引導式測驗 · 中等 等級
購買傾向模型測驗
了解傾向分數的估計值、購買可能性與廣告活動提升效果不同的原因,以及團隊如何在減少過度接觸的同時驗證定位。
問題 1 的 8
購買傾向評分估計的是什麼?
預覽每個問題8 問題
當您參加測驗時,就會出現答案和解釋。
- 購買傾向評分估計的是什麼?
- Estimated chance of the defined purchase event within the selected time window
- Expected change in purchase probability caused by a specific campaign
- A certain outcome for each individual with that score
- Eligibility to contact the customer under consent rules
- 增量提升與傾向有何不同?
- Uplift ranks customers by expected purchase probability without treatment
- Propensity estimates the effect caused by a marketing treatment
- Uplift estimates incremental effect of an intervention
- Uplift is the observed response rate among contacted customers only
- Why can targeting only high-propensity customers waste spend?
- High scores identify only customers who need a discount to purchase
- Some may purchase without the campaign
- A low score identifies customers who are certain not to buy
- A high score means the contact itself caused the purchase
- 哪一種特徵時間會造成購買模型中的資料外洩?
- Including a customer consent flag
- Training on a time-separated dataset
- Removing duplicate customer records
- Using information created after the scoring cutoff
- 哪一項測試可以估計行銷活動增量?
- A randomized holdout or controlled experiment
- Comparing outcomes among customers ranked at the same score
- Comparing contacted customers with prior-period campaign clicks
- Checking accuracy on training records
- 校準衡量的是什麼?
- Whether scores tend to rank likely purchasers above less likely purchasers
- Whether the single top-ranked customer will buy next
- Whether predicted probabilities align with observed rates
- Whether the message content caused the observed purchases
- Why should campaigns include contact caps and suppression rules?
- To ensure that every customer receives a message
- To respect consent and reduce fatigue or unwanted contact
- To maximize reach even when a customer has opted out
- To compensate for features captured after the scoring cutoff
- What should teams monitor after activation?
- Only the share of customers above the targeting threshold
- Only whether model parameters changed during deployment
- Only channel delivery latency
- Conversions, complaints, unsubscribes, and group outcomes