Audience Management In Sift

Sift Audience Management through Behavioural Clusters and Relationships


Contextual Marketing has risen to an all new level with the real time propellant to it. The Real Time-ness of the context is providing businesses very high values and much deeper insights, which helps them to design and strategise their marketing plans in a very different way. This also means that the marketing strategies needs to be very agile in order to catch up to the changing contexts with time. Real time analysis can play a much important role in analysing different customers and their behaviours with time but performing this analysis on the customer behaviours in real time is very daunting task, as behavioural patterns tend to change with time and the numbers one is looking for can be very unpredictable. That is why, one of the major challenges in real time contextual marketing is Audience Management. Customers with different behavioural patterns fall into different categories or we can say in different clusters. And with time the counts of customers falling in a particular cluster changes very rapidly. In order to explore the accurate audience cluster to contact or target, a system needs to capture the behavioural patterns in real time and keeps on updating itself on any change encountered in the behaviour. Real Time analysis of these clusters can enhance the business potential for marketing and overwhelming customer satisfaction.

Sift’s Audience Management Module provide all that is need to track, analyse different audience clusters in absolute real time. Sift uses Active Behavioural Clusters to manage these sets of customers. Each cluster in sift comprises a set of customers falling under same behavioural pattern. These clusters are updated in real time by sift, giving the user a perfect perspective of what a target audience cluster may look like. Sift also provides an analysis mechanism to mix and match and perform certain combinations to get to the specific target audience according to specific needs. And once the real time analysis is satisfying, these customers can directly be contacted with some offers or promotions.

One of my drive was to make this miles ahead of what current tools offer. I had endless brainstorming with Raja and helped to shape the current state of this module with very clear differentiators and unique selling propositions.


Lets look at some very key differentiators of audience management in sift and what value they can provide for business

Real time and Trend based counts

Sift’s audience management and active behavioural clusters provides real time counts of the customers, pertaining to certain behaviour. user can also track certain clusters and can get the trends of constantly changing counts of customers in the specific cluster.

Business Value:This provides business a great value of accurately targeting specific set of customers based on their real time patterns.

Related Entities as Audience

While traditional audience management mostly look at the entity being analysed, Sift brings in the differentiator of linking the entity with every other relationships. These linkages pave way for semantic analysis of the data and thus, a much more comprehensive audience.  We have built a special purpose database to serve these kind of knowledge representation.

Business Value: this helps business to increase their reach in market and also helps to discover new clusters.

API Based

All realtime counts and trends can be opened for 3rd party consumption. Through the provided API, 3rd parties and consume these data and then use them for different predictive analysis.

Business Value:This kind of third party data consumption may lead business to perform data monetisation.

Launch real time or scheduled campaigns instantly

Sift provides the mechanism of launching a real time campaign or a scheduled batch campaign immediately after performing the analysis on the clusters. With minimum hassle, accurate audience can be targetted straight away.

Business Value:This provides business a value of very accurate real time contextual audience targeting.

For information on our thought process and API details, you may refer to Raja’s post “Sift Audience Data Query Language (DQL)”. In this post, I will focus purely on how we envisage marketing users to access and manage the audience.

How it works

Lets look in detail now how it actually works.

Table View

Active Behavioural Cluster are shown in tabular view with counts and different actions which can be performed on each cluster. This provides user a paginated view of cluster and standard table operations like searching, sorting can be performed. Following screen illustrates table view of the audience management module.

Sift Audience Management : Table View

Category View

Active Behaviour Clusters are shown as cards and categorised. This provides a more user friendly way of looking at the clusters

Sift Audience Management : Category View

Funnel Analysis

Funnel Analysis provides the user with capability to narrow down to the specific target audience. It forms a funnel view to illustrate the filtering of the customers as the clusters are being added in the funnel. User can click on the cluster in list, to add or remove it from the funnel. Any cluster added goes at the bottom of funnel and the customer counts keep getting filtered. Summary of the operation performed by user is displayed on the right summary panel.

Sift Audience Management : Funnel Analysis

Query Analysis

This provides user with the flexibility of creating free form queries using different combinations of the clusters. As every requirement cannot be fulfilled by fixed analysis mediums, query analysis give user a huge amount of flexibility to meet the specific needs

Sift Audience Management : Query Analysis

Use Cases

Lets now look at two very interesting use cases where this audience management module can play a great role and can enhance the business outreach.

Event at a location

lets assume that an event is happening at a certain location. Now, one wants to target certain customers which are in the vicinity of the event location. So the requirement of the cluster can be specified as all customers within the vicinity of event location and having high data usage. In Sift a cluster can be created by using the geofences and that cluster will give you all the customers currently in that geofence. An already existing cluster of high data usage customers is already present. So creating a funnel analysis of these two clusters will give accurate number of customers that can be targeted at that specific point of time. User can straight away target them by using the launch program.


Share This Post
Have your say!
5 0
1 Comment

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>