My presentation given at the Association of Subscription Agents annual conference, Feb 2013.
It was titled Understanding how researchers and practitioners use STM information, but the specific theme was understanding how to design information products and services for researchs and practitioners against a background of information abundance (aka information overload).
Reading studies go back decades e.g. average numbers of readings have increased ( Tenopir) Source: Tenopir, C (2007). What does usage data tell us about our users? Online Information, London
Reading studies go back decades & reading behaviour varies across disciplines ( Tenopir) Source: Tenopir, C (2007). What does usage data tell us about our users? Online Information, London
So publishers can still lackin-depth understanding of:• how researchers use content• how it integrates with other information• the context in which content used• which articles were used, by whom, where and when?• or which parts of articles were used?
It may be even worse ...Percentage of unique visitors that do not come from recognisedsources (known IP ranges, authenticated, or registered)Geoff Bilder (2009) Brave Adventures: New Publishing Models for the Now World, SSP, Baltimore
Why was this?• cost & complexity of ﬁnding out• intermediation – libraries and agents• less value in print world anyway• but also, publishers may have thought they understood enough
The wider information ecosystem is complexRIN (2009) Patterns of information use and exchange: case studies of researchers in the life sciences
Case studies can provide a fuller understanding of differences between disciplines Humanities Physical SciencesRIN (2011) Collaborative yet independent: Information practices in the physical sciences
Large-scale surveys can provide insight, especially if repeated Inger/Gardner: How Readers Discover Content in Scholarly Journals (Renew, 2012) http://www.renewtraining.com/How-Readers-Discover-Content-in-Scholarly-Journals-summary-edition.pdf
Whats new• lots of data!• near-real-time data collection• mobile devices = personal data• point-of-care use & similar• Big Data analytics• altmetrics – using data to measure impact• CRIS and research metrics/evaluation• and coming up, distributed annotation (Hypothes.is)
Deep log analysis (e.g. CIBER) offers one approach • what they actually do (online), not what they say or wish they do. E.g.: • very little time reading in the digital environment • Only 1–3 pages viewed & >50% of all visitors never come back • PDFs downloaded, but saved rather than read ofﬂineSource: Nicholas & Clark (2012) Reading in the digital environment. Learned Publishing doi: 10.1087/20120203
More granular data onreading history now possible
Information abundance is a fact ... BUT What keeps us awake at night is notthat all this information will cause us tohave a mental breakdown but that weare not getting enough of theinformation that we need —David Weinberger [my emphasis]
Designing products forinfo-overloaded users• Data/Information pyramid: knowing- by-reducing • selective, or ﬁltering out• Better ﬁlters – ﬁltering forward • surfacing relevant information, at the right time, in the right context
Workﬂow solutions • Combining (ﬁltered) content & software tools, integrated with user work/information environment • Improved certainty and consistency of decision making • Enhanced of productivity • Certainty in terms of compliancedepositphotos.com
Designing workﬂow solutions: contextual enquiry • Combines multiple methods, e.g. • surveys • cluster / conjoint analysis • on-the-job observation • Three minutes method (Thomson) • 25–50 interviews per user • behaviour 3 mins before/after using the information / serviceHarrington & Tjan 2008 Transforming Strategy One Customer at a Time, Harvard Business Review
User segmentation • We ask editors: Do you know the proﬁle of speciﬁc users? Who are you targeting? The CHOs? The Male Social Glue inﬂuencers? We ask: who is more valuable? Which segment? • Our audience follows an 80-20 rule: 20% of the audience is of high value to us. 80% cost us more than the revenue they generate, for example, if they watch many long videos. Source: Outsell (2010) eMedia Organization Part III: Analytics-Wired Content www.outsellinc.com
User segmentation: goals• to identify differentiated segments• clear identiﬁable differences• representing real behaviour and/or attitudinal differences• allowing prediction of behaviour of future users
User segmentation: goals• to use data to identify differentiated segments• clear identiﬁable statistically signiﬁcant differences• representing real behaviour and/or attitudinal differences• allowing statistically valid prediction of behaviour of future users
User segmentation: approach• Large, detailed surveys• Factor analysis ➜ correlated, differentiating statements• Cluster analysis ➜ possible segmentations• Test potential segmentations by interviewing
OvidMD and ClinicalKey Comprehensive? Trusted? Fast?Source: Wolters Kluwer; Elsevier
What sort of questions might we answer (or try to)?• What are the different barriers potential users face?• Who are the potential customers for possible new services?• How do different market segments value different features, and how might these be grouped? • What new products / services are missing from out portfolios?
Why should we bother? • If your market is experiencing discontinuity • If you lack clear value propositions • If you rely too heavily on channel segmentation • If you sense that you face new customer demands and competition Harrington & Tjan 2008 Transforming Strategy One Customer at a Time, Harvard Business Review
Some conclusions• Analytics capabilities are now a core requirement• Opportunities to borrow from B2C• As content commoditises, new ways of adding value become critical• Content / Data are likely to be distributed across the web ➜ open for new entrants to create new services