Loading presentation...

Present Remotely

Send the link below via email or IM


Present to your audience

Start remote presentation

  • Invited audience members will follow you as you navigate and present
  • People invited to a presentation do not need a Prezi account
  • This link expires 10 minutes after you close the presentation
  • A maximum of 30 users can follow your presentation
  • Learn more about this feature in our knowledge base article

Do you really want to delete this prezi?

Neither you, nor the coeditors you shared it with will be able to recover it again.



No description

Gavin Storey

on 24 April 2013

Comments (0)

Please log in to add your comment.

Report abuse

Transcript of Scribblelytics

Historical Real-time Social Predictive Domain specific Publishing/ Content Marketing Page Info State of Analytics Mobile Covering events in the past Covering current events One size fits all Generic stats for all IT/Infrastructure gavstorey@gmail.com #gavstorey Gavin Storey Founder + Developer Predictive Analytics and Marrying Goats Questions?
(please be kind) Covering events in a specific industry or domain Predicting future events Obama predicted his second election win
IBM wins Jeopardy! with Watson computer
Microsoft predicts a person's future location
Google predicts flu trends, outbreaks and ads
Tesco personalizes coupons to retain visitors
Target predicts customer pregnancy by purchases
Netflix recommends movies
Insurance company predicts the death of elderly within 18 months in order to trigger counseling Examples of fancy predictive analytics Some quite spooky if you ask me likes Cat tracks mice What is Scribblelytics? Cat sees a benefit but refines further Journalists + Content Producers Publishers + Brands Simple, addictive analytics for journalists, bloggers and publishers He likes mice and wants to catch them a lot* Cat monitors their location over time (identifies a predictor) Cat watches more closely. Mice like cheese, but which mice? No good, so cat tries another predictor 10% sugar 20% fat chocolate 30% cheese 40% Cat watches the mice and decides to monitor what they eat instead. *no mice have been injured in the making of these slides OK, give me a simple yet cute example Cat defines goal Cat sees trend and reacts mmm nice mice pieces! Next up...birds! Cat predicts mouse likes cheese and targets Bob (validates model) yummy cheese 10% 10% 10% 10% 10% 10% 10% 10% 10% 10% cheese 100% cheese 100% cheese 50% cheese 50% cheese 50% chocolate 100% chocolate 50% fat 100% 50% chocolate 50% chocolate 50% fat 50% sugar chocolate 50% 50% sugar chocolate 50% 50% fat Bob Bo So... what does this have to do with publishing? Why should I be bothered? And why haven't I seen a goat? Equality is just a dream The phantom menace: crosswords? Believing PA is automagical*
Not defining the goal or metric for success
Lack of enough data or relevant data
Incorrect or poor choice of statistical model
Believing your model is always right
Only using one model It's not all catnip and roses Nooooo *Automatic, but with an apparent element of magic Not just small publishers have kid issues Here are some amusing examples of some problems I have dealt with 1st publishing problem, 1st bollocking and why basic models and loyal readers can cause problems J E D I Y O D A L I G H T S A B E R S = Predicting content is not as simple as it first seems, with changing and complicated influences on predictions RIP Rose 2007, who sadly died after 1 year of marriage. She loved the coverage. Top 10 content through 2007
Millions of visitors/page views
Self-propagated embarrassment
Competing factions Section Publication Content Social Home 40% now 7% of page views Search Marketing 20% now 3% of page views 60% now 90% of page views Opinion Slideshows Reviews Features News Events Investigative Comics Other was 50% now 30% was 0% now 40% 10% was 40% now 10% International = regional = local, right? Still no goat? ~1k page views a week
800 visitors a week
avg 1.5 mins on page
= lower than average traffic What the hell is going on? Loyal crossword enthusiasts flock in at 8am, leave shortly after
= prediction failure
= first lesson sources *before and after based on Google stats from 2008 to 2013 National
50% of visitors
$15-5cpm Regional
30% of visitors
$25-8cpm International
20% of visitors
$0cpm National
40% of visitors
-$12cpm Regional
45% of visitors
-$17cpm International
15% of visitors
-10% 2008 2012 Page views = $$$ = bad assumptions National
10% ad block
+1s per ad Regional
5% ad block International
20% ad block
+1s per ad National
40% of visitors
20% ad block
+2s per ad Regional
45% of visitors
10% ad block
+1s per ad International
15% of visitors
50% ad block
+3s per ad 2008 2012 Simple stats customized to match your priorities and publication, with easy to use views for each department
Motivate and empower your editorial staff
Identify loyalty and retention locally or nationally
Predictions on content, traffic and ads
Get the wider picture by comparison to competitors
Integration with web, mobile, tablets and print distribution
Powerful API with exports, widgets and notifications to enhance your business or site Simple real-time leaderboards showing your impact
Compare your content across publications, within editorial cycles and topics
Identify loyal readers and sources of your traffic
See what people are saying and react
Review your content and find out what works Deeper influence of location and a rare metric Scribblelytics Publisher
Info Visitor
Info Social Google Analytics Author
info Trending Tweets Score Semantic + content analysis Special Sauce gavstorey@gmail.com #gavstorey Gavin Storey Founder + Developer Coming out of secret ninja mode soon Scribblelytics *I can hear your thoughts, and yes some of these do more than just historic data A potentially misunderstood and yet hot term *Potential competitor alert Big boys Recent publishing upstarts* Big data*
Cloud computing
Software, tools, languages + methods
Cost Predictive Analytics isn't new, so what happened? *Oh I feel trendy Now things are getting interesting Not great at this end Not very happy Getting happier Things are not looking good for Bob + Bo Diagram by no means represents the actual occurrence and joy of log file parsing and comparison Crossword instead of Breaking News? = boss is angry! Then as if by magic... *this includes direct traffic + email if you are wondering why it is so high *if you go back further, search accounted for 70% of all traffic and Google 60% This may seem low, especially when you can receive large amounts of traffic from Digg, Reddit etc. However, smaller publishers struggled to keep consistently in such charts. Not a shocker but Facebook + Twitter dominate this space list of news, arts, music coverage Audio + Video Blog
Posts largely curated list of content Why did I go into publishing?
Do the stats look ok?*
Is it a fault, error, system or coding problem?
Is it artificial traffic, robots etc? It happens again: lots of types of content, with each publisher
having a separate secret sauce + flavor Infographics Graphics Interactive Polls Discussion Breaking Even here further fragmentation PERFORMANCE Content also drives traffic to other content through links, loyalty, authors, tags Last but not least National
40% of visitors
$3cpm Regional
51% of visitors
$8cpm International
8% of visitors 2012 Multiple factors: = visitors with income potential Predictions Direct measurement Publisher-based weightings So how does all this magic work? = Better informed publishers, staff + decisions sadly no magic unicorns light hacking light reading *if you are from these companies I mean no harm Content (text, structure, quality), author, publisher, site, social, promotion, marketing CONTENT That was a lot of info. I am now losing focus, so just tell me what it all means. Until then you may have a little time for some light reading *please note this was back in the day when we only had daily stats Some of these are directly quantifiable, others are not however they can be indicative How much for a pig? How much will you earn from your content $$$ = (PageViews * Ads * CPM/1000) = right? Is location universal? No! New metrics pop up all of the time I predict this presentation runs over = 65 slides * time = too long
Full transcript