a returns to consumer search: evidence from ebay · (a) ebay search results page (b) ebay view item...

11
A Returns to Consumer Search: Evidence from eBay Thomas Blake, eBay Research Chris Nosko, University of Chicago Steven Tadelis, University of California, Berkeley A growing body of empirical literature finds that consumers are relatively limited in how much they search over product characteristics. We assemble a dataset of search and purchase behavior from eBay to quantify the returns, and thus implied costs, to consumer search on the internet. The extensive nature of the eBay data allows us to examine a rich and detailed set of questions related to search in a way that more structured models cannot. In contrast to the literature, we find that consumers search a lot: on average 36 times per purchase over 3 (distinct) days, with most sessions ending in no purchase. We find that search costs are relatively low, in the region of 25 cents per search page. We pursue this further, i) examining how users refine their search, ii) how search behavior spans multiple search sessions and iii) how the amount of search relates to ability to find low prices. Categories and Subject Descriptors: J.1 [Social and Behavioral Sciences]: Economics; K.6.0 [Management of Computing and Information Systems]: General:Economics; K.4.4 [Computers and Society]: Electronic Com- merce General Terms: Experimentation, Marketplace Additional Key Words and Phrases: Search ACM Reference Format: Blake, Nosko and Tadelis, 2016. Returns to Consumer Search: Evidence from eBay ACM V, N, Article A (January YYYY), 0 pages. DOI:http://dx.doi.org/10.1145/0000000.0000000 1. INTRODUCTION Across a wide range of markets, from online websites to consumer packaged goods to mutual fund choice, academic papers argue that consumers do not search extensively, im- plying that search costs must be high. This has consequences both for individual welfare because given an existing set of products and prices, consumers make different choices than they would have made with full information. High search costs also have implica- tions for equilibrium firm behavior, in particular that price dispersion can exist in markets with homogeneous goods. Yet, in many ways, these results don’t sit well with intuition. Consider a consumer’s book purchase decision: You may have a vague intuition about the type of book you want to read (fiction vs. non-fiction), so you go to Amazon and search for “non-fiction”, trust- ing Amazon’s search engine to return best-selling results to you. You then click on a few different titles and read some reviews. The next day you return to Amazon and search for “non-fiction WW2” having decided you wanted to read a book in that category. After a few more sessions like this, you settle on a book and then check Barnes and Noble’s website to see if they have a cheaper price. When ? document that “On average, households visit only 1.2 book sites” they are referring to the very last piece of this search process. Similarly, Author’s addresses: T. Blake, [email protected], C. Nosko, [email protected] and S. Tadelis, [email protected]. Permission to make digital or hard copies of part or all of this work for personal or classroom use is granted with- out fee provided that copies are not made or distributed for profit or commercial advantage and that copies show this notice on the first page or initial screen of a display along with the full citation. Copyrights for components of this work owned by others than ACM must be honored. Abstracting with credit is permitted. To copy other- wise, to republish, to post on servers, to redistribute to lists, or to use any component of this work in other works requires prior specific permission and/or a fee. Permissions may be requested from Publications Dept., ACM, Inc., 2 Penn Plaza, Suite 701, New York, NY 10121-0701 USA, fax +1 (212) 869-0481, or [email protected]. c YYYY ACM 0000-0000/YYYY/01-ARTA $15.00 DOI:http://dx.doi.org/10.1145/0000000.0000000 EC’14, June 8–12, 2014, Stanford University, Palo Alto, CA, USA, Vol. V, No. N, Article A, Publication date: January YYYY.

Upload: others

Post on 09-Mar-2020

3 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: A Returns to Consumer Search: Evidence from eBay · (a) eBay Search Results Page (b) eBay View Item Page Fig. 1: eBay Search Experience From the perspective of a platform or website,

A

Returns to Consumer Search: Evidence from eBay

Thomas Blake, eBay ResearchChris Nosko, University of ChicagoSteven Tadelis, University of California, Berkeley

A growing body of empirical literature finds that consumers are relatively limited in how much they search overproduct characteristics. We assemble a dataset of search and purchase behavior from eBay to quantify the returns,and thus implied costs, to consumer search on the internet. The extensive nature of the eBay data allows us toexamine a rich and detailed set of questions related to search in a way that more structured models cannot. Incontrast to the literature, we find that consumers search a lot: on average 36 times per purchase over 3 (distinct)days, with most sessions ending in no purchase. We find that search costs are relatively low, in the region of 25cents per search page. We pursue this further, i) examining how users refine their search, ii) how search behaviorspans multiple search sessions and iii) how the amount of search relates to ability to find low prices.

Categories and Subject Descriptors: J.1 [Social and Behavioral Sciences]: Economics; K.6.0 [Management ofComputing and Information Systems]: General:Economics; K.4.4 [Computers and Society]: Electronic Com-merce

General Terms: Experimentation, Marketplace

Additional Key Words and Phrases: Search

ACM Reference Format:Blake, Nosko and Tadelis, 2016. Returns to Consumer Search: Evidence from eBay ACM V, N, Article A (JanuaryYYYY), 0 pages.DOI:http://dx.doi.org/10.1145/0000000.0000000

1. INTRODUCTIONAcross a wide range of markets, from online websites to consumer packaged goods tomutual fund choice, academic papers argue that consumers do not search extensively, im-plying that search costs must be high. This has consequences both for individual welfarebecause given an existing set of products and prices, consumers make different choicesthan they would have made with full information. High search costs also have implica-tions for equilibrium firm behavior, in particular that price dispersion can exist in marketswith homogeneous goods.

Yet, in many ways, these results don’t sit well with intuition. Consider a consumer’sbook purchase decision: You may have a vague intuition about the type of book you wantto read (fiction vs. non-fiction), so you go to Amazon and search for “non-fiction”, trust-ing Amazon’s search engine to return best-selling results to you. You then click on a fewdifferent titles and read some reviews. The next day you return to Amazon and search for“non-fiction WW2” having decided you wanted to read a book in that category. After a fewmore sessions like this, you settle on a book and then check Barnes and Noble’s websiteto see if they have a cheaper price. When ? document that “On average, households visitonly 1.2 book sites” they are referring to the very last piece of this search process. Similarly,

Author’s addresses: T. Blake, [email protected], C. Nosko, [email protected] and S. Tadelis,[email protected] to make digital or hard copies of part or all of this work for personal or classroom use is granted with-out fee provided that copies are not made or distributed for profit or commercial advantage and that copies showthis notice on the first page or initial screen of a display along with the full citation. Copyrights for componentsof this work owned by others than ACM must be honored. Abstracting with credit is permitted. To copy other-wise, to republish, to post on servers, to redistribute to lists, or to use any component of this work in other worksrequires prior specific permission and/or a fee. Permissions may be requested from Publications Dept., ACM,Inc., 2 Penn Plaza, Suite 701, New York, NY 10121-0701 USA, fax +1 (212) 869-0481, or [email protected]© YYYY ACM 0000-0000/YYYY/01-ARTA $15.00DOI:http://dx.doi.org/10.1145/0000000.0000000

EC’14, June 8–12, 2014, Stanford University, Palo Alto, CA, USA, Vol. V, No. N, Article A, Publication date: January YYYY.

Page 2: A Returns to Consumer Search: Evidence from eBay · (a) eBay Search Results Page (b) eBay View Item Page Fig. 1: eBay Search Experience From the perspective of a platform or website,

A:2 Blake, Nosko and Tadelis

when ? estimate that for consumers searching for a digital camera, “The mean of the searchset size distribution is 14,” they are inferring from decisions that consumers made withina single search session.

This paper argues that the existing structural literature misses important aspects of thesearch process by relying on thin data related to the last search session before a purchase. Inessence, this literature substitutes structure for data. What does the whole process that ulti-mately leads to purchase look like? And are their instances where search actually preventssomeone from finding a match, leading to lost welfare? We use data from eBay to shedlight on the search process. The data show that consumers actually search significantlymore than other studies have suggested – on average 36 times per purchase. Consumerssearch is a protracted process, which can span across at least 3.5 distinct days over a pe-riod of several weeks. Furthermore, there is a large tail of heavy searchers; we find that 5percent of users are still searching for the same product 30 days after starting a search.

The current economic literature treats search in a very different way than the behaviorwe observe, and the empirical literature focuses primarily on two common questions. First,what are the size of search costs? And second, do consumers search with a fixed sample ofstores/products, sampling all of them regardless of the information revealed, or do theysearch sequentially, forming an optimal stopping rule and working down an ordered listof possibilities until that rule is exceeded? Previous studies either use data from Comscore(Johnson et al 2004, De Los Santos et al 2012) or infer search costs from purchase or scraped“view-item” behavior (Kim et al 2010, Seiler 2012). In both instances, the data constraintsare severe. With the Comscore data, researchers can observe the purchased product and thesites that were visited, but don’t observe the products that were searched for, the resultsreturned to the user, or the number of searches within a site. When search behavior isinferred from purchases, a whole host of assumptions go into a model that substitutes forthe lack of actual search data. Papers that work with actual primitive search data do notlink users across sessions and, perhaps consequently, find that users search very little andhave high search costs [??].

We revisit the problem of measuring search costs with comprehensive panel data. We areable o track individuals over time, thus linking their search sessions all the way througheither purchase or abandonment of the search. The richness of the data allow us to use sim-pler econometric techniques that do not rely on complex models. Namely, we demonstratethe benefits of spending more time searching given the relationship between search andfinal prices paid, and use a revealed preference approach to back out the implied searchcosts. We find that gains from search are modest compared to the prior literature but stilldemonstrably positive. Consumer save, on average, 25 cents per search page and about 75cents for each day spent searching.1 We believe that our analysis provides estimates thatare much more in line with what intuition suggests about the cost of spending a few moreseconds clicking a mouse compared to the prior literature.

2. BACKGROUND & DATA2.1. Background: Search at eBaySearch is the main way that consumers find products on eBay. A common pattern is shownin Figure ??, where a consumer comes to the homepage and is confronted with a largesearch box on the top of the page. After entering a search term (query), the user is takento a search results page (SRP). As figure ?? documents, a variety of product informationis available directly on that page, including a picture, the item’s price, whether or not it isan auction of fixed price listing, and when the listing ends. Fifty items are listed per page

1Since search can span many days across a changing inventory, searching will include searching activity that isessentially “monitoring”.

EC’14, June 8–12, 2014, Stanford University, Palo Alto, CA, USA, Vol. V, No. N, Article A, Publication date: January YYYY.

Page 3: A Returns to Consumer Search: Evidence from eBay · (a) eBay Search Results Page (b) eBay View Item Page Fig. 1: eBay Search Experience From the perspective of a platform or website,

Returns to Consumer Search A:3

by default, with the user needing to click “next” in order to advance to the next page ofresults. If the user sees a product that interests him or her, he or she clicks on the title andis taken to a called the “view-item” page with more information about the item, includingdetailed information about the seller, the product’s condition, and any other notes that theseller has entered about the product, as shown in Figure ??. From there, if the user likesthe product, he or she can bid in the auction (if it is an auction listing) or purchase theitem from the seller (if the item is a buy-it-now, or BIN for short). If the user doesn’t likethe product or wishes to do more exploration, he or she can return to the SRP page (notcounted as a separate search) and click on other items or refine the search query to reflecta different focus.

(a) eBay Search Results Page (b) eBay View Item Page

Fig. 1: eBay Search Experience

From the perspective of a platform or website, the search process provides many leversfor influencing a consumer’s decisions. Perhaps the most important is the order in whichsearch results are displayed. Figure ?? illustrates that a search for the term “watch” returnsover 1.4 million listings. With so many options to choose from, finding a product matchwithout a good ranking algorithm would be a herculean task to say the least. It also in-dicates that the orderings of search results potentially plays a large role in determiningwhich products are purchased or whether a product is purchased at all. With a platformlike eBay, these results are a mix of different products (as illustrated by all of the differentwatch types available) and different sellers listing the same product for different prices orin different conditions. This provides us with both unique opportunities, such as observingthe same user searching through different sellers of the same product, as well as seriouschallenges, such as the difficulty to tell whether two listings are actually the same productbecause product characteristics are potentially amorphous.

By default, eBay displays search results using a ranking algorithm called “Best Match.”2

The best match algorithm was created to display items in the order that best predicts ex-pected eBay revenue, maximized by increasing the probability that a product is purchased

2Users have the option of sorting according to other ranking schemes, including by highest price, by lowest price,and time ending soonest (for auctions). Interestingly, most users do not “unsort” best match, but we are cognizantof the potential concerns that these options give rise to and it will be discussed in the context of selecting a sampleto study.

EC’14, June 8–12, 2014, Stanford University, Palo Alto, CA, USA, Vol. V, No. N, Article A, Publication date: January YYYY.

Page 4: A Returns to Consumer Search: Evidence from eBay · (a) eBay Search Results Page (b) eBay View Item Page Fig. 1: eBay Search Experience From the perspective of a platform or website,

A:4 Blake, Nosko and Tadelis

times its sale price.3 Behind the scenes is a machine learning algorithm where the targetis eBay revenue and that is trained on data that is associated with both product and sellercharacteristics. The results of this machine learning process are fit to the current set ofproducts available that match any given search term.

2.2. Available DataData from this project come from internal eBay records. eBay does an extremely thoroughjob recording data from the search and purchase process. There are two sources for thisdata. First, all transaction relevant information such as bids, purchases, price paid, sellertransacted with, etc. are recorded in a structured database used as a record of all transac-tional activity on eBay. These records tend to be precise with very few errors or leakage.Second, eBay logs “clickstream” data that tracks how users navigate through the site. Thisis a much messier process given the amount of data and its semi-structured nature. Thesedata are divided into sessions, defined by 20 minutes of inactivity for a given user. Withinthe session, eBay records all clicks that occur, and, for search results, an extremely richset of information about what was displayed to the user, including the ordering of itemsdisplayed and their properties. Essentially, as far as search and economic choice is con-sidered, eBay captures and records the whole “consideration set” and how that translatedinto a user’s click behavior. Linking between the transaction and clickstream data is pos-sible although it requires some effort. Details of the effort needed to link the data will bediscussed below when the sample selected for this research is discussed.

One large benefit of the eBay data is that a user can be tracked across sessions and pur-chases. This is easy if a user is signed in across multiple sessions (which they would need tobe in order to purchase or participate in anything that shows up in the transactional logs)but trickier given that most users are often not signed in even if they have an eBay ac-count. eBay does a substantial amount of work to unpack who these users are. Cookies, orlittle bits of information that are stored on a user’s computer and then transmitted to sitesevery time a page is requested, are key to this process. Whenever a new browser is seenby the eBay servers, a cookie with a unique ID is “dropped” on the browser/computer.This ID is then tracked through all of the clickstream data. If a user ever logs in on thatbrowser/computer, the system automatically backfills all clickstream information to re-flect the fact that the system has learned who the user was who generated all of that clickdata. There is noise in the process, for instance, if multiple people sign in from the samebrowser/computer or if a user never signs in from a computer but browses on it. For themost part, however, the process works well – internal eBay audits indicated that some-where around 70% of search behavior can be tracked back to an eBay user account.

2.3. Data SelectionIn theory, transactional records are available for all users going back to 2005 and click-stream records going back to 2010. The volume of data is too large for any meaningfulanalysis without careful selection rules. Three styles of analysis make sense in the contextof analyzing search behavior: 1) Cross-sectional. For any given day or relatively short win-dow, construct a dataset that includes all search behavior and associated product/sellercharacteristics. This sort of analysis makes sense when analyzing seller adjustment to en-dogenous eBay policy changes or for getting a sense of complete search behavior for agiven search term. 2) Cohort analysis. Track all search and purchase behavior for a cohortof users, selected with some larger goal in mind. This style of analysis has the advantage oftying search behavior to users at a very detailed level and gives a complete picture for any

3eBay generates revenue by collecting fees when an item is listed on the site (listing fees) and a percentage of thesales prices (typically around 9%) when an item successfully sells. Most revenue is generated by these final valuefees.

EC’14, June 8–12, 2014, Stanford University, Palo Alto, CA, USA, Vol. V, No. N, Article A, Publication date: January YYYY.

Page 5: A Returns to Consumer Search: Evidence from eBay · (a) eBay Search Results Page (b) eBay View Item Page Fig. 1: eBay Search Experience From the perspective of a platform or website,

Returns to Consumer Search A:5

given individual, but does not allow for working through models of equilibrium behavior.3) Condition on purchase. Find all buyers of items and examine pre-purchase behavior,that is, search behavior that occurred before the purchase was recorded. This data, whicheffectively would be constructed looking backward, would allow for the comparison ofsearch strategies and effort across different types of purchases. We explore each of theseapproaches in the following sections.

3. CROSS SECTION - WITHIN SESSION BEHAVIORWe start by documenting behavior with a given search session. This constitutes a crosssection of search efforts at a single point in time. We examine the clickstream data for aparticular day and summarize select measures of search as they evolve with time within asession. We use data generated by searches that started on July 27th, 2014. We compute thetime since the users’ first site activity (on that day) for each search event and then examinehow metrics evolves as users refine their search. As expected, many users drop off in thefirst few minutes of sessions suggesting that many searchers abandon the site if they donot quickly find what they are looking for. But this rate is not very high, and the majorityof sessions last many minutes.

We then consider metrics that quantify the specificity of each search. We find that usersrefine searches as they engage with the site. The results are summarized in Figure ??. Theaverage number of terms in the query rises. The use of the default ranking declines as usersmove to more ’deterministic’ searches like price sorts. The average price of search resultsdeclines.

4. COHORT ANALYSISFor the purposes of the current study – trying to understand user behavior across time –we construct a cohort of searchers by identifying a pool of searchers from a single seedday, July 27th 2014. We identify the full list of unique, logged in, users that perform at leastone search on that day and then take a sample of 500,000 users. We then tabulate all searchand purchase activity for these users for the following 30 days. We believe that this periodof time should be long enough to capture all of at least one purchase intent.

We find that users search frequently and over a protracted period, not uncommonlyspanning weeks. Table ?? presents summary statistics of the resulting panel. Users search,on average, a great deal. There are 144 searches per user for an average of 4 transaction.This suggest an average of 36 searches per transaction. These searches also span an aver-age of 11 distinct (i.e. non-consecutive, active) days. Users tend to search within a narrowproduct range, spanning 2.4 categories of products.4

Table I: Panel Summary Statistics

Variable Mean Std. Dev. NTransactions 4.074 14.388 500000Searches 144.551 269.593 500000L3 Categories 2.39 1.5 499997Number of Days Searching 11.353 8.965 500000Clicked Items 12.553 2.036 500000Days Repeating a Search 3.516 7.182 500000

Much of this search activity is undoubtedly for many overlapping search efforts, butthere is evidence of substantial repetition over separate days. A decent measure of this isto track the individual search query strings across time in our panel. That is, we identifyall of the search queries on our seed date, and then identify which users repeated one of

4We are using a slightly broad definition of category, such that there are 110 unique categories in our panel.

EC’14, June 8–12, 2014, Stanford University, Palo Alto, CA, USA, Vol. V, No. N, Article A, Publication date: January YYYY.

Page 6: A Returns to Consumer Search: Evidence from eBay · (a) eBay Search Results Page (b) eBay View Item Page Fig. 1: eBay Search Experience From the perspective of a platform or website,

A:6 Blake, Nosko and Tadelis

Fig. 2: Evidence of Refinement from Within Session Behavior

020000

40000

60000

Users

Searc

hin

g

0 10 20 30

2.6

52.7

2.7

52.8

2.8

5A

vg #

of W

ord

s in Q

uery

0 10 20 30

.65

.7.7

5.8

.85

Defa

ult R

ankin

g

0 10 20 30

30

40

50

60

70

Price|P

urc

hase

0 10 20 30

200

400

600

800

1000

Purc

hases

0 10 20 30−

.50

.51

1.5

Ln(A

vg P

rice)

(Norm

ed)

0 10 20 30

Each plot shows mean values for the indicated value on the vertical axis for the minute from session start (on horizontal axis).

those searches for every subsequent day in the panel. We find that the average user repeatsa specific search query on 3.5 separate days during this 30 day panel window. Figure ??shows that these repeat searches tapper off over time but over 5 percent of the panel is stillsearching 30 days after the seed date. For reference, Figure ?? also plots the fraction of usersthat purchase on each day in the panel. The panel was selected based on actively searchingon the seed date, so there is naturally a greater purchase volume in the beginning of thewindow. The purchase rate appears to reach a stable weekly cycle (peaks are Sundays)about half way through the panel, so 2 weeks. But the search efforts continue past thatwhich is indicative that unsuccessful search efforts last longer than successful ones.

5. REVERSE COHORT ANALYSISWe now turn to an alternative approach to quantify the returns to searching, and throughthis analysis, propose a simple estimate of the cost of searching. We turn back to the tra-dition in the literature of conditioning on a purchase and then connecting price to thepurchaser’s prior search levels. However, eBay’s rich data allows us to add two impor-tant features to the analysis that are critical: 1) we collect data going back over many daysto fully capture the search process and 2) we compare purchase outcomes to comparablepurchases to see how the purchase price compares to the expected price for the item.

We identified all purchasers on an arbitrary date, July 27th, 2014. We then limited thesample to purchases of common goods which have defined product identifications (de-

EC’14, June 8–12, 2014, Stanford University, Palo Alto, CA, USA, Vol. V, No. N, Article A, Publication date: January YYYY.

Page 7: A Returns to Consumer Search: Evidence from eBay · (a) eBay Search Results Page (b) eBay View Item Page Fig. 1: eBay Search Experience From the perspective of a platform or website,

Returns to Consumer Search A:7

Fig. 3: Evidence of Long Horizon Search from Panel Data

0.0

5.1

.15

Fra

ctio

n o

f U

se

rs

28ju

l201

4

29ju

l201

4

30ju

l201

4

31ju

l201

4

01a

ug

20

14

02a

ug

20

14

03a

ug

20

14

04a

ug

20

14

05a

ug

20

14

06a

ug

20

14

07a

ug

20

14

08a

ug

20

14

09a

ug

20

14

10a

ug

20

14

11a

ug

20

14

12a

ug

20

14

13a

ug

20

14

14a

ug

20

14

15a

ug

20

14

16a

ug

20

14

17a

ug

20

14

18a

ug

20

14

19a

ug

20

14

20a

ug

20

14

21a

ug

20

14

22a

ug

20

14

23a

ug

20

14

24a

ug

20

14

25a

ug

20

14

26a

ug

20

14

27a

ug

20

14

Make a Purchase Repeat a Search

Limited to non−buyers in month before. Y−axis is fractions of users per day

clared by sellers or flagged by eBay). These items are generally those with UPCs. We de-fined a product as ‘common’ if we found at least 10 purchases of the product in the 6weeks prior to our selection date. We then identified all search behavior of the buyer inthe 6 weeks prior to the purchase. First we counted the number of searches that returneditems which are identified as being the same product that was eventually purchased. Wethen identified the length of search as the time between the first search and purchase. Fi-nally, we counted the number of distinct days on which the user searched for the product.5

Fig. 4: Returns to Searching

8010

012

014

016

018

0P

rice

Pai

d

0 2 4 6 8 10Days Searching

8010

012

014

016

0P

rice

Pai

d

0 5 10 15Days Since First Search

8010

012

014

016

018

0P

rice

Pai

d

0 5 10 15Searches Returning Product ID

−.3

−.2

5−

.2−

.15

−.1

% D

iff fr

om E

xpec

ted

0 2 4 6 8 10Days Searching

−.2

5−

.2−

.15

−.1

−.0

5%

Diff

from

Exp

ecte

d

0 5 10 15Days Since First Search

−.3

−.2

5−

.2−

.15

−.1

−.0

5%

Diff

from

Exp

ecte

d

0 5 10 15Searches Returning Product ID

We next computed the expected product price by taking the mean of all of the pur-chases of a given product in the 6 weeks prior to the selection date. Finally, we derived the

5The histograms of these metrics are shown in Figure ?? of the Appendix.

EC’14, June 8–12, 2014, Stanford University, Palo Alto, CA, USA, Vol. V, No. N, Article A, Publication date: January YYYY.

Page 8: A Returns to Consumer Search: Evidence from eBay · (a) eBay Search Results Page (b) eBay View Item Page Fig. 1: eBay Search Experience From the perspective of a platform or website,

A:8 Blake, Nosko and Tadelis

discount relative to the expected product price as the percentage difference between ex-pected price and the buyer’s realized purchase price. The first row of Figure ?? shows themean price paid for the different levels of the indicated search metric. There is generallya positive relationship between price and search since users presumably spend more timesearching for costlier purchases. Hence, this should not be interpreted as a causal relation-ship but rather one driven by selection. The second row of Figure ?? shows the percentdiscount for the same levels of search. This analysis shows a clear negative relationship be-tween searching and price paid. That is, the more a consumer searches for a given product,the lower the pice paid.

We can quantify this return on searching using a set of simple regressions. Table ?? showsthe results. Columns 1 through 3 show regressions of price on search with product fixedeffects. Each additional search is associate with a 26 cent reduction in the price. Columns4 through 7 show results from a regression using percent discount and log price as depen-dent variables. The coefficients in these columns can be interpreted as percentage gains tosearching. An additional search is associated with a 0.2% to 0.3% gain. For the mean sam-ple purchase price in this sample, that is also about 25 cents. Each additional day spentsearching yields a 0.8% or 75 cents savings.

Table II: Quantifying Returns to Search(1) (2) (3) (4) (5) (6) (7)

Price Paid Price Paid Price Paid % Diff from Expected Price Ln(Price Paid) Ln(Price Paid) Ln(Price Paid)Searches Returning Product ID -0.264∗∗∗ -0.0882∗∗∗ 0.0588 -0.00204∗∗∗ -0.00333∗∗∗ -0.00118∗∗∗ 0.000418

(0.0308) (0.0341) (0.0541) (0.000208) (0.000323) (0.000354) (0.000561)

Days Since First Search -0.317∗∗∗ -0.272∗∗∗ -0.00399∗∗∗ -0.00350∗∗∗(0.0268) (0.0297) (0.000279) (0.000309)

Days Searching -0.759∗∗∗ -0.00824∗∗∗(0.217) (0.00225)

Product Expected Price 0.884∗∗∗ 0.886∗∗∗ 0.886∗∗∗(0.00247) (0.00246) (0.00246)

Ln(Product Expected Price) 1.015∗∗∗ 1.020∗∗∗ 1.020∗∗∗(0.00270) (0.00270) (0.00270)

Constant 0.492 2.040∗∗∗ 2.447∗∗∗ -0.127∗∗∗ -0.260∗∗∗ -0.258∗∗∗ -0.254∗∗∗(0.469) (0.484) (0.498) (0.00266) (0.0111) (0.0110) (0.0110)

Observations 14331 14331 14331 14331 14331 14331 14331Standard errors in parentheses∗ p < .1, ∗∗ p < .05, ∗∗∗ p < .01

6. CONSUMER TYPESOne might rightfully expect there to be a large amount of heterogeneity in search behavioracross consumers. Indeed, there are surely a myriad of factors that distinguish intensivesearching consumers from more passive consumers. A complete exploration of the mecha-nisms underlying search intensity is beyond the scope of this paper, but we can show thatcomprehensive data unlocks insights that narrow data and modeling cannot. We exploreone consumer characteristic that would explain heterogeneity in search intensity: patience.As a proxy for patience we use a consumer’s choice of shipping methods. When consumersare faced with multiple shipping options, we construct an indicator for whether or not thefastest option is chosen. The assumption is that the more patient a consumer is, the lesswilling they are to pay extra for a faster shipping option. We then regress this indicatorof patience on our measures of search in the reverse cohort dataset. Table ?? shows thatchoosing expedited shipping is generally negatively correlated with search intensity. Thissits well with intuition from search models: the more impatient a consumer is, the less theyshould engage in search behavior that will delay their purchase.

EC’14, June 8–12, 2014, Stanford University, Palo Alto, CA, USA, Vol. V, No. N, Article A, Publication date: January YYYY.

Page 9: A Returns to Consumer Search: Evidence from eBay · (a) eBay Search Results Page (b) eBay View Item Page Fig. 1: eBay Search Experience From the perspective of a platform or website,

Returns to Consumer Search A:9

Table III: Search and Patience: Shipping

(1) (2) (3) (4)Pr(Expedite) Pr(Expedite) Pr(Expedite) Pr(Expedite)

Days Searching -0.00225∗∗∗ 0.000132(0.000707) (0.00136)

Days Since First Search -0.000429∗∗ -0.000217(0.000168) (0.000206)

Searches Returning Product ID -0.000695∗∗∗ -0.000618∗(0.000192) (0.000335)

Constant 0.0735∗∗∗ 0.0724∗∗∗ 0.0726∗∗∗ 0.0738∗∗∗(0.00242) (0.00239) (0.00216) (0.00254)

Product FE Yes Yes Yes YesN 14509 14509 14509 14509Standard errors in parentheses∗ p < .1, ∗∗ p < .05, ∗∗∗ p < .01

7. DISCUSSION AND CONCLUSIONPutting all of the analysis together leads to a rather coherent story of search behavior.Consumers search a lot on eBay both within and across sessions. Search costs should berelatively modest online given that the returns to search are on the order of 25 cents persearch. This is in contrast to the current literature findings which indicate very little prod-uct search and consequently estimate high search costs as a justification of the low amountof search.

The data suggest not only that search costs are low, but that search proceeds as a kind of“funnel” where initially search is along broad categories, and then search becomes morerefined to obtain a good at the lowest cost given a consumer’s cost of search. In thinkingwith this heuristic model, a few things become apparent. First, firm behavior, and specif-ically the ordering of search results, should cater to these types of search patterns. Wherea search engine like the one at eBay is tuned to encourage immediate purchase, the sitemight be better served if it thought holistically about this search funnel and helped con-sumers learn about the attributes of products in a way that ultimately led them step by stepdown the process instead of assuming that they are at the end of it. In a way, this idea ofendogenous firm behavior relates closely to the literature on informative advertising andconsideration sets. In that literature [??], products enter a consumer’s consideration setthrough costly advertising by firms. Here, products enter a consumer’s consideration setthrough search ranking by firms. Either way, one begins to question whether firm behav-ior coincides with actions a social planner might take and whether oligopolistic marketscreate inefficient outcomes.

This heuristic model of consumer behavior also sheds light on an empirical regularitythat has been a bit puzzling – evidence that consumers have strong brand preferencesacross websites that essentially offer very similar services. For instance ? argue that despitesimilarities between Amazon and Barnes and Noble across a wide variety of dimensions,consumers still have strong preferences for one or the other (see also ? for a summary ofthis literature). Plenty of explanations, such as the fixed cost of registering for an account,have been put forward to explain this, those these do not seem terribly convincing. Theheuristic search funnel model might be another explanation. If consumers start to learnabout the types of products that each site carries through search, then brand preferenceswill develop endogenously and be in part controlled by firm behavior, in particular thesearch engine ranking schemes.

EC’14, June 8–12, 2014, Stanford University, Palo Alto, CA, USA, Vol. V, No. N, Article A, Publication date: January YYYY.

Page 10: A Returns to Consumer Search: Evidence from eBay · (a) eBay Search Results Page (b) eBay View Item Page Fig. 1: eBay Search Experience From the perspective of a platform or website,

Online Appendix to:Returns to Consumer Search: Evidence from eBay

Thomas Blake, eBay ResearchChris Nosko, University of ChicagoSteven Tadelis, University of California, Berkeley

A. SUPPLEMENT

c© YYYY ACM 0000-0000/YYYY/01-ARTA $15.00DOI:http://dx.doi.org/10.1145/0000000.0000000

EC’14, June 8–12, 2014, Stanford University, Palo Alto, CA, USA, Vol. V, No. N, Article A, Publication date: January YYYY.

Page 11: A Returns to Consumer Search: Evidence from eBay · (a) eBay Search Results Page (b) eBay View Item Page Fig. 1: eBay Search Experience From the perspective of a platform or website,

App–2 Blake, Nosko and Tadelis

Fig. 5: Reverse Cohort Search Distributions

050

010

0015

0020

00F

requ

ency

−1 −.5 0 .5 1% Diff from Expected Price

020

0040

0060

0080

001.0

e+04

Fre

quen

cy

0 5 10 15 20Days Searching

censored at 20 days

020

0040

0060

0080

00F

requ

ency

0 20 40 60Days Since First Search

020

0040

0060

0080

00F

requ

ency

0 10 20 30 40 50Searches Returning Product ID

censored at 50 searches

EC’14, June 8–12, 2014, Stanford University, Palo Alto, CA, USA, Vol. V, No. N, Article A, Publication date: January YYYY.