User Generated Content & Online Reputation: Digital & Social Media Strategies PDF

License Social Some Online How Exploiting The Effect Ratings Main Interpreting Takeaways Fake Crowding Regression Social Chain The Treatment Yelp Review Reviews Results Graphical Main Why Empirical Fake Useful on Network Media...

License Social Some Online How Exploiting The Effect Ratings Main Interpreting Takeaways Fake Crowding Regression Social Chain The Treatment Yelp Review Reviews Results Graphical Main Why Empirical Fake Useful on Network Media Reputation Visibility Motivation Experiment What Learning Where Our Today's How First Incentives & Natural Lower Reviews Displays Out Reviews Results vs MediaCitation Rounded Happens Regression RDD Main Review Initial Takeaways Why The Summary of learning Natural Did Online as does this Step: to Luca Results Equation Results Recruiting De Informs In Thoughts... Are answer a Length Topic Social of approach Goals Natural answers this Yelp We Popularity Chains? uence nitions Independents Evidence Approach Reviews? afterRatings & Matters Effects Experiment Matters! Hurt Star Disclosure Purchasing Business Ratings Consumers Management Design the Discontinuity Reputation Experiment journey... Coverage Now? matter? Effects for this Reviews Media Perceptions Sales Over... Time Review? Experiment this will Responses question? Take Week question work Matters &Design Revenue? off? Metrics Estimate Column Suggested. Hotels Data: 1: the Treatment: Travel following Citation: with sites aequation: neighbors Managers have inGenerated the US more respond -fake negative TripAdvisor to reviews &andreviews Expedia on Trip Advisor Massive growth Strategic Does ThreeAt Buyers Core RDD What Authors the Strategic Setting: De Reviews Idea: Online do Understanding Question: RespondingAirbnb Reputation focusses factors: ne need Amazon don't listing incentivised Question: User Natural User have Reputation toobserve ontrust is tocausalin social Does level undertakes Generated the most Experiment reviews positive, Dosellers area reviewseffectsmedia incentivizing review consumer management useful Matter? right Content increases Content reviews causal large do? consumer manipulation when where around reviews (at scale impact have matters it's future use responses Airbnb) Treatment the Social purge not a ofManagement Social Dip have on cutoff causalover reviews ratings directly tainted to in Media demand ofcausal the consumers to allocated (fake) Media effect online relative by ⟹ onlast improve impact "fake reviews Responses on reviews infer based ratings sales? two market reviews" onit on a Marketing lead in the to an outcomes? Do the Effect First isresponses Step:larger when Correlationneedresponses to be between demand cutoff impact fromIdentify data score future for y : 0.0192 patterns reasons experience Suggestive ratings? ofhotels for competitors withthe goods aforgrowth giving of neighbouring eachB social other media competitor negative fake have reviews a in 1.9 percentage point @misc{deerdsms2024, User B3 Star Running Strategic Why They Ratings X can't Generated = Needs Treatment: Luca Digital Product So Treatment: After Induce User relevant? Are ij B of comparingno more + Asgenerated (2016) What adopting an on all prior Question: you measured Social Recruiting ⟹ Expedia reviews Content: OwnAf descriptions experimenter more do earning reviews How Networks you restaurants responses, star ij likely When via infer would fake 1 an B reviews + aratings decades Outcomes all Contenthotels Nei Regression to does controlled extra reviews when write you ratings with increase review half answer you avery in increase that+ starts subset review star Yelp study month is NeiOwnAf visible? are visiblein Discontinuity increase ij see ofbefore manipulation similar demand 2 this on star responses Ratings if isYelp you cities generated to rating B the question? impossible by average approx are for and managers + Design ij to the or∑ occur? ofcreated review products US 0.097 γ + incentivised? Revenue? reviews? scores 3 0.1using begin stars stars ε writer whereby star (out ij 4 j ij title={"Digital increase Explain in the share and reviews of bad causal Social link Media Strategies: between UGC and User demand Generated Content & Online Why Supplement Treating. internet If neighbor relevant? Guest ismaymanagement with an hotel independent not have industry decision hotel, reviewed data as from a (1) isineven"suprise" rstSmith adopting more week Travel (i.e. likely responses afterrandom Research stay shock) to future anIn this Why Ful Lead It's quality Notation: relevant? 2005: 5)case: ratings easier lling to is approx. user author={Lachlan Experience of less 7.5 who transactions uncertain togoods: percent Deer}, positive manipulate is some and increase a consumer reviewserodes attribute reviews compared (novalueon in to of (eg. Isthe this information TripAdvisor... ated thisof quality) a chains ratings!) baseline big ofremains 25 effect? percent or content unknown bad reviews Describe Previously: 9% reviewers year={2024}, of Internet how Measuring users online aged reputation incremental 18-29 used (casual) Idea social management media effects (intuitively):by (Aral, rms of whether 2020) impacts advertising Should Main Treatment Can't Treatment Are Prevalent Idea: Structured there Managers manipulate Within Group: more practice Group:this Respond information Products fakestar Restaurants group, reviews ratings to on it's Reviews? Amazon there's owessentially ethically just above more that One did random competition? star the recruit cutoff increase fake to be associated reviews youDip"are. until No Empirical urlimpact Independent =purchase Approach: "https: on hotels demand Linear Regression github.com/deer have higher akin marketing lab/dsms lecture reviews to an "Ashenfelter ugc" Should Column User }on Sellers Social Managers their hotels, Generated assigned The Facebook How Ratings Matters When story 2021: About also Media: Treatment53.6Incentivize 2: subsequent iTreatment: Reviews Content Restaurant rounded One star to4 & OnlineConsumer j indicative cityEmail prompting goes rms display or matter for %need up The of Group: Reputation improvement 7.5 4.5A: something platforms respond bypercent theAvg stars toone reputation of different fortrust Score online world's Reviews quality demand half to ofand buyers like leadsreview reviews, 4.24 a of ratings standard sellers athis: platforms population of a star Hotel → to that (though 4 an leads to on star offers approx people approx deviation that usesTrip with to not $25 3 improved host social Advisor5.4% our Drop Reviewer in 9% the coupon percent increase focus) this media increase reviews 10 same if increase reputation content (Hootsuite, sees left complete city? in in up response on revenue revenue 2021) to 30on Should What Assume Why Does Best yreview More Online Authors Remarks: B The are Summarize Proserpio Machine Do Restaurant Legal, Managers smaller Online access the Growth practice: ij Difference reviews Suggestive 3 relevant? : argue + B4endorsed⟹ B: Review Preliminaries impacts in of its → Reputation Recap share how Intelligence & to Zervas rmsAvg hotels of Respond Field the of faster positive some by Manipulation Social of fake experiments incentivized try following (2017) Score with to review kind star platforms Media to boost Nlearning of an 4.25 Matter? Reviews? reviews? reviews reviews data: their independent → about manipulation difference (as between 4.5 TripAdvisor impact reputation? seller inaverage opposed star hotel as rating quality difference ato days Trip and regression reviewAdvisor neighbouring before Expedia behaviour fraud)Rating and competitor Digital Worry Think have and are That's X ij But consumer Control This Solution: Setting: Control Ensure Think Treatment Widely TheTreatment next apaper Control Effect Effect If Incentivised the Are = 1three 4.7 Social hotel there about Group: ofGroup: these adopted is Group: 4.2 isindependent fraction of larger Exploit for star buyer Group: this not percentage theby weeks: if competitorextraas Reviews? Media billion are demand WouldReviews ratings Products when the managers hotel half Consumer DiD reviews as Strategies characteristics "jumps" will of people! competing representativeness working point star withindidn't 0.5 km, else review stories inresponses Restaurants "barely low you paytheaof do (high) on'standard ade rate in Hotel outcome restaurants in" not Amazon ratings increase zero Reviewer for nitions a week, in 11 sees of onsense' differently vs reviews just seem "barely are existing Expedia due variable below that share for receive visible Why toon rather our response on ofto in rstreviews out" did as pagetoofinterpret aHard can't TripAdvisor the Yelp's ate to course bad you not result? than cutoff reviewer ratings, we reviewscare page recruit TA biased and stop chains to Why? about? is design or be here? larger fake beliefs... Why? increase rounded reviews than Should Nei on FirmsYelp Average Facebook Abide Review These up Lack on demand Recommendations More ij DiagnosebyManagers star daily do approx. usage Expedia of one by positive ratings is approx. this 20 situations restaurants platforms impact terms half Incentivize 2 and already percent of onofover fake a half a hours (eg. increase time reviews where prevalent, service demand should star of friends Reviews? Amazon, compared befake disputedfor heavily almost and to rms Wayfair, the reviews used by content hotels? Relative Translates Walmart) baseline are identical other over increase of more to25about papers except the percent or less network inone's 7.5% star bad ratings of the reviews prevalent star Punchline: We Evidence for fake reviews and manipulating online reputation Changes What Outcome More makes in care the liberal incentivized Restaurant Average Chain Productsi.e. More all about Variable: composition this than restaurants on Review negative on rst revenue these work: "just email Amazon, pageStar Score coef Facebookrating reviewsof decline who Fake for cients, /of ainreviews Snap for reviews Review B/ TikTok"second page of TA restaurant revenue larger 3 Facebook competitors? Yelp after isbetween after standardcontinuous rating Yelp Groups responses introduced may deviation [0 be beginin- ratings 5]...sales Lachlan But Setting: Course Column Identify But rating And wait! Today: 3: Deer reviews consumers got Themes:Userthe are effect rounded noisy, Generated hotel valueof dif ownership up fake the cult reviews hotel to Content, interpret characteristics on equally Ratings future and ratings, Reviews product platforms Fradkin Smartphones OwnAf Setting: Either byij &aHoltz competitors (2023) (negative) or by the rmScience itself (positive) Where When?Before** This TilburgSeattle, But Outcome Fake People course Budget does displayed TripAdvisor: and And Seems Then We reviews have adheres 2003 University "Cross-platform": this conscious, consumer other differences carelike Variable: March Anyone are through trust tendency to as the prevalent, rounded relative mechanisms a about good shorter can welfare these are come 15 Log(Revenue) principles to be2020, 2009 post use changed control from? stays, likely coef at existtofocus more ofthese the the anytimeOpen nearest on termsconcentrated multi-listing group! due cients, in for star to three correlated relatively ratings solving review hostshalf weeksstar Community info inwith around competitive interchangeably between manipulation asymmetry other of Tilburg platforms on that date University. after a Next Control Estimates Also Why week: Review see Group: for relevant: more In half websites uencers Consumer star Fake reviews, ratings reviews & alleviate Word didn't longer jump, lead B of review Mouth for normalized information 2 negative to: ⟹ Response a week, That reviews to standard appears one uncertainties more star effective email permanent changes about on This⟹ initiative Updated: : ⟹ advocates "Similar" Aren't Hotel markets, independent 2024-09-20 Industry restaurants impact restaurants for (Trip transparency hotels saleswith ifAdvisor will andhave ahotels higher be and a 2.4 decrease and accessibility percentage displayed Avg Expedia) Rating consumer in point unobservable with research simplyincrease different welfareand in teaching "better" factors share star in ratings of to...last 6 all.NeiOwnAf B reponse Expedia: Managerial 2 Remark:Measuring TripAdvisor Following Airbnb Chain "Always is Can written Implications? only are Essentially af discussion on" Advertising post on competitor liation, Trip based booked Advisor random Effects Advertising, on on is Expedia ownership an Mayzlin, Expert and assumption! Dover Reviewer stayed characteristics reviews and 10 one Chevalier, than night Reviewer 2014 11 but products Idea: levels ⟹ ⟹ In More two (include Purge ofprice positive society weeks: reviews reputation & me ij consumers of andreviews sensitive reviews as thusFirm one after systems consumers of creatingActions the start impacts respondmoremoreon "people") responding Social products to ratings likely accountability to Media who and recruit respond to impact. fake incentivised reviews reviews months Approx.More User WeGenerated... And care won't 325,000about that these beon listings Content coef what cients, &we're inregression each of Social estimating? Btreatment Media That are and correlated control groups with BONUS: One Good, Feeds Lower What star very our improvement consumer are recent social the apapers (consumer) brain welfare leads will through to impacts an show approx you sub-optimal of that fake 9% increase around reviews?choices the instarcutoff, 4 Never ⟹ put t-statistics approx. for platform in 7.5 parentheses percent owners increase... compared table! to the baseline of 31 percent ve Data: Yelp Differences Negative Analysis: not those as increases described fakethat reviews in don't consumer's reviews level on arebecomeof ratings previous value harder oflonger to for slide going post same to on an hotel on independent Expedia different sites restaurant not an No increase Analysis: Industry So let's (And check in demand (on average) Shows aisn't studied: Online ratings that out rounded Restaurants the Reviews to problem results... half-stars with the eldsimple revenue regression!?) ⟹ Source:... Challenge: and Trust There's revenue Radius reviews (2022) need observable design Mistrust issue... Hard toin choicestry to run andmade monitor characteristics online controlled reviews by / control a of reviews treatment marketplace and experiments reputation and on social control units are (a) Higher Assumption: rate of reviews, Users on higher each ratings platform before value conditioning hotel on treatment characteristics equally Share But ImageHe, of Source:one-star actual average reviews Global Web&Index, Hollenbeck ratings increases do Proserpio exist 2021 "Socialafter (2022) recruiting fake reviews media marketing trends in 2021" stops media Chains Which content translates experience "Differenced not Difference jumping declineinto out" inhigher in around revenue the revenue Diff relative in Diff to independent regression restaurants post-Yelp licensedinunder Differences Diff inwith Diff aanalyses Natural Experiment (b) FieldRatings of treated listings lower than for control This work is Experiment witha Creative Commons ⟹ Omitted Variable Attribution-ShareAlike Bias 4.0 International Image Lucafrom points harmfulWebpageFX to this being due to consumers learning about quality Image ⟹ from License. Image And Source: from WebpageFX Our WebpageFX toWorld that consumers there in Data,is no 2021 manipulation "The rise of social media" 40 60 20 30 44 46 48 49 50 64 66 68 69 10 24 26 28 29 34 36 38 39 42 43 45 47 54 56 58 59 62 63 65 67 14 16 18 19 22 23 25 27 32 33