=Paper=
{{Paper
|id=None
|storemode=property
|title=A Model of Consumer Search Behaviour
|pdfUrl=https://ceur-ws.org/Vol-909/paper4.pdf
|volume=Vol-909
|dblpUrl=https://dblp.org/rec/conf/eurohcir/Russell-RoseM12
}}
==A Model of Consumer Search Behaviour==
https://ceur-ws.org/Vol-909/paper4.pdf
A Model of Consumer Search Behaviour
Tony Russell-Rose Stephann Makri
UXLabs University College London Interaction Centre,
London University College London, Gower St.
UK London, WC1E 6BT, UK
+44 (0)7779 936191 +44 (0)20 7679 0696
tgr@uxlabs.co.uk s.makri@ucl.ac.uk
ABSTRACT In this paper, we examine the needs and behaviours of individuals
across a range of site search scenarios. These are based on an
In order to design better search experiences, we need to analysis of user needs derived from a series of customer
understand the complexities of human information-seeking engagements involving the development of customised site search
behaviour. In previous work [13], we proposed a model of applications. In so doing, we extend and validate a model of
information behavior based on an analysis of the information information behaviours derived from a previous study of
needs of knowledge workers within an enterprise search context. enterprise search users [13].
In this paper, we extend this work to the site search context,
examining the needs and behaviours of users of consumer- The model is based on a set of ‘search modes’ that users employ
oriented websites and search applications. to satisfy their information search and discovery goals. It extends
the IR concept of information-seeking to embrace a broader
We found that site search users presented significantly different notion of discovery-oriented problem solving, addressing a wider
information needs to those of enterprise search, implying some range of information interaction and information use behaviours.
key differences in the information behaviours required to satisfy The overall structure of the model reflects Marchionini’s [9]
those needs. In particular, the site search users focused more on framework, and consists of three lower-level ‘lookup’ modes
simple “lookup” activities, contrasting with the more complex, (locate, verify and monitor), three “learn” modes (compare,
problem-solving behaviours associated with enterprise search. We comprehend and explore) and three higher-level “investigate”
also found repeating patterns or ‘chains’ of search behaviour in modes (analyze, evaluate and synthesize).
the site search context, but in contrast to the previous study these
were shorter and less complex. These patterns can be used as a We investigate the degree to which the model extends to
framework for understanding information seeking behaviour that accommodate the domain of site search (i.e. consumer-oriented
can be adopted by other researchers who want to take a ‘needs websites and search applications) and discuss some of the
first’ approach to understanding information behaviour. differences between the needs and goals of enterprise search users
versus those of site search. We conclude by exploring the ways in
Categories and Subject Descriptors which these modes combine to form distinct chains or patterns,
H.3.3 [Information Search and Retrieval]: Search process; and reflect on the value this offers as a framework for expressing
H.3.5 [Online Information Services]: Web-based services complex patterns of behaviour.
General Terms 2. MODELS OF INFORMATION SEEKING
Human Factors. The framework investigated in this study is influenced by a
number of existing models. For example, Bates [1] identified a set
Keywords of 29 search ‘tactics’ which she organised into four broad
Site search, enterprise search, information seeking, user categories, including monitoring (“to keep a search on track”).
behaviour, search modes, information discovery, user experience Likewise, O’Day & Jeffries [11] examined the use of information
design. search results by clients of professional information intermediaries
and identified three categories of behaviour, including monitoring
a known topic or set of variables over time and exploring a topic
1. INTRODUCTION in an undirected fashion. They also observed that a given search
Classic IR (information retrieval) is predicated on the notion of scenario would often evolve into a series of interconnected
users searching for information in order to satisfy a particular searches, delimited by triggers and stop conditions that signalled
'information need'. However, it is now accepted that much of what transitions between modes within an overall scenario.
we recognize as search behaviour is often not informational per
se. For example, Broder [2] has shown that the need underlying a Cool & Belkin [3] proposed a classification of interaction with
given web search could in fact be navigational (e.g. to find a information which included evaluate and comprehend. They also
particular site) or transactional (e.g. through online shopping, proposed create and modify, which together reflect aspects of our
social media, etc.). Similarly, Rose & Levinson [12] have synthesize mode.
identified the consumption of online resources as a further Ellis and his colleagues [4, 5, 6] developed a model consisting of
common category of search behaviour. a number of broad information seeking behaviours, including
Presented at EuroHCIR2012. Copyright © 2012 for the individual papers monitoring and verifying (“checking the information and sources
by the papers' authors. Copying permitted only for private and academic found for accuracy and errors”). In addition, his browsing mode
purposes. This volume is published and copyrighted by its editors. (“semi-directed searching in an area of potential interest”) aligns
with our definition of explore. He also noted that it is possible to Moreover, the scope and focus of these scenarios represents a
display more than one behaviour at any given time. In revisiting further point of differentiation. In previous studies, (e.g. [8]),
Ellis’s findings among social scientists, Meho and Tibbo [10] measures have been taken to address the limitations of using
identified analysing (although they did not elaborate on it in interview data by combining it with direct observation of
detail). More recently, Makri et al [8] proposed searching information seeking behaviour in naturalistic settings. However,
(“formulating a query in order to locate information”), which the behaviours that this approach reveals are bounded by the
reflects to our own definition of locate. functionality currently supported by existing systems and working
practices, and as such do not reflect the full range of aspirational
In addition to the research-oriented models outlined above, we
or unmet user needs encompassed by the scenarios in this study.
should also consider practitioner-oriented views. Spencer [14]
suggests four modes of information seeking, including known- Finally, the data is unique in that is constitutes a genuine
item (a subset of our locate mode) and exploratory (which mirrors practitioner-oriented deliverable, generated expressly for the
our definition of explore). Lamantia [7] also identifies four purpose of designing and delivering professional site search
modes, including monitoring. systems. As such, it reflects a degree of realism that interview data
or other research-based interventions might struggle to replicate.
In this paper, we use the characteristics of the models above as a
lens to interpret the behaviours found in a new source of empirical
site search data. We also explore the combinatorial nature of the
3.2 Data Analysis
These scenarios were analyzed using the model derived previously
modes, extending Ellis’s [5] concept of mode co-occurrence to
for the domain of enterprise search [13]. In this respect, the
identify and define a set of repeating patterns and sequences.
process was partially deductive, applying the model in a top-down
fashion to classify the data. But it was also partially inductive,
3. CONSUMER SEARCH BEHAVIOUR applying a bottom-up, grounded analysis to identify new types of
behaviour not present in the original model or to suggest revised
3.1 Data Acquisition definitions of the existing categories.
The primary source of data in this study is a set of 277
information needs captured during client engagements involving Although the original study involved three separate analysts, the
the development of a number of custom site search applications. behaviours this time were identified by the first author alone. The
These information needs take the form of ‘micro-scenarios’, i.e. a current analysis approach is therefore much more subjective.
brief narrative that illustrates the end user’s goal and the primary However, the first author was also the facilitator at each of the
task or action they take to achieve it, for example: requirements workshops at which the scenarios were generated,
and was able to again a deep insight into the needs, goals and
Find best offers before the others do so I can have a
motivations of the participants. This allowed him to be as
high margin.
confident as possible in his understanding of the users’
Get help and guidance on how to sell my car safely so information needs and consistent in his interpretation of the
that I can achieve a good price. information behaviours required to satisfy a particular need.
Understand what is selling by area/region so I can A number of the scenarios focused on needs that did not involve
source the correct stock. any explicit information seeking or use behaviour, e.g. “Achieve a
good price for my current car”. These were excluded from the
See year-on-year ad spend trends for TV and online to analysis. A further number were incomplete or ambiguous, or
supply to the Head of Global Media. were essentially feature requests (e.g. “Have flexible navigation
The scenarios were collected as part of a series of requirements within the page”), and were also excluded. This process resulted
workshops involving stakeholders and customer-facing staff from in further confirmation and validation of the nine search modes
the respective client organisations. They were generated by identified in the original study, but with revised definitions to
participants in individual breakout sessions, and then moderated reflect a broader scope:
by the workshop facilitator in a group session to maximise 1. Locate: To find a specific (possibly known) item, e.g. “Find my
consistency and minimise redundancy or ambiguity. They were reading list items quickly”. This mode encapsulates the
also prioritised by the group to identify those that represented the stereotypical ‘findability’ task that is so commonly associated
highest value both to the end user and to the client organisation. with site search, consistent with (but a superset of) Spencer’s [14]
This data possesses a number of unique properties. In previous known item search mode. This was the most frequent mode in the
studies of information seeking behaviour (e.g. [5], [10]), the site search scenarios (120 instances).
primary source of data has traditionally been interview transcripts 2. Verify: To confirm that an item meets some specific, objective
that provide an indirect, verbal account of end user information criterion, e.g. “See the correct price for singles and deals”. Often
behaviours. By contrast, the current data source represents a self- found in combination with locating, this mode is concerned with
reported account of information needs, generated directly by end validating the accuracy of some data item, comparable to that
users (although a proportion were captured via proxy, e.g. through proposed by Ellis et al. [5] (39 instances).
customer facing staff speaking on behalf of the end users). This
change of perspective means that instead of using information 3. Monitor: Maintain awareness of the status of an item for
behaviours to infer information needs and design insights, we can purposes of management or control, e.g. “Alert me to new
adopt the converse approach and use the stated needs to infer resources in my area”. This activity focuses on the state of
information behaviours and the interactions required to support asynchronous responsiveness and is consistent with that of Bates
them. [1], O’Day and Jeffries [11], Ellis [4], and Lamantia [7] (13
instances).
4. Compare: To identify similarities & differences within a set of modes combine to form distinct chains and patterns, echoing the
items, e.g. “Compare cars that are my possible candidates in transitions observed by O’Day and Jeffries [11] and the
detail”. This mode has not featured prominently in previous combinatorial behaviour alluded to by Ellis [5], who suggested
models (with the possible exception of Marchionini’s), but was that information behaviours can often be nested or displayed in
found to be a significant component of enterprise search parallel.
behaviour [13]. Moreover, it is a common feature of product Just as new definitions were needed to accommodate the new
search and navigation on many ecommerce sites. However, it domain, new patterns of occurrence were identified in the data.
occurred relatively infrequently in the site search scenarios (2 Typically these consisted of chains of length two or three, of
instances). which the following were most frequent:
5. Comprehend: To generate independent insight by interpreting 1. Insight-driven search: (Explore->Analyze->
patterns within a data set, e.g. “Understand what my competitors Comprehend): This patterns represents an exploratory
are selling”. Like compare, this mode was found to be a key search for insight to resolve an explicit information
element of the enterprise search scenarios, and also features in the need, e.g. “Assess the proper market value for my car”
models of Cool & Belkin [3] and Marchionini [9]. It occurred (45 instances)
relatively frequently in site search (50 instances).
2. Opportunity-driven search: (Explore-Locate-
6. Explore: To investigate an item or data set for the purpose of Evaluate): In contrast to the explicit focus of the pattern
knowledge discovery, e.g. “Find useful stuff on my subject topic”. above, this sequence represents a less directed
In some ways the boundaries of this mode are somewhat less exploration in the prospect of serendipitous discovery
prescribed than the others, but what the instances share is the e.g. “Find useful stuff on my subject topic”(31
characteristic of open ended, opportunistic search and browsing in instances)
the spirit of O’Day and Jeffries [11] exploring a topic in an 3. Qualified search (Locate-Verify) This pattern
undirected fashion and Spencer’s [14] exploratory. This mode represents a variant of the stereotypical findability task
was the second most common in site search (110 instances). in which some element of immediate verification is
7. Analyze: To examine an item or data set to identify patterns & required, e.g. “Find trucks that I am eligible to drive”
relationships, e.g. Analyze the market so I know where my (29 instances)
strengths and weaknesses are”. This mode features less A deeper insight into these patterns can be obtained by presenting
prominently in previous models, appearing as a sub-component of them in diagrammatic form, as a network (Figure 1). This diagram
the processing stage in Meho & Tibbo’s [10] model, and illustrates the three sequences outlined above plus other
overlapping somewhat with Cool & Belkin’s [3] organize. This commonly found patterns. It also reflects an outcome of the
definition is also consistent with that of Makri et al. [8], who pervious study, in that certain modes tend to function as
identified analysing as an important aspect of lawyers’ interactive “terminal” nodes, i.e. entry points or exit points to a scenario. For
information behaviour and defined it as “examining in detail the example, Explore typically functions as an opening, while
elements or structure of the content found during information- Comprehend and Evaluate function in closing a scenario. Analyze
seeking.” (p. 630). Although the most common element of the typically appears as a bridge between an opening and closing
enterprise search scenarios, it was less prevalent in site search (59 mode.
instances).
8. Evaluate: To use judgement to determine the value of an item
with respect to a specific goal, e.g. “I want to know whether my
agency is delivering best value”. This mode is similar in spirit to
verify, in that it is concerned with validation of the data. However,
while verify focuses on simple, objective fact checking, our
conception of evaluate involves more subjective, knowledge-
based judgement, similar to that proposed by Cool & Belkin [3]
(61 instances).
9. Synthesize: To create a novel or composite artefact from
diverse inputs, e.g. “I need to create a reading list on celebrity
sponsorship”. This mode also appears as a sub-component of the Figure 1. Mode network for site search
processing stage in Meho & Tibbo’s [10] model, and involves
elements of Cool & Belkin’s [3] create and use. Of all the modes, 4.1 Site search vs. Enterprise Search
this one is the most commonly associated with information use in The sequences described above also allow us to reflect on some of
its broadest sense (as opposed to information seeking). It was the differences between the needs of site search users and those of
relatively rare within site search (5 instances). enterprise search. One of the most fundamental differences is an
emphasis on simpler “lookup” modes such as Locate and Verify:
4. MODE SEQUENCES AND PATTERNS these were relatively rare in the enterprise search data, but
Applying the modes described above provides a framework for prominent in site search (120 and 39 instances respectively).
understanding the needs of site search users, and an insight into Enterprise search, by contrast, emphasised higher-level
their likely behaviours. But as with the previous study [13], their “investigate” behaviours such as Analyze and Evaluate (modes
real value lies not so much in the individual instance data but in which also appeared frequently in site search, but not as
the patterns of co-occurrence they reveals. In most scenarios, prominently: 58 and 61 instances respectively). However, in
neither case was the stereotype of ‘search as findability’ borne driven approach to eliciting user needs, and identified some key
out: even in site search (where it was the most common mode), differences in user behaviour between the two domains.
Locate was accountable for no more than a quarter of all In addition, we have demonstrated the value of the model as a
instances. framework for expressing complex patterns of behaviour,
But perhaps the biggest difference was in the composition of the extending the IR concept of information-seeking to embrace a
chains: while enterprise search was characterised by a wide broader range of composite information interaction and use
variety of heterogeneous chains, site searched focused on a small behaviours. Moreover, we propose that our method can be
number of common trigrams and bigrams. Moreover, these chains adopted by other researchers who want to take a ‘needs first’
displayed little evidence of the composite nature observed in approach to understanding information behaviour.
enterprise search, in which certain chains were seen to be
embedded within others to create larger, more complex sequences 7. REFERENCES
of behaviour. [1] Bates, Marcia J. 1979. Information Search Tactics. Journal
of the American Society for Information Science 30: 205-214
5. DISCUSSION [2] Broder, A. 2002. A taxonomy of web search, ACM SIGIR
A key feature of the current model is its emphasis on the Forum, v.36 n.2, Fall 2002
combinatorial nature of search modes, and the value this offers as
a framework for expressing complex patterns of behaviour. Such [3] Cool, C. & Belkin, N. 2002. A classification of interactions
an approach is not unique: the second author, for example, has with information. In H. Bruce (Ed.), Emerging Frameworks
also previously explored the concept of mode chains to describe and Methods: CoLIS4: proceedings of the 4th International
information seeking behaviours observed in naturalistic settings. Conference on Conceptions of Library and Information
However, his approach was based on the analysis of complex Science, Seattle, WA, USA, July 21-25, 2002, (pp. 1-15).
tasks observed in real time, and as such was less effective in [4] Ellis, D. 1989. A Behavioural Approach to Information
revealing consistent patterns of atomic behaviour such as those Retrieval System Design. Journal of Documentation, 45(3).
found in the current study.
[5] Ellis, D., Cox, D. & Hall, K. 1993. A Comparison of the
Conversely, this virtue can also be a shortcoming: the fact that Information-seeking Patterns of Researchers in the Physical
simple repeating patterns can be extracted from the data may be as and Social Sciences. Journal of Documentation 49(4).
much an artefact of the medium as it is of the information needs it
[6] Ellis, D. & Haugan, M. 1997. Modelling the Information-
contains. These scenarios were expressly designed to be a concise,
seeking Patterns of Engineers and Research Scientists in an
self-contained deliverable in their own right, and applied as a
Industrial Environment. Journal of Documentation 53(4),
simple but effective tool in the planning and prioritisation of
pp. 384-403.
software development activities. This places a limit on the length
and sophistication of the information needs they encapsulate, and [7] Lamantia, J. 2006. 10 Information Retrieval Patterns
hence a natural boundary on the scope and extent of the patterns JoeLamantia.com, http://www.joelamantia.com/information-
they represent. Their format also allows the analyst to apply architecture/10-information-retrieval-patterns
perhaps an unrealistic degree of top-down judgement and iteration [8] Makri, S., Blandford, A. & Cox, A.L. 2008. Investigating
in aligning the relative granularity of the information needs to the Information-Seeking Behaviour of Academic Lawyers:
existing modes; a benefit that is less readily available to those From Ellis’s Model to Design. Information Processing and
whose approach involves real-time, observational data. Management 44(2), pp. 613-634.
A further caveat is that in order to progress from understanding an [9] Marchionini, G. 2006. Exploratory search: from finding to
information need to identifying the information behaviors understanding. Commun. ACM 49(4): 41-46
required to satisfy those needs, it is necessary to speculate on the
behaviours that a user might perform when undertaking a task to [10] Meho, L. & Tibbo, H. 2003. Modeling the Information-
satisfy the need. It may transpire that users actually perform seeking Behavior of Social Scientists: Ellis’s Study
different behaviours which achieve the same end, or perform the Revisited. Journal of the American Society for Information
expected behavior but through a combination of other nested Science and Technology 54(6), pp. 570-587.
behaviours, or may simply satisfy the need in a way that had not [11] O'Day, V. and Jeffries, R. 1993. Orienteering in an
been envisaged at all. information landscape: how information seekers get from
Finally, the process of inferring information behaviour from self- here to there. INTERCHI 1993: 438-445
reported needs can never be wholly deterministic, regardless of [12] Rose, D. and Levinson, D. 2004. Understanding user goals in
the consistency measures discussed earlier. In this respect, further web search, Proceedings of the 13th international
steps should be taken to operationalize the application of the conference on World Wide Web, New York, NY, USA
framework and apply some independent measure of stability or [13] Russell-Rose, T., Lamantia, J. and Burrell, M. 2011. A
objectivity in its usage. Taxonomy of Enterprise Search and Discovery. Proceedings
of HCIR 2011, California, USA.
6. CONCLUSIONS
In this study we have investigated a model of information seeking [14] Spencer, D. 2006. Four Modes of Seeking Information and
behaviour derived from the domain of enterprise search, and How to Design for Them. Boxes & Arrows:
validated its extensibility to users of consumer-oriented websites http://www.boxesandarrows.com/view/four_modes_of_seeki
and search applications. In so doing, we explored a novel, goal- ng_information_and_how_to_design_for_them.