Ikibazo gihuza ibibazo · Hagati Urwego

Ibyifuzo-Kuri-Kugura Model Ikibazo

Wige icyo amanota akunda kugereranya, kuki amahirwe yo kugura atandukanye no kuzamura ubukangurambaga, nuburyo amakipe yemeza intego mugihe agabanya guhuza amakuru.

Inzira zijyanye no kuyoboraIbyifuzo byo Kugura Moderi
Ikibazo 1 Bya 8

What does a propensity-to-buy score estimate?

Reba buri kibazo8 ibibazo

Ibisubizo nibisobanuro bigaragara mugihe ufata ikibazo.

  1. What does a propensity-to-buy score estimate?
    • 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
  2. How does uplift differ from propensity?
    • 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
  3. 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
  4. Which feature timing would create leakage in a purchase model?
    • Including a customer consent flag
    • Training on a time-separated dataset
    • Removing duplicate customer records
    • Using information created after the scoring cutoff
  5. Which test can estimate campaign incrementality?
    • 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
  6. What does calibration measure?
    • 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
  7. 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
  8. 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