an engaging click
DESCRIPTION
A good search engine is one when users come very regularly, type their queries, get their results, and leave quickly. With user engagement metrics from web analytics, these translate to a low dwell time, often low CTR, but a very high return rate. But user engagement is not just about this. User engagement is a multifaceted, complex phenomenon, giving rise to a number of approaches for its measurement: self-reporting (e.g. questionnaires); observational methods (e.g., facial expression analysis, desktop actions); and of course web analytics using online behavior metrics. These methods represent various trade-offs between the scale of the data analyzed and the depth of understanding. For instance, surveys are hardly scalable but offer rich, qualitative insights, whereas click data can be collected on a large-scale but are more difficult to analyze. This talk will present various efforts aiming at combining approaches to measure engagement and seeking to provide insights into what makes an engaging experience. The talk will focus of what makes users click or not click, and what this means in terms of user engagement. SIGIR 2013 Industry Track: Keynote by Ricardo Baeza-Yates - VP, Yahoo! Research Europe & Latin AmericaTRANSCRIPT
Ricardo Baeza-Yates SIGIR 2013 – Industry Talk
An Engaging Click
Why is it important to engage users?
• In today’s wired world, users have enhanced expectations about their interactions with technology
… resulting in increased competition amongst the purveyors and designers of interactive systems. • In addition to utilitarian factors, such as usability, we must
consider the hedonic and experiential factors of interacting with technology, such as fun, fulfillment, play, and user engagement.
2 An Engaging Click
CTR and user engagement
CTR
3 An Engaging Click
Multimedia search activities often driven by entertainment needs, not by information needs
CTR and entertainment driven search
(Slaney, 2011)
An Engaging Click 4
I just wanted the phone number … I am totally satisfied J
CTR and factual needs
An Engaging Click 5
This talk
What is user engagement? What are the characteristics of user engagement? How to measure user engagement? What is an engaging click?
1. inter-session metric 2. online multi-tasking 3. serendipity
6 An Engaging Click
Work on user engagement across
web applications
Implications to search
http://thenextweb.com/asia/2013/05/03/kakao-talk-rolls-out-plus-friend-home-a-revamped-platform-to-connect-users-with-their-favorite-brands/
Engagement is on everyone’s mind
http://socialbarrel.com/70-percent-of-brand-engagement-on-pinterest-come-from-users/51032/
http://iactionable.com/user-engagement/
http://www.cio.com.au/article/459294/heart_foundation_uses_gamification_drive_user_engagement/
http://www.localgov.co.uk/index.cfm?method=news.detail&id=109512
http://www.trefis.com/stock/lnkd/articles/179410/linkedin-makes-a-90-million-bet-on-pulse-to-help-drive-user-engagement/2013-04-15
An Engaging Click 8
What is user engagement?
User engagement is a quality of the user experience that emphasizes the positive aspects of interaction – in particular the fact of being captivated by the technology (Attfield et al, 2011).
user feelings: happy, sad, excited, …
emotional, cognitive and behavioural connection that exists, at any point in time and over time, between a user and a technological resource
user interactions: click, read, comment, buy…
user mental states: involved, lost, concentrated…
9 An Engaging Click
Considerations in the measurement of user engagement
• Short term (within session) and long term (across multiple sessions)
• Laboratory vs. field studies • Subjective vs. objective measurement • Large scale (dwell time of 100,000 people) vs.
small scale (gaze patterns of 10 people) • User engagement as process vs. product
One is not better than other; it depends on what is the aim.
10 An Engaging Click
Characteristics of user engagement (I)
• Users must be focused to be engaged • Distortions in the subjective perception of time used to
measure it
Focused attention (Webster & Ho, 1997; O’Brien,
2008)
• Emotions experienced by user are intrinsically motivating • Initial affective “hook” can induce a desire for exploration,
active discovery or participation
Positive Affect (O’Brien & Toms, 2008)
• Sensory, visual appeal of interface stimulates user & promotes focused attention
• Linked to design principles (e.g. symmetry, balance, saliency)
Aesthetics (Jacques et al, 1995; O’Brien,
2008)
• People remember enjoyable, useful, engaging experiences and want to repeat them
• Reflected in e.g. the propensity of users to recommend an experience/a site/a product
Endurability (Read, MacFarlane, & Casey,
2002; O’Brien, 2008)
12 An Engaging Click
Characteristics of user engagement (II) • Novelty, surprise, unfamiliarity and the unexpected • Appeal to users’ curiosity; encourages inquisitive
behavior and promotes repeated engagement
Novelty (Webster & Ho, 1997; O’Brien,
2008)
• Richness captures the growth potential of an activity • Control captures the extent to which a person is able
to achieve this growth potential
Richness and control (Jacques et al, 1995; Webster &
Ho, 1997)
• Trust is a necessary condition for user engagement • Implicit contract among people and entities which is
more than technological
Reputation, trust and expectation (Attfield et al,
2011)
• Difficulties in setting up “laboratory” style experiments • Why should users engage?
Motivation, interests, incentives, and
benefits (Jacques et al., 1995; O’Brien & Toms, 2008)
13 An Engaging Click
14
Measuring user engagement Measures Characteristics
Self-reported engagement
Questionnaire, interview, report, product reaction cards, think-aloud
Subjective Short- and long-term Lab and field Small-scale Product outcome
Cognitive engagement
Task-based methods (time spent, follow-on task) Physiological measures (e.g. EEG, SCL, fMRI, eye tracking, mouse-tracking)
Objective Short-term Lab and field Small-scale and large-scale Process outcome
Interaction engagement
Web analytics metrics + models
Objective Short- and long-term Field Large-scale Process outcome
15 An Engaging Click
Large-scale measurements of user engagement – Web analytics
Intra-session measures Inter-session measures
• Dwell time / session duration
• Play time (video) • (Mouse movement) • Click through rate (CTR) • Mouse movement • Number of pages viewed
(click depth) • Conversion rate (mostly for
e-commerce) • Number of UCG
(comments)
• Fraction of return visits • Time between visits (inter-session
time, absence time) • Total view time per month (video) • Lifetime value (number of actions) • Number of sessions per unit of time • Total usage time per unit of time • Number of friends on site (social
networks) • Number of UCG (comments)
• Intra-session engagement measures our success in attracting the user to remain on site for as long as possible.
• Inter-session engagement can be measured directly or, for commercial sites, by observing lifetime customer value.
16 An Engaging Click
Dependency on task • Engagement varies by task:
– user who accesses a website to check for emails (goal-specific) has different engagement patterns from one browsing for leisure.
• In (Yom-Tov et al, 2013), sessions in which 50% or more of the visited sites belonged to the 5 most common sites (for each user) were classified as goal-specific. – 38% sessions were goal-specific – most users (92%) both goal-specific and non-goal-
specific sessions – average downstream engagement in goal-
specific sessions was 0.16 vs. 0.2 during non-goal-specific sessions
17 An Engaging Click
18
User engagement in search – “relevance”
• Click-through rate (CTR) • Dwell time (search result) • Time to first click
• Skipping
• Abandonment rate • Number of query reformulations
• Search engine switching
• Interleaving
• Cumulative gain family of metrics
• …
An Engaging Click 19
20
Click vs cursor – heat-map Estimate search result relevance
(Bing - Microsoft employees – 366,473 queries; 21,936 unique cookies; 7,500,429 cursor move or click) the role of hovering
(Huang et al, 2011)
21 An Engaging Click
Mouse movement – what can hovering tell about relevance?
Click-through rate: % of clicks when URL Shown (per query) Hover rate: % hover over URL (per query) Unclicked hover: Media time user hovers over URL but no click (per query) Max hover time: Maximum time user hover over a result (per SERP)
(Huang et al, 2011)
22 An Engaging Click
• Domain: Yahoo! Answers Japan • Study: Inter-session engagement metric
23
(Dupret & Lalmas, 2013)
If users find a web application interesting, engaging or useful, they will return to it sooner.
Absence time and survival analysis
Easy to implement and interpret Can compare many things in one go No need to estimate baselines But need lots of data to account for noise
(Dupret & Lalmas, 2013)
24 An Engaging Click
Survival Analysis: high hazard rate = short absence
Using absence time to compare 6 ranking functions (buckets) on Yahoo! Answers Japan
1. Returning relevant results is important, but is not enough to keep returning to the search application
2. Clicks after the 5th results reflect poorer user experience; users cannot find what they are looking for
3. No click means a bad user experience 4. Clicking lower in the ranking suggests more careful choice
from the user 5. Clicking at bottom is a sign of low quality overall ranking 6. Users finding their answers quickly (click sooner) return
sooner to the search application 7. Returning to the same search result page is a worse user
experience than reformulating the query. An Engaging Click 25
26
Online multi-tasking
users spend more and more of their online session multi-tasking, e.g. emailing, reading news, searching for information à ONLINE MULTI-TASKING navigating between sites, using browser tabs, bookmarks, etc seamless integration of social networks platforms into many services
leaving a site is not a “bad thing!”
(fictitious navigation between sites within an online session)
181K users, 2 months browser data, 600 sites, 4.8M sessions • only 40% of the sessions have no site revisitation
• hyperlinking, backpaging and teleporting
An Engaging Click 27
• Domain: 700+ web applications • Study: Online multi-tasking
28
(Lehmann et al, 2013)
Online multi-tasking affects the way users interact (or engage) with sites.
Online multi-tasking – and search
181K users, 2 months browser data, 600 sites, 4.8M sessions • only 40% of the sessions have no site revisitation
• commonly accessed sites between visits à search 22%, navigation 12%, social 8% • for some sites (e-commerce) same sites are accessed between visits à one task? • no patterns for sites such as mail, social à anchor, habit?
• longer time between visits à a different task (new search) • more vs less times spent at each revisit à increased vs shift of attention
An Engaging Click 29
Navigating between sites – hyperlinking, backpaging and teleporting
timestamp page navi1346242507 1 T1346242567 2 L1346242627 3 L(1346242687) 1 B1346242687 4 L1346242747 5 T1346329147 6 L(1346329207) 5 B1346329207 7 L(1346329267) 2 B1346329267 8 L
2
3
1
4
8
5
76
click-tree 1 click-tree 2
1 - 2 - 3 - 1 - 4 - 5 - 6 - 5 - 7 - 2 - 8timestamp page referral1346242507 1 -1346242567 2 11346242627 3 21346242687 4 11346242747 5 -1346329147 6 51346329207 7 51346329267 8 2
8
7
6
2
3
1
4
5
2
3
1
4
8
5
76
click-tree 1 click-tree 2
(a) Interaction dataclick-stream
(b) Navigation pathclick-stream
(c) Logical navigationclick-trees
(d) Interaction datatree-stream
(e) Navigation pathtree-stream
Page [L] Hyperlinking [B] Backpaging [T] Teleportingn
Number of backpaging actions is an under-estimate! (using browser back button, or user returns to one of several open tabs/windows)
An Engaging Click 30
Revisitation and navigation patterns auction sites [complex attention]
●
●
●●
●
●●
●
●
10
11
12
1 2 3 4 5 6 7 8 9
p-value = 0.24m = 0.142
100% 67% 54% 46% 41% 35% 31% 29% 26%
search sites [increasing attention]
●●
●●
● ●
●
●
●
10.8
11.0
11.2
1 2 3 4 5 6 7 8 9
100% 69% 54% 44% 38% 33% 29% 26% 23%
p-value < 0.05m = 0.063
●
●
●
● ●
●●
● ●10.8
11.2
1 2 3 4 5 6 7 8 9
p-value < 0.05����������
100% 54% 36% 26% 20% 17% 14% 12% 10%proportion of users
% o
f to
tal page v
iew
s o
n s
ite
% o
f n
avig
atio
n t
yp
e
Hyperlinking
mail sites [decreasing attention]
●
●●
● ●
●●
●●1
011
12
13
1 2 3 4 5 6 7 8 9
100% 62% 41% 29% 21% 16% 13% 10% 8%
p-value < 0.05m = -0.288
average attention
1 2 3 4 5 6 7 8 90.0
0.4
0.8
0.0
0.4
0.8
1 2 3 4 5 6 7 8 9 0.0
0.4
0.8
1 2 3 4 5 6 7 8 9 0.0
0.4
0.8
1 2 3 4 5 6 7 8 9
k [kth visit on site] k [kth visit on site] k [kth visit on site] k [kth visit on site]Teleporting Backpaging
An Engaging Click 31
Online multi-tasking – and web search • 48% sites visited at least 9 times • Revisitation “level” depends on site
• 10% users accessed a site 9+ times (23% for search sites); 28% at least four times (44% for search sites)
• Activity on site decreases with each revisit but activity on many search (and adult) sites increases
• Backpaging usually increases with each revisit but hyperlinking remains important means to navigate between sites
An Engaging Click 32
33
Networked user engagement: engagement across a network of sites
• Large online providers (AOL, Google, Yahoo!, MSN, etc.) offer not one service (site), but a network of sites
• Each service is usually optimized individually, with some effort to direct users between them
• Success of a service depends on itself, but also
on how it is reached from other services (user traffic)
An Engaging Click 34
Measuring downstream engagement
User session
Pro
vide
r site
s
Downstream engagement for site A
(% remaining session time)
Site A
35
(Yom-Tov etal, 2012)
Influential features o Time of day
o Number of (non-image/non-video) links to Yahoo! sites in HTML body o Average rank of Yahoo! links on page o Number of (non-image/non-video) links to non-Yahoo! sites in HTML body
o Number of span tags (tags that allow adding style to content or manipulating content, e.g. JavaScript)
o Link placements and number of Yahoo! links can influence downstream engagement o Not new, but here shown to hold also across sites
o Links to non-Yahoo! sites have a positive effect on downstream engagement o Possibly because when users are faced with abundance of outside links
they decide to focus their attention on a central content provider, rather than visiting multitude of external sites
(Yom-Tov et al, under submission)
• Domain: social media (Yahoo! Answers and Wikipedia) • Study: serendipity (in entity search)
37
(Bordino, Mejova & Lalmas, 2013)
Interesting search results may promote serendipitous browsing.
Yahoo! Answers vs Wikipedia community-driven question & answer portal • 67 336 144 questions &
261 770 047 answers • January 1, 2010 –
December 31, 2011 • English-language
community-driven encyclopedia • 3 795 865 articles • as of end of
December 2011 • English Wikipedia
curated high-quality knowledge variety of niche topics
minimally curated opinions, gossip, personal info
variety of points of view
38 An Engaging Click
Entity Search
we build an entity-driven serendipitous search system based on entity networks extracted from Wikipedia and Yahoo! Answers
Serendipity finding something good or useful while not specifically looking for it, serendipitous search systems provide relevant and interesting results
Wikipedia
39
Yahoo! Answers
An Engaging Click
Retrieval
Wikipedia Yahoo! Answers
Combined
Precision @ 5 0.668 0.724 0.744 MAP 0.716 0.762 0.782
Justin Bieber, Nicki Minaj, Katy Perry, Shakira, Eminem, Lady Gaga, Jose Mourinho, Selena Gomez, Kim Kardashian, Miley Cyrus, Robert Pattinson, Adele %28singer%29, Steve Jobs, Osama bin Laden, Ron Paul, Twitter, Facebook, Netflix, IPad, IPhone, Touchpad, Kindle, Olympic Games, Cricket, FIFA, Tennis, Mount Everest, Eiffel Tower, Oxford Street, Nubcrburgring, Haiti, Chile, Libya, Egypt, Middle East, Earthquake, Oil spill, Tsunami, Subprime mortgage crisis, Bailout, Terrorism, Asperger syndrome, McDonal's, Vitamin D, Appendicitis, Cholera, Influenza, Pertussis, Vaccine, Childbirth
3 labels per query-result pair gold standard quality control
Yahoo! Answers Jon Rubinstein Timothy Cook Kane Kramer
Steve Wozniak Jerry York
Wikipedia System 7
PowerPC G4 SuperDrive
Power Macintosh Power Computing Corp.
Steve Jobs • Annotator agreement
(overlap): 0.85 • Average overlap in
top 5 results: <1 40
retrieve entities most related to a query entity using random walk
An Engaging Click
| relevant & unexpected | / | unexpected | number of serendipitous results out of all of the unexpected results retrieved
| relevant & unexpected | / | retrieved | serendipitous out of all retrieved
41
Baseline Data Top: 5 en//es that occur most frequently WP 0.63 (0.58) in top 5 search from Bing and Google YA 0.69 (0.63) Top –WP: same as above, but excluding WP 0.63 (0.58) Wikipedia page from results YA 0.70 (0.64) Rel: top 5 en//es in the related query WP 0.64 (0.61) sugges/ons provided by Bing and Google YA 0.70 (0.65) Rel + Top: union of Top and Rel WP 0.61 (0.54) YA 0.68 (0.57)
Serendipity “making fortunate discoveries by accident” Serendipity = unexpectedness + relevance
“Expected” result baselines from web search
An Engaging Click
Interestingness ≠ Relevance Interesting > Relevant
Relevant > Interesting
Oil Spill à Penguins in Sweaters WP
Robert Pattinson à Water for Elephants WP
Lady Gaga à Britney Spears WP
Egypt à Cairo Conference WP
Netflix à Blu-ray Disc YA
Egypt à Ptolemaic Kingdom WP & YA
42 An Engaging Click
Similarity (Kendall’s tau-b) between result sets and reference ranking
43
Data tau-‐b Which result is more WP 0.162 relevant to the query? YA 0.336 If someone is interested in the query, would WP 0.162 they also be interested in the result? YA 0.312 Even if you are not interested in the query, WP 0.139 is the result interes;ng to you personally? YA 0.324 Would you learn anything new about WP 0.167 the query from the results YA 0.307
Following (Arguello et al, 2011) 1. Labelers provide pairwise
comparisons between results 2. Combine into a reference ranking 3. Compare result ranking to optimal
ranking using Kendall’s tau
Assessing “interestingness”
An Engaging Click
44
Take-away messages
• Search is not just about specific information needs • People search for many other reasons
– Navigation – Transaction – Fun (ECIR 2012 workshop) – Etc.
• Engagement in search is to view search activities as part of the current overall task of a user
• We never know what we get if we are ready to explore – Users do things that no one expects, not even them!
(like staying inside Yahoo! in spite of having many links to go elsewhere) – So a link is not everything, for search too!
• Summarizing, we need to look at engagement in a broader way
Thank you
Acknowledgements: Mounia Lalmas, Jahnette Lehmann, George Dupret, Ilaria Bordino, Yelena Mejova and Elad Yom-Tov.
An Engaging Click 46