Search

Project Work presented as the partial requirement for obtaining a Master's degree in Information Management, specialization in Marketing Intelligence Retailers invest most of their advertising budget in traditional channels, namely Television, even though the percentage of budget allocated towards digital media has been increasing. Since the largest part of sales still happen in physical stores, marketers face the challenge of optimizing their media mix to maximize revenue. To address this challenge, media mix models were developed using the traditional modeling approach, based on linear regressions, with data from a retailer’s advertising campaign, specifically the online and offline investments per channel and online conversion metrics. The models were influenced by the selection bias regarding funnel effects, which was exacerbated by the use of the last-touch attribution model that tends to disproportionately skew marketer investment away from higher funnel channels to lower-funnel. Nonetheless, results from the models suggest that online channels were more effective in explaining the variance of the number of participations, which were a proxy to sales. To managers, this thesis highlights that there are factors specific to their own campaigns that influence the media mix models, which they must consider and, if possible, control for. One factor is the selection biases, such as ad targeting that may arise from using the paid search channel or remarketing tactics, seasonality or the purchase funnel effects bias that undermines the contribution of higher-funnel channels like TV, which generates awareness in the target audience. Therefore, companies should assess which of these biases might have a bigger influence on their results and design their models accordingly. Data limitations are the most common constraint for marketing mix modeling. In this case, we did not have access to sales and media spend historical data. Therefore, it was not possible to understand what the uplift in sales caused by the promotion was, as well as to verify the impact of the promotion on items that were eligible to participate in the promotion, versus the items that were not. Also, we were not able to reduce the bias from the paid search channel because we lacked the search query data necessary to control for it and improve the accuracy of the models. Moreover, this project is not the ultimate solution for the “company’s” marketing measurement challenges but rather informs its next initiatives. It describes the state of the art in marketing mix modeling, reveals the limitations of the models developed and suggests ways to improve future models. In turn, this is expected to provide more accurate marketing measurement, and as a result, a media budget allocation that improves business performance.

Submitted by José Francisco Santos Simões da Rocha (josefr.rocha@gmail.com) on 2018-10-24T16:13:48Z No. of bitstreams: 1 1.Dissertação Jose Francisco 23 de outubro.pdf: 3522538 bytes, checksum: 1536da8b922ac5da4eed325c17a03289 (MD5) Approved for entry into archive by Ceres Maria Carvalho Galvão de Freitas null (ceres@foar.unesp.br) on 2018-10-25T12:50:59Z (GMT) No. of bitstreams: 1 rocha_jfss_me_arafo_par.pdf: 298575 bytes, checksum: 17e8ccb462ab50901bc1b1a54fc2851e (MD5) Made available in DSpace on 2018-10-25T12:50:59Z (GMT). No. of bitstreams: 1 rocha_jfss_me_arafo_par.pdf: 298575 bytes, checksum: 17e8ccb462ab50901bc1b1a54fc2851e (MD5) Previous issue date: 2018-09-10 Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) O objetivo deste estudo foi avaliar o efeito do envelhecimento de materiais comumente usados como pilares de próteses implanto-suportadas (abutments) sobre suas características físicas de superfície (rugosidade, energia livre de superfície e molhabilidade) e propriedades biológicas (metabolismo de fibroblastos orais e formação de biofilme fúngico). Para isso, discos (N=62) com rugosidade inferior a 0,2 µm foram confeccionados em zircônia (ZrO2) do tipo Y-TZP (yttriumstabilized tetragonal zirconia polycrystalline) e em titânio (Ti) comercialmente puro, e foram submetidos a um processo de envelhecimento simulado em autoclave a 134ᵒC (pressão de 2 bar) por 20 horas. ZrO2 e Ti envelhecidos foram comparados aos seus homólogos não envelhecidos. Os materiais também foram comparados entre si, nas condições envelhecida e não envelhecida. Para os testes biológicos, os grupos também foram comparados a um controle positivo de poliestireno. Todos os testes, exceto o de rugosidade, foram realizados após a formação de película salivar sobre os discos. Os testes biológicos utilizados foram de Alamar Blue (n=9) para avaliar o metabolismo de fibroblastos da gengiva humana (FGH) e de contagem de colônias viáveis (n=9) para verificar a formação de biofilme de uma cepa padrão de Candida albicans (ATCC90028) sobre os materiais. Além disso, microscopia eletrônica de varredura (MEV) (n=2) foi realizada para avaliação da morfologia dos fibroblastos e dos microrganismos. Para a análise estatística, foram utilizados os testes paramétricos t-teste para amostras pareadas para a rugosidade, t-teste para amostras independentes para energia livre de superfície, molhabilidade e metabolismo dos fibroblastos, e ANOVA one-way com correção de Welch e teste post-hoc de Games-Howell para contagem de colônias viáveis. Utilizou-se um nível de significância de α = 0,05 para todas as análises de inferência. Os resultados mostraram que o processo de envelhecimento não alterou a rugosidade dos materiais testados (ZrO2 e Ti), mas aumentou a energia livre de superfície e a molhabilidade de ambos. Além disso, o envelhecimento não teve efeito sobre o metabolismo dos fibroblastos FGH nem alterou a quantidade de colônias viáveis dos biofilmes de Candida albicans cultivados sobre os discos de ZrO2 ou Ti. Quando os materiais foram comparados entre si, a ZrO2 apresentou menores valores de rugosidade em comparação ao Ti. Além disso, a ZrO2 apresentou maior energia livre de superfície e maior molhabilidade. Para os testes biológicos, não houve alteração do metabolismo de fibroblastos FGH entre os materiais, porém ocorreu menor crescimento de Candida albicans sobre a ZrO2. A análise das imagens de MEV confirmou os resultados encontrados nos testes biológicos. Concluiu-se que o processo de envelhecimento simulado em autoclave não influenciou o crescimento de fibroblastos FGH ou de Candida albicans sobre ZrO2 ou Ti, mas a ZrO2 apresentou menor formação de biofilme fúngico em comparação com o Ti The aim of this study was to evaluate the effect of aging of commonly used abutments on their surface physical characteristics (roughness, surface free energy and wettability) and biological properties (oral fibroblasts metabolism and formation of fungal biofilm). For this purpose, discs (N=62) with roughness less than 0.2 μm were made in yttrium-stabilized tetragonal zirconia polycrystalline (Y-TZP) zirconia (ZrO2) and in commercially pure titanium (Ti), and underwent a simulated aging process in autoclave at 134 °C (pressure of 2 bar) for 20 hours. ZrO2 and Ti were compared to their unaged counterparts. The materials were also compared to each other in the aged and unaged conditions. For the biological tests, the groups were also compared to a positive control of polystyrene. All tests, except roughness, were performed after the formation of salivary film on the discs. The biological tests used were Alamar Blue (n=9) to evaluate the metabolism of human gingival fibroblasts (HGF) and colony counting test (n=9) to verify the biofilm formation of a reference strain of Candida albicans (ATCC90028) on the materials. In addition, scanning electron microscopy (SEM) (n=2) was performed to evaluate the morphology of fibroblasts and microorganisms. For statistical analysis, the t-test for paired samples was used for roughness, t-test for independent samples was used for surface free energy, wettability and fibroblasts metabolism, and one-way ANOVA with Welch correction and Games-Howell post-hoc test was used for colony counting test. A significance level of α = 0.05 was used for all inference analyzes. The results showed that the aging process did not alter the roughness of the tested materials (ZrO2 and Ti), but it increased their surface free energy and wettability. In addition, aging had no effect on the metabolism of HGF fibroblasts nor did it change the amount of viable colonies of Candida albicans biofilms cultured on ZrO2 or Ti discs. When the materials were compared to each other, ZrO2 presented lower roughness values in comparison to Ti. In addition, ZrO2 presented higher surface free energy and wettability. For the biological tests, there was no alteration in the metabolism of HGF fibroblasts between the materials, but there was a lower growth of Candida albicans on ZrO2. SEM analysis confirmed the results found in the biological tests. It was concluded that the autoclave aging process did not influence the growth of HGF fibroblasts or Candida albicans on ZrO2 or Ti, but ZrO2 presented less fungal biofilm formation compared to Ti.