and those of AdSend Data.
02:52Sorry.
02:53This is done through data cleanroom technology.
02:57So, this enables different companies
03:00to unlock their databases
03:03in a very secure manner
03:05within the data cleanroom.
03:07And this is done, of course,
03:10fully compliant with GDPR and privacy regulations.
03:15We work together with two data cleanroom partners,
03:19Infosum and LifeRamp.
03:21Based on the database matching,
03:25we can perform some really relevant insights.
03:31So, for example,
03:32we can give you some insights
03:34on the media consumption of your customer
03:38with AdSend Data,
03:39or even the behavioral
03:41and socio-demographic profile of your customer.
03:45Then, we proceed with creating the data segments
03:52on which the campaign has to be activated on.
03:56And we see here two important use cases.
04:00So, the first use case is targeting your own customers.
04:05So, based on the overlap data,
04:07you can, as an advertiser,
04:10upsell a new product to your existing customers.
04:15Sorry.
04:16But you can also activate inactive customers.
04:21And the second use case,
04:24we're not there yet.
04:25The second use case is targeting your prospects.
04:31So, based on the database matching,
04:33we create look-alike audiences.
04:36And this is done through AI.
04:37And so, we really create a digital twin
04:41of your best customers.
04:42So, which means that
04:43we look at those with AdSend Data
04:47who have similar characteristics
04:50and profiles
04:52and interests and behaviors
04:54as your customers.
04:56So, because of the...
04:58There is a high affinity
04:59between the look-alikes
05:01and your customers.
05:03The look-alikes are more likely
05:05to be interested by your products
05:07and are more likely
05:08to buy your products.
05:11So, this enables you,
05:13as an advertiser,
05:14to fully focus on your prospects
05:16while optionally excluding
05:19your own customers.
05:21And after the campaign,
05:23we link the viewing data
05:26coming from AdSend Data,
05:28the viewing data of the campaign,
05:30to the sales data
05:32of the advertiser
05:33or the retailer.
05:34and we perform
05:35a closed-loop measurement.
05:37So, let's go into
05:39the closed-loop measurement.
05:41So, here,
05:42we really measure
05:44the campaign impact
05:45on the sales results.
05:46So, we don't even
05:49measure the clicks.
05:50We go further
05:52into the funnel
05:52and really go
05:54to measure the conversion.
05:57So, we know
05:57that our media
05:59are very performant
06:01in the upper part
06:02of the funnel.
06:02so, for building
06:03brand awareness.
06:05But, until now,
06:06measuring of conversion
06:08was really a blind spot.
06:10And thanks to
06:10the closed-loop measurement,
06:11we are capable
06:12in going
06:13at the bottom
06:15of the marketing funnel
06:16and measure
06:17the conversion.
06:18So, of course,
06:23this is again
06:24done in the secure
06:27environment
06:28of the data
06:29cleanroom.
06:30So, we link
06:32the viewing data
06:33of the campaign
06:33to the sales data
06:35of the advertiser
06:36or the retailer.
06:37So, as you see,
06:39we need two types
06:40of data.
06:41The campaign data,
06:42which comes from
06:44AdSend Data,
06:45if it's an online campaign,
06:46or a BeVote campaign,
06:48or a CTV campaign.
06:50But,
06:50the viewing data
06:51can also come
06:52from the telco partners,
06:54Telenet and Proximus,
06:55if it's
06:56a linear TV campaign
06:58or even addressable
07:00TV campaign,
07:01which we want
07:02to measure.
07:02and the second type
07:04of data
07:05we need
07:06is purchase data
07:07or sales data.
07:08Ideally,
07:09as an advertiser,
07:10you dispose
07:11of your own
07:12sales data,
07:13but,
07:13for example,
07:14if you are a CPG brand
07:16or an FMCG brand,
07:18you don't dispose
07:19of all the sales data
07:20and we can rely
07:21on the sales data
07:22of the retailer.
07:24So,
07:24retail media,
07:26again,
07:26also a hot topic.
07:28And,
07:29so,
07:30we are announcing
07:31very soon
07:32in our partnership
07:32with media marketing
07:34Deleize,
07:35so,
07:36which is the retail
07:36media division
07:37of the Deleize group.
07:40And,
07:40here,
07:41it will be soon
07:42possible
07:43to,
07:44for brands
07:46which are sold
07:47at Deleize
07:47to activate
07:49online display campaigns,
07:51not yet video,
07:53so,
07:54only online display campaigns
07:55for now
07:56on our network.
07:57And,
07:58that,
07:58these campaigns
07:59are based
08:00on data
08:01of Deleize,
08:02which is,
08:03of course,
08:04based on real
08:05purchasing behavior
08:06with Deleize.
08:11So,
08:12what we do
08:13in this analysis,
08:14so,
08:14we create,
08:16so,
08:16we divide
08:17the buyers
08:17into two different groups,
08:19the target group
08:20on the one side
08:21and the control group
08:23on the other side.
08:24so,
08:24the targeted group
08:25is the target group
08:28on which the campaign
08:29has been activated.
08:30So,
08:31it consists both
08:32of viewers
08:32and non-viewers.
08:33And,
08:34the control group,
08:35on the other hand,
08:36has been set aside
08:37before activating
08:39the campaign.
08:40So,
08:40it's not targeted
08:41and it consists
08:43of non-viewers.
08:45So,
08:46and,
08:47in the closed-loop
08:48measurement,
08:48we will compare
08:49the performances
08:50of these two groups
08:52to see whether,
08:54what is the impact
08:55of the campaign.
08:57We can measure,
08:58for example,
08:58conversion
08:59and buy rate
09:00of the two different groups,
09:01but we also can measure
09:03other parameters.
09:05I don't have the time
09:06to go into
09:07all the use cases
09:09we worked on,
09:11but a use case
09:12we did together
09:13with Kineso
09:15and Media Brands,
09:16we had some
09:18really convincing results.
09:21There,
09:21we saw
09:22that the conversions
09:23of those
09:24who were exposed
09:25to the campaign
09:26was twice as high
09:27as those
09:28who didn't see
09:29the campaign.
09:30So,
09:30these results
09:31are really
09:31very convincing
09:32and makes us
09:35even go further.
09:36Let's talk
09:38about EVEF.
09:40EVEF is
09:41a fashion retailer
09:42and EVEF
09:44wanted to attract
09:45new clients
09:47and activate
09:48inactive ones.
09:50So,
09:51they launched
09:51a BVOT,
09:52a CTV campaign
09:53on our network
09:55and they added
09:57a call to action.
09:59And so,
09:59based on the database
10:00matching
10:01of the active clients,
10:03we created
10:03lookalikes
10:04and their
10:06we also
10:08activated
10:08the inactive
10:10clients
10:10to the campaign.
10:12And in the
10:12closed loop
10:13analysis
10:13afterwards,
10:15we saw
10:16that EVEF
10:17was able
10:18to reach
10:19its objectives.
10:20so,
10:21we had,
10:22there were
10:23851
10:24customers
10:25who had
10:27bought,
10:28who made
10:29a,
10:29who,
10:30yeah,
10:30who made
10:31a,
10:32who bought
10:33at EVEF
10:34mode
10:34after seeing
10:35the campaign
10:36on our network.
10:38So,
10:38these clients
10:39were either
10:40completely new
10:41clients
10:42or clients
10:43or clients
10:43who were
10:44inactive
10:44and who
10:45were
10:45activated
10:46by buying,
10:48by doing
10:49a purchase
10:49with EVEF.
10:51And what we
10:52saw is that
10:53the majority
10:54of those
10:55customers
10:56came from
10:57the lookalikes
10:58that we have
10:58created.
11:00So,
11:00these were
11:01completely
11:01new clients
11:03for EVEF.
11:04and the
11:06analysis
11:07also gave
11:09some really
11:09interesting
11:10insights.
11:11First of all,
11:12we saw that
11:12repeated exposure
11:14led to
11:15higher conversion.
11:17So,
11:17if you want
11:17to attract
11:18new clients,
11:19you really
11:19need to have
11:20more repetition,
11:22more frequency,
11:23more OTS.
11:24And we also
11:25saw a big
11:26correlation
11:26with the
11:28local shoppers.
11:29And local shoppers
11:29is a basic
11:30targeting segment
11:31that we offer
11:32and it's based
11:33on,
11:34surfers who
11:35read a lot
11:35about buying
11:39locally
11:39and about
11:40Belgian brands.
11:41And there
11:41we saw
11:42that the
11:43conversion of
11:44those who
11:44had been
11:45exposed to
11:46the campaign
11:46was twice
11:48as high
11:48as those
11:49who hadn't
11:50been exposed
11:51to the
11:51campaign.
11:52And of
11:53course,
11:53with closed
11:54loop measurement,
11:54we are able
11:55to measure
11:56the online
11:57conversion
11:58on the
11:59webshop,
12:00but also
12:00the offline
12:01conversion,
12:02so what
12:02it has,
12:03what
12:03result it
12:04has in
12:05the physical
12:05shops.
12:07And of
12:07course,
12:08just something
12:10we found
12:11out also
12:12with other
12:13use cases,
12:14the lookalikes
12:15we had
12:16created were
12:16really very
12:17performant,
12:19which means
12:19that the
12:21actual buyers
12:22closely matched
12:24the lookalikes
12:25that we had
12:25created,
12:26and so
12:27they are
12:28really having
12:29strong predictive
12:30power and
12:31they really
12:31can anticipate
12:32buying behavior.
12:34And the
12:34second conclusion
12:36of the
12:36lookalikes,
12:37of course,
12:37is that the
12:37majority of
12:39the new
12:40clients were
12:41attracted by
12:42the lookalikes
12:43that we had
12:43created.
12:46So we're
12:47almost at the
12:48end of the
12:48presentation.
12:50Let's
12:50summarize the
12:51key conclusions.
12:51so database
12:53matching is
12:54based on
12:54three important
12:55building blocks,
12:56insights,
12:57activation of
12:58data,
12:59and measurement.
13:00It can be
13:01used in a
13:03full closed
13:04loop,
13:05and closed
13:05loop enables
13:06us really to
13:07measure the
13:08conversion and
13:09the sales
13:10impact.
13:11So this is
13:11really very
13:12innovative.
13:15And of
13:16course,
13:16what did we
13:17learn about
13:18the use cases
13:19we did?
13:20So these
13:20lookalikes are
13:21really having
13:23a close
13:23match with
13:24the actual
13:25buyers.
13:26So that was
13:27one conclusion.
13:28The second
13:28conclusion is
13:29that we have
13:30really
13:31performant
13:32lookalikes,
13:34so they
13:34really generate
13:35new customers
13:36for you.
13:37And of
13:38course,
13:38we saw that
13:39first-party
13:40data in
13:40combination with
13:41campaign exposure
13:43led to the
13:45doubling of
13:45conversion.
13:47And the
13:48last slide,
13:49we are
13:50organizing a
13:51legal webinar
13:52on the
13:5218th of
13:53November,
13:53because we
13:54all know
13:54that when
13:56we want to
13:56set up a
13:58successful
13:59database
13:59cooperation,
14:01we need
14:02different teams
14:03to communicate
14:04with each
14:05other,
14:05the marketing
14:06team,
14:07the technical
14:07team,
14:08and the
14:08legal team.
14:09and it
14:11might seem
14:12the legal
14:12aspects
14:13might seem
14:14a little
14:14bit
14:14complex,
14:16but it
14:17doesn't have
14:18to be an
14:18obstacle,
14:19because we
14:19really have
14:20a lot of
14:20DPO
14:20expertise with
14:21Edson Data,
14:22and that's
14:22why we are
14:23organizing this
14:24webinar.
14:25so if you
14:26want to
14:27invite your
14:28legal or
14:29DPO team,
14:30so please
14:31feel free to
14:32do that,
14:32because it
14:34will be
14:34really
14:35interesting.
14:36I won't be
14:36giving the
14:37webinar,
14:38it will be
14:38my DPO
14:39colleague,
14:40Bart van
14:41Holdenhoven.
14:42So,
14:43yeah,
14:44I think we
14:44are at the
14:45end of the
14:45presentation,
14:46within the
14:47timing.
14:47Thank you.
14:47Thank you.
14:47Thank you.
14:48Thank you.
14:49Thank you.
14:49Thank you.
14:50Thank you.
14:51Thank you.
14:52Thank you.
14:53Thank you.
14:54Thank you.
15:02Thank you.
15:03Thank you.
Commentaires