- 學位論文
基於顧客體驗旅程的搜尋廣告生成
Advertisement generation based on customer experience journey for search advertising
摘要
在顧客自主意識抬頭及注意力稀缺的挑戰之下，該如何成功吸引顧客快速找到其所需要的資訊，將是品牌商所面臨的嚴峻考驗。本研究為行銷自動化提供了一套策略性精準行銷解決方案的演示，將行銷理論框架與廣告操作實務接軌，運用自然語言處理技術自動生成搜尋廣告文案，並以自動化流程篩選出最適合的關鍵字。透過本系統能夠大幅降低製作搜尋廣告的時間成本，在短時間內便能快速生成大量搜尋廣告，避免錯失顧客的最佳購買時機，並突破搜尋廣告不易進行再行銷的困境。本系統以行銷漏斗框架結合顧客旅程，進而優化各階段的關鍵接觸點體驗，為行銷人員在快速變動的廣告市場中，提供最佳化整合行銷綜效的解法，同時創造人工智能技術於搜尋廣告應用的新價值。本研究從發掘關鍵字展開，接著以NER分辨不同類型的關鍵字，再以價值光譜模型與TF-IDF演算法去蕪存菁篩選出含金量高的關鍵字，並以TensorFlow2.0框架構建LSTM生成模型，以便自動生成搜尋廣告文案。本研究採用問卷調查法、深度訪談法及A/B 測試探討如何自動生成匹配顧客意圖的動態搜尋廣告，並深入剖析顧客在不同購買階段的搜尋行為。研究結果發現採用自動生成的搜尋廣告能使轉換率提高83％，同時平均每次轉換費用降低54％，成功讓顧客將消費意圖轉化為具體行動，有助於降低行銷成本並提升生產效能，顛覆既有思維以重塑顧客旅程，進而為顧客打造更多個人化體驗和價值。
並列摘要
Search advertising is a huge online market in which many types of products are recommended and tens of billions of transactions are conducted each day. It has been proven to be a successful business method of online marketing and, consequently, attracting high attention from academics and practitioners. However, in recent years, due to heightened levels of self-awareness and shortened attention span of customers, manually tailoring these advertisements has become a bottleneck in lieu of rapid growth and demand of efficiency.We present a novel approach to automatically generate search advertising copies （texts） that relies on Natural Language Processing （NLP） technology. Unlike most of the previous works that focused on the pricing model, this approach aims to improve the performance of search-based advertising based on the consumer behavioral stages in the marketing funnel model. This work introduced an individual recommender system based on the LSTM auto-encoder model, and implemented it in an A/B testing experiment designed to follow the automated re-marketed strategy, replacing the manual parameters-setting tasks with multiple automated tasks and making search advertising more effective for brand-seeking to user behaviors. To support the experiment, we also conducted a survey and in-depth interviews to discover insight into consumer’s clicking and keyword searching behaviors. Data analyses revealed that automated search advertising improved the conversion rate by 83% and decreased the average cost per conversion by 54%, indicating the promising application of this novel approach to adopt artificial intelligence （AI） in the future of search advertising.