27. Campaign improvement
KPI’s:
number of conversions
cost per conversion decline in July = holidays
decline in December = Xmas, New Year
28. Keyword Extractor
Keyword list for each industry
Corpus of human-made catalogues
Subtitles corpus
Twitter corpus
Mathematic models = language independent
32. Analysis output
Category (words, relevance):
Apple Inc. (17, 185)
Main topic classifications (53, 183)
Apple Inc. Hardware (11, 164)
Society (44, 157)
IPhone (15, 144)
CSV with distances
Keywords (relevance):
Apple (174494) right (13156)
Iphone (165368) law (12760)
Legal (20880) iphone apple (8859)
Twitter (18249) person (8424)
Iphones (17718) home (6488)
people (17091) app store (5616)
apple iphone (14765) wrong (4076)
Jailbreaking (13771)
33. Selection = relevance
Google normalized distance
Spectral analysis
Historical data
... and 10 more filters depending on type of campaign
57. 1. 2.
Keyword &!copy Ad Systems (AdWords)
keyword extractor dynamic copy
search engine distance QS based preferences
description analysis, etc.
58. 1. 2.
Keyword &!copy Ad Systems (AdWords)
keyword extractor dynamic copy
search engine distance QS based preferences
description analysis, etc.
3.
Media
proprietary auctioning
online reports from all systems, etc.
66. Twitter Corpus
Czech tweets (language and location)
geo-location and time-based influence of language
deviations
btw. did you know that people are in better mood on
Tuesdays than on Fridays?
combined with Aboutness allows creation of Twitter
stream on a certain topic
combined with mood analysis can find out positive or
negative feeling towards a topic or brand
can generate local news