Daniel Acosta
Big Data Visualisations with React
Lead Front end engineer at Fospha
Previously at Transferwise, Thomas Cook
At Fospha.
Were solving the biggest problem marketers
have today, understanding their true cost of
acquisition across channels
Old World
Customer Journey
Platform
Basic rule based
Interest in broader rule Hidden
based attribution models 1 2 3
attribution models states
Output A B C
states
D E
Beginnings First non-linear view
Product evolution to MTA
On-site Analytics Data engineering MTA
2000 Panorama Site reporter with SQL
Proto-GA web analytics Write SQL queries against multi-channel BI data warehouse
Console Visitor stitching
2010 BI Dashboards Proprietary tech to track visitors over multiple touches, visits
and devices, then stitch together for single multi-touch view of
Content Performance visitor journey from acquisition through to repeat purchase
2015 Applying data science to digital content consumption
with proprietary algorithm measuring human Customer data platform
interaction with content Evolving iJento's journey stitching technology and extensible
customer data model enable data environment to allow multi- MTA Query
2016 H1 Customer Journey Optimisation touch attribution and activation of customer data Query reporting suite expanded to include MTA
Analyses visitor journeys through digital and offline focussed reports, combining data from PPC and
domains across multi-touch visits Google & Adobe clickstream data ingestion Social (cost) and CRM (revenue) sources to prove the
2016 H2 Capability to ingest Google clickstream data from GCS
actual ROI of channels across multi-touch customer
journeys
into Fospha processing as if native
Data-driven attribution
Integration with bid mgmt. platforms to operationalise Development of algorithmic attribution modelling
2017 Q1 Fospha insight using Markov Chains
Integration to push our derived datasets to automated
bidding platform, optimising PPC strategy to actual ROAS Facebook Cost API
2017 Q2 Ingestion of Facebook Ad cost for ROAS calculation
The Data Library (start Q3 WIP)
Re-write of data storage, processing & transport framework MTA app (start Q2 - WIP)
to accelerate ingestion of new data sources and availability of MTA app including performance dashboards with rules-
2017 Q3 derived datasets based and data-driven models, with APIs to Facebook,
Adwords and other major advertising platforms
2017 Q4
Facebook impressions integration (pending FB
approval)
Layering Facebook impressions into MTA sequencing for
2018 to prove value of FB as channel vs PPC
MTRA (Multi-touch re-allocation)
Evolving data-driven MTA to automate re-allocation of
4 marketing channel spend
How we started Visualizing Big Data
1 Year Ago
Data Visualisation Custom Bar Charts
Custom
Bar Charts
Adding clear separation with components
=
Making human readable code for a
better developer experience.
Can you guess what part of the
funnel each component relates to?
Visualising data with
Recharts
Fospha Attribution
1
7
0.6 9
5
0.1
5 10
0.
1
20
1
3
Start with a touchpoint Build the transitions from Find the effect of
network touchpoint to touchpoint removing states from
your system to assign
the state value
How our customers consume Fospha MTA
Fospha Query Fospha MTA (WIP)
Reporting Dedicated front-end
for dashboards
Third Party Exporting
Visualisations CSV, Excel, APIs
Tableau
12
12
From paper to prototype
Lightbulb demo built with
Recharts
https://fospha.com/meetup
15
