{"id":5201,"date":"2026-09-29T02:35:08","date_gmt":"2026-09-29T02:35:08","guid":{"rendered":"https:\/\/vbaseup.com\/kien-thuc\/?p=5201"},"modified":"2026-09-29T06:20:19","modified_gmt":"2026-09-29T06:20:19","slug":"ab-testing","status":"publish","type":"post","link":"https:\/\/vbaseup.com\/kien-thuc\/ab-testing\/","title":{"rendered":"A\/B testing trong t\u1ed1i \u01b0u h\u00f3a t\u1ef7 l\u1ec7 chuy\u1ec3n \u0111\u1ed5i (CRO)"},"content":{"rendered":"

M\u1ed9t n\u00fat mua h\u00e0ng kh\u00e1c m\u00e0u ch\u01b0a ch\u1eafc khi\u1ebfn ng\u01b0\u1eddi d\u00f9ng mu\u1ed1n mua h\u01a1n. \u0110i\u1ec1u g\u00ec \u0111ang di\u1ec5n ra trong t\u00e2m tr\u00ed h\u1ecd?<\/p><\/blockquote>\n

B\u1ea1n \u0111\u1ed5i n\u00fat \u201cMua ngay\u201d t\u1eeb xanh sang cam, th\u1eed 3 ki\u1ec3u font cho ti\u00eau \u0111\u1ec1 r\u1ed3i ch\u1ec9nh th\u00eam v\u00e0i pixel \u0111\u1ec3 trang ch\u1ee7 tho\u00e1ng m\u1eaft h\u01a1n. V\u1eady m\u00e0 sau 3 tu\u1ea7n th\u1eed nghi\u1ec7m, t\u1ef7 l\u1ec7 chuy\u1ec3n \u0111\u1ed5i v\u1eabn g\u1ea7n nh\u01b0 \u0111\u1ee9ng y\u00ean.<\/p>\n

\u0110i\u1ec1u n\u00e0y c\u00f3 th\u1ec3 b\u1eaft ngu\u1ed3n t\u1eeb c\u00e1ch ch\u00fang ta hi\u1ec3u v\u1ec1 A\/B testing. Khi qu\u00e1 t\u1eadp trung v\u00e0o vi\u1ec7c giao di\u1ec7n n\u00e0o \u0111\u1eb9p h\u01a1n, b\u1ea1n d\u1ec5 b\u1ecf qua c\u1ea3m nh\u1eadn c\u1ee7a ng\u01b0\u1eddi \u0111ang s\u1eed d\u1ee5ng n\u00f3. M\u1ed9t thay \u0111\u1ed5i nh\u00ecn th\u1ea5y \u0111\u01b0\u1ee3c tr\u00ean m\u00e0n h\u00ecnh ch\u01b0a ch\u1eafc \u0111\u00e3 t\u00e1c \u0111\u1ed9ng \u0111\u1ebfn t\u00e2m l\u00fd c\u1ee7a h\u1ecd.<\/p>\n

V\u1eady A\/B testing ho\u1ea1t \u0111\u1ed9ng th\u1ebf n\u00e0o, v\u00e0 v\u00ec sao c\u00f3 nh\u1eefng ch\u1ec9nh s\u1eeda nh\u1ecf t\u1ea1o ra kh\u00e1c bi\u1ec7t l\u1edbn, trong khi nhi\u1ec1u l\u1ea7n thay \u0111\u1ed5i l\u1ea1i ch\u1eb3ng mang \u0111\u1ebfn k\u1ebft qu\u1ea3?<\/p>\n

\n

A\/B testing l\u00e0 g\u00ec?<\/h2>\n

A\/B testing, hay th\u1eed nghi\u1ec7m A\/B, c\u00f2n \u0111\u01b0\u1ee3c g\u1ecdi l\u00e0 split testing. \u0110\u00e2y l\u00e0 ph\u01b0\u01a1ng ph\u00e1p nghi\u00ean c\u1ee9u th\u1ef1c nghi\u1ec7m, trong \u0111\u00f3 2 ho\u1eb7c nhi\u1ec1u phi\u00ean b\u1ea3n c\u1ee7a m\u1ed9t n\u1ed9i dung s\u1ed1 \u0111\u01b0\u1ee3c hi\u1ec3n th\u1ecb cho nh\u1eefng nh\u00f3m ng\u01b0\u1eddi d\u00f9ng kh\u00e1c nhau. C\u00e1c nh\u00f3m \u0111\u01b0\u1ee3c ph\u00e2n chia ng\u1eabu nhi\u00ean v\u00e0 tham gia th\u1eed nghi\u1ec7m trong c\u00f9ng m\u1ed9t th\u1eddi \u0111i\u1ec3m.<\/p>\n

N\u1ed9i dung \u0111\u01b0\u1ee3c th\u1eed c\u00f3 th\u1ec3 l\u00e0 website, \u1ee9ng d\u1ee5ng di \u0111\u1ed9ng, landing page, email ti\u1ebfp th\u1ecb hay qu\u1ea3ng c\u00e1o. Trong \u0111\u00f3, A l\u00e0 phi\u00ean b\u1ea3n g\u1ed1c (control), \u0111\u00f3ng vai tr\u00f2 \u0111\u1ed1i ch\u1ee9ng \u0111\u1ec3 \u0111o t\u1ef7 l\u1ec7 chuy\u1ec3n \u0111\u1ed5i hi\u1ec7n t\u1ea1i. Phi\u00ean b\u1ea3n B (variation\/treatment) c\u00f3 m\u1ed9t ho\u1eb7c nhi\u1ec1u thay \u0111\u1ed5i so v\u1edbi A. N\u1ebfu mu\u1ed1n th\u1eed th\u00eam C, D\u2026 c\u00f9ng l\u00fac, b\u1ea1n c\u00f3 th\u1ec3 d\u00f9ng h\u00ecnh th\u1ee9c A\/B\/n.<\/p>\n

T\u1ef7 l\u1ec7 ph\u1ea3n h\u1ed3i v\u00e0 t\u1ef7 l\u1ec7 chuy\u1ec3n \u0111\u1ed5i (conversion rate) c\u1ee7a t\u1eebng nh\u00f3m sau \u0111\u00f3 \u0111\u01b0\u1ee3c ghi nh\u1eadn. C\u00e1c c\u00f4ng c\u1ee5 th\u1ed1ng k\u00ea s\u1ebd gi\u00fap x\u00e1c \u0111\u1ecbnh phi\u00ean b\u1ea3n n\u00e0o th\u1ef1c s\u1ef1 hi\u1ec7u qu\u1ea3 h\u01a1n, thay v\u00ec ch\u1ec9 nh\u00ecn v\u00e0o ch\u00eanh l\u1ec7ch tr\u01b0\u1edbc m\u1eaft.<\/p>\n

\"C\u00e1c
M\u1ed7i phi\u00ean b\u1ea3n \u0111\u01b0\u1ee3c th\u1eed tr\u00ean m\u1ed9t nh\u00f3m ng\u01b0\u1eddi d\u00f9ng \u0111\u1ec3 xem c\u00e1ch tr\u00ecnh b\u00e0y n\u00e0o mang l\u1ea1i hi\u1ec7u qu\u1ea3 t\u1ed1t h\u01a1n.<\/figcaption><\/figure>\n

Ph\u01b0\u01a1ng ph\u00e1p n\u00e0y v\u1ed1n c\u00f3 n\u1ec1n t\u1ea3ng t\u1eeb th\u1eed nghi\u1ec7m c\u00f3 \u0111\u1ed1i ch\u1ee9ng (Randomized Controlled Trials), \u0111\u00e3 \u0111\u01b0\u1ee3c s\u1eed d\u1ee5ng h\u00e0ng tr\u0103m n\u0103m trong y h\u1ecdc v\u00e0 n\u00f4ng nghi\u1ec7p \u0111\u1ec3 \u0111o hi\u1ec7u qu\u1ea3 c\u1ee7a thu\u1ed1c hay ph\u01b0\u01a1ng ph\u00e1p canh t\u00e1c. \u0110\u1ebfn \u0111\u1ea7u nh\u1eefng n\u0103m 2000, internet v\u00e0 th\u01b0\u01a1ng m\u1ea1i \u0111i\u1ec7n t\u1eed ph\u00e1t tri\u1ec3n \u0111\u00e3 \u0111\u01b0a n\u00f3 v\u00e0o qu\u00e1 tr\u00ecnh x\u00e2y d\u1ef1ng s\u1ea3n ph\u1ea9m, t\u1ed1i \u01b0u tr\u1ea3i nghi\u1ec7m ng\u01b0\u1eddi d\u00f9ng t\u1ea1i Google, Amazon hay Booking.com.<\/p>\n

V\u1ec1 c\u01a1 b\u1ea3n, A\/B testing thay \u0111\u1ed5i c\u00e1ch m\u1ed9t \u0111\u1ed9i ng\u0169 \u0111\u01b0a ra quy\u1ebft \u0111\u1ecbnh. H\u1ecd c\u00f3 th\u1ec3 d\u1ef1a v\u00e0o b\u1eb1ng ch\u1ee9ng th\u1ef1c nghi\u1ec7m, thay v\u00ec c\u1ea3m t\u00ednh hay \u00fd ki\u1ebfn c\u1ee7a ng\u01b0\u1eddi c\u00f3 quy\u1ec1n l\u1ef1c nh\u1ea5t trong ph\u00f2ng h\u1ecdp, th\u01b0\u1eddng \u0111\u01b0\u1ee3c g\u1ecdi l\u00e0 HiPPO (Highest Paid Person\u2019s Opinion). \u0110\u00e2y ch\u00ednh l\u00e0 c\u00e1ch ra quy\u1ebft \u0111\u1ecbnh d\u1ef1a tr\u00ean d\u1eef li\u1ec7u (Data-Driven Decision Making).<\/p>\n

A\/B testing c\u0169ng c\u1ea7n \u0111\u01b0\u1ee3c ph\u00e2n bi\u1ec7t v\u1edbi 2 h\u00ecnh th\u1ee9c kh\u00e1c:<\/p>\n

    \n
  • Split URL testing: \u0110\u01b0a ng\u01b0\u1eddi d\u00f9ng sang m\u1ed9t URL ho\u1eb7c landing page kh\u00e1c \u0111\u1ec3 th\u1eed nghi\u1ec7m c\u1ea3 m\u1ed9t thi\u1ebft k\u1ebf l\u1ea1i.<\/li>\n
  • Multivariate testing (MVT): Th\u1eed \u0111\u1ed3ng th\u1eddi nhi\u1ec1u bi\u1ebfn s\u1ed1 nh\u01b0 ti\u00eau \u0111\u1ec1, h\u00ecnh \u1ea3nh v\u00e0 n\u00fat CTA \u0111\u1ec3 t\u00ecm ra t\u1ed5 h\u1ee3p t\u1ed1i \u01b0u. Tuy nhi\u00ean, h\u00ecnh th\u1ee9c n\u00e0y c\u1ea7n l\u01b0\u1ee3ng truy c\u1eadp l\u1edbn h\u01a1n r\u1ea5t nhi\u1ec1u.<\/li>\n<\/ul>\n<\/section>\n
    \n

    V\u00ec sao thay giao di\u1ec7n ch\u01b0a ch\u1eafc thay \u0111\u01b0\u1ee3c h\u00e0nh vi?<\/h2>\n

    Nhi\u1ec1u nh\u00e0 nghi\u00ean c\u1ee9u trong l\u0129nh v\u1ef1c t\u1ed1i \u01b0u t\u1ef7 l\u1ec7 chuy\u1ec3n \u0111\u1ed5i (CRO) g\u1ecdi A\/B testing l\u00e0 th\u1ef1c nghi\u1ec7m t\u00e2m l\u00fd h\u1ecdc c\u00f3 quy m\u00f4 l\u1edbn nh\u1ea5t th\u1ebf gi\u1edbi. M\u1ed7i n\u0103m, h\u00e0ng tr\u0103m ngh\u00ecn th\u1eed nghi\u1ec7m \u0111\u01b0\u1ee3c ti\u1ebfn h\u00e0nh, ghi nh\u1eadn tr\u1ef1c ti\u1ebfp h\u00e0nh vi c\u1ee7a h\u00e0ng tr\u0103m tri\u1ec7u ng\u01b0\u1eddi d\u00f9ng.<\/p>\n

    Nh\u00ecn t\u1eeb g\u00f3c \u0111\u1ed9 n\u00e0y, m\u1ed7i bi\u1ebfn th\u1ec3 \u0111\u1ec1u l\u00e0 m\u1ed9t c\u00e1ch t\u00e1c \u0111\u1ed9ng v\u00e0o c\u1ea3m nh\u1eadn v\u00e0 h\u00e0nh vi. M\u00e0u n\u00fat hay font ch\u1eef l\u00e0 nh\u1eefng g\u00ec b\u1ea1n thay \u0111\u1ed5i tr\u00ean giao di\u1ec7n, c\u00f2n ng\u01b0\u1eddi d\u00f9ng c\u1ea3m th\u1ea5y th\u1ebf n\u00e0o m\u1edbi l\u00e0 \u0111i\u1ec1u c\u1ea7n t\u00ecm hi\u1ec3u ph\u00eda sau.<\/p>\n

    Ng\u01b0\u1eddi d\u00f9ng c\u00f3 nh\u1eadn ra b\u1ea1n v\u1eeba thay \u0111\u1ed5i g\u00ec?<\/h3>\n

    Ho\u1ea1t \u0111\u1ed9ng t\u1ed1i \u01b0u CRO c\u00f3 th\u1ec3 \u0111\u01b0\u1ee3c nh\u00ecn \u1edf 3 c\u1ea5p \u0111\u1ed9:<\/p>\n

      \n
    • Th\u1eed nghi\u1ec7m\/Th\u1ef1c thi: Thay \u0111\u1ed5i nh\u1eefng y\u1ebfu t\u1ed1 c\u1ee5 th\u1ec3 v\u1ec1 giao di\u1ec7n v\u00e0 tr\u1ea3i nghi\u1ec7m ng\u01b0\u1eddi d\u00f9ng (UI\/UX).<\/li>\n
    • Gi\u1ea3 thuy\u1ebft: Ki\u1ec3m ch\u1ee9ng c\u00e1c nguy\u00ean l\u00fd h\u00e0nh vi v\u00e0 nh\u1eefng r\u00e0o c\u1ea3n t\u00e2m l\u00fd.<\/li>\n
    • H\u00e0nh tr\u00ecnh kh\u00e1ch h\u00e0ng: T\u00ecm hi\u1ec3u \u0111\u1ed9ng l\u1ef1c, c\u00e1c ph\u00e2n \u0111o\u1ea1n t\u00e2m l\u00fd v\u00e0 nh\u1eefng \u0111i\u1ec3m khi\u1ebfn ng\u01b0\u1eddi d\u00f9ng b\u1ecb m\u1eafc l\u1ea1i.<\/li>\n<\/ul>\n

      N\u1ebfu ch\u1ec9 d\u1eebng \u1edf c\u1ea5p \u0111\u1ed9 \u0111\u1ea7u ti\u00ean, b\u1ea1n c\u00f3 th\u1ec3 s\u1eeda r\u1ea5t nhi\u1ec1u th\u1ee9 m\u00e0 v\u1eabn ch\u01b0a t\u00e1c \u0111\u1ed9ng \u0111\u01b0\u1ee3c \u0111\u1ebfn t\u00e2m tr\u00ed ng\u01b0\u1eddi d\u00f9ng. Theo nhi\u1ec1u chuy\u00ean gia trong ng\u00e0nh CRO, \u0111\u00e2y l\u00e0 l\u00fd do khi\u1ebfn h\u01a1n 75-90% th\u1eed nghi\u1ec7m A\/B kh\u00f4ng t\u1ea1o ra kh\u00e1c bi\u1ec7t \u0111\u00e1ng k\u1ec3.<\/p>\n

      M\u1ed9t c\u00f4ng th\u1ee9c trong gi\u1edbi CRO di\u1ec5n gi\u1ea3i \u0111i\u1ec1u n\u00e0y nh\u01b0 sau:<\/p>\n

      T\u00e1c \u0111\u1ed9ng = S\u1ef1 nh\u1eadn th\u1ee9c \u00d7 S\u1ef1 li\u00ean quan \u00d7 L\u01b0\u1ee3ng truy c\u1eadp<\/p>\n

      Theo \u0111\u00f3, n\u1ebfu ng\u01b0\u1eddi d\u00f9ng kh\u00f4ng nh\u1eadn ra thay \u0111\u1ed5i, y\u1ebfu t\u1ed1 \u201cs\u1ef1 nh\u1eadn th\u1ee9c\u201d b\u1eb1ng 0 v\u00e0 t\u00e1c \u0111\u1ed9ng c\u0169ng b\u1eb1ng 0. N\u00f3i c\u00e1ch kh\u00e1c, \u00fd t\u01b0\u1edfng c\u00f3 hay \u0111\u1ebfn \u0111\u00e2u c\u0169ng kh\u00f3 t\u1ea1o \u1ea3nh h\u01b0\u1edfng n\u1ebfu ng\u01b0\u1eddi s\u1eed d\u1ee5ng kh\u00f4ng c\u1ea3m nh\u1eadn \u0111\u01b0\u1ee3c n\u00f3.<\/p>\n

      \u0110\u00e2y ch\u00ednh l\u00e0 \u00fd ngh\u0129a c\u1ee7a nguy\u00ean t\u1eafc \u201cNo Perception = No Impact\u201d. N\u00f3 c\u0169ng l\u00fd gi\u1ea3i v\u00ec sao ph\u1ea7n l\u1edbn th\u1eed nghi\u1ec7m vi m\u00f4, nh\u01b0 \u0111\u1ed5i m\u00e0u m\u1ed9t n\u00fat nh\u1ecf hay ch\u1ec9nh nh\u1eb9 l\u1ec1 trang, kh\u00f4ng mang l\u1ea1i k\u1ebft qu\u1ea3 c\u00f3 \u00fd ngh\u0129a.<\/p>\n

      B\u1ed9 n\u00e3o th\u01b0\u1eddng ch\u1ecdn c\u00e1ch \u00edt t\u1ed1n s\u1ee9c<\/h3>\n

      Theo l\u00fd thuy\u1ebft ti\u1ebfn tr\u00ecnh k\u00e9p (Dual-Process Theory), n\u00e3o b\u1ed9 c\u00f3 2 h\u1ec7 th\u1ed1ng x\u1eed l\u00fd th\u00f4ng tin l\u00e0 System 1 v\u00e0 System 2.<\/p>\n

      System 1 ho\u1ea1t \u0111\u1ed9ng nhanh, t\u1ef1 \u0111\u1ed9ng, thi\u00ean v\u1ec1 c\u1ea3m x\u00fac v\u00e0 di\u1ec5n ra v\u00f4 th\u1ee9c. H\u1ec7 th\u1ed1ng n\u00e0y l\u00e0m vi\u1ec7c li\u00ean t\u1ee5c nh\u01b0ng ti\u00eau t\u1ed1n \u00edt n\u0103ng l\u01b0\u1ee3ng, ch\u1ecbu tr\u00e1ch nhi\u1ec7m cho h\u01a1n 90-95% quy\u1ebft \u0111\u1ecbnh h\u00e0ng ng\u00e0y c\u1ee7a ch\u00fang ta.<\/p>\n

      Trong khi \u0111\u00f3, System 2 x\u1eed l\u00fd ch\u1eadm h\u01a1n, d\u1ef1a v\u00e0o logic v\u00e0 s\u1ef1 ch\u00fa \u00fd c\u00f3 \u00fd th\u1ee9c. N\u00f3 ch\u1ec9 \u0111\u01b0\u1ee3c k\u00edch ho\u1ea1t khi g\u1eb7p v\u1ea5n \u0111\u1ec1 ph\u1ee9c t\u1ea1p, \u0111\u1ed3ng th\u1eddi d\u1ec5 r\u01a1i v\u00e0o tr\u1ea1ng th\u00e1i \u201cc\u1ea1n pin\u201d nh\u1eadn th\u1ee9c (cognitive depletion).<\/p>\n

      H\u1ea7u h\u1ebft ng\u01b0\u1eddi d\u00f9ng l\u01b0\u1edbt web v\u00e0 ra quy\u1ebft \u0111\u1ecbnh b\u1eb1ng System 1. V\u00ec v\u1eady, n\u1ebfu bi\u1ebfn th\u1ec3 ch\u1ec9 b\u1ed5 sung th\u00f4ng tin logic \u0111\u1ec3 h\u1ecd ph\u1ea3i suy ngh\u0129 b\u1eb1ng System 2, h\u1ecd s\u1ebd th\u1ea5y qu\u00e1 t\u1ea3i v\u00e0 r\u1eddi \u0111i. Nh\u1eefng bi\u1ebfn th\u1ec3 t\u1ea1o ra m\u1ee9c c\u1ea3i thi\u1ec7n l\u1edbn th\u01b0\u1eddng t\u00e1c \u0111\u1ed9ng tr\u1ef1c ti\u1ebfp v\u00e0o System 1, th\u00f4ng qua thi\u1ebft k\u1ebf tr\u1ef1c quan, gi\u00e1 tr\u1ecb c\u1ea3m x\u00fac (Emotional Value Proposition) v\u00e0 c\u00e1c thi\u00ean ki\u1ebfn h\u00e0nh vi.<\/p>\n<\/section>\n

      \n

      T\u00e2m l\u00fd ng\u01b0\u1eddi d\u00f9ng t\u1ea1o ra kh\u00e1c bi\u1ec7t th\u1ebf n\u00e0o?<\/h2>\n

      H\u00ecnh \u1ea3nh g\u1ee3i c\u1ea3m x\u00fac c\u00f3 th\u1ec3 khi\u1ebfn kh\u00e1ch mu\u1ed1n mua h\u01a1n<\/h3>\n

      M\u1ed9t v\u00ed d\u1ee5 \u0111i\u1ec3n h\u00ecnh l\u00e0 th\u1eed nghi\u1ec7m v\u1edbi m\u1eabu gi\u00e0y Boxfresh Sparko t\u1ea1i m\u1ed9t c\u1eeda h\u00e0ng th\u1eddi trang tr\u1ef1c tuy\u1ebfn l\u1edbn \u1edf \u0110\u1ee9c. C\u1eeda h\u00e0ng th\u1eed t\u1ed1i \u01b0u khung m\u1eabu trang, \u0111i\u1ec1u ch\u1ec9nh kho\u1ea3ng tr\u1eafng nh\u01b0ng gi\u1eef nguy\u00ean h\u00ecnh \u1ea3nh s\u1ea3n ph\u1ea9m. K\u1ebft qu\u1ea3, t\u1ef7 l\u1ec7 th\u00eam v\u00e0o gi\u1ecf h\u00e0ng t\u0103ng 8%.<\/p>\n

      \u1ede m\u1ed9t h\u01b0\u1edbng ti\u1ebfp c\u1eadn kh\u00e1c, h\u00ecnh \u1ea3nh s\u1ea3n ph\u1ea9m \u0111\u01b0\u1ee3c t\u00e1i \u0111\u1ecbnh v\u1ecb d\u1ef1a tr\u00ean nh\u1eefng gi\u00e1 tr\u1ecb c\u1ea3m x\u00fac c\u1ee7a ng\u01b0\u1eddi d\u00f9ng, nh\u01b0 s\u1ef1 \u0111\u1ed9c \u0111\u00e1o hay \u0111\u1ecba v\u1ecb. L\u1ea7n n\u00e0y, t\u1ef7 l\u1ec7 th\u00eam v\u00e0o gi\u1ecf h\u00e0ng t\u0103ng t\u1edbi 80%, doanh s\u1ed1 th\u1ef1c t\u1ebf t\u0103ng 40%, trong khi t\u1ef7 l\u1ec7 tr\u1ea3 h\u00e0ng kh\u00f4ng t\u0103ng.<\/p>\n

      C\u00f9ng m\u1ed9t s\u1ea3n ph\u1ea9m, nh\u01b0ng c\u00e1ch t\u00e1c \u0111\u1ed9ng v\u00e0o c\u1ea3m nh\u1eadn c\u1ee7a ng\u01b0\u1eddi xem \u0111\u00e3 mang l\u1ea1i nh\u1eefng k\u1ebft qu\u1ea3 kh\u00e1c nhau.<\/p>\n

      C\u1ea3m gi\u00e1c \u0111\u01b0\u1ee3c l\u1ef1a ch\u1ecdn khi\u1ebfn vi\u1ec7c \u0111i\u1ec1n form d\u1ec5 ch\u1ecbu h\u01a1n<\/h3>\n

      Khi mu\u1ed1n t\u0103ng t\u1ef7 l\u1ec7 ho\u00e0n th\u00e0nh form \u0111\u1eb7t ph\u00f2ng, chu\u1ed7i kh\u00e1ch s\u1ea1n Van Der Valk t\u1ea1i H\u00e0 Lan th\u1eed thay \u0111\u1ed5i c\u00e1ch \u0111\u00e1nh d\u1ea5u c\u00e1c tr\u01b0\u1eddng th\u00f4ng tin. Thay v\u00ec ch\u1ec9 ra nh\u1eefng tr\u01b0\u1eddng b\u1eaft bu\u1ed9c ph\u1ea3i \u0111i\u1ec1n nh\u01b0 th\u00f4ng th\u01b0\u1eddng, h\u1ecd chuy\u1ec3n sang \u0111\u00e1nh d\u1ea5u c\u00e1c tr\u01b0\u1eddng t\u1ef1 ch\u1ecdn.<\/p>\n

      C\u00e1ch th\u1ec3 hi\u1ec7n n\u00e0y khi\u1ebfn ng\u01b0\u1eddi d\u00f9ng c\u00f3 c\u1ea3m gi\u00e1c \u0111\u01b0\u1ee3c t\u1ef1 do l\u1ef1a ch\u1ecdn h\u01a1n, thay v\u00ec b\u1ecb y\u00eau c\u1ea7u ph\u1ea3i l\u00e0m m\u1ed9t vi\u1ec7c. K\u1ebft qu\u1ea3, t\u1ef7 l\u1ec7 ho\u00e0n th\u00e0nh form t\u0103ng 24.7%. \u0110i\u1ec1u \u0111\u01b0\u1ee3c thay \u0111\u1ed5i l\u00e0 c\u00e1ch g\u1ecdi t\u00ean, nh\u01b0ng \u0111i\u1ec1u ch\u1ecbu t\u00e1c \u0111\u1ed9ng l\u1ea1i l\u00e0 c\u1ea3m gi\u00e1c c\u1ee7a ng\u01b0\u1eddi \u0111ang \u0111i\u1ec1n th\u00f4ng tin.<\/p>\n

      M\u1ed9t h\u1ea1n ch\u00f3t c\u00f3 th\u1eadt t\u1ea1o c\u1ea3m gi\u00e1c c\u1ea5p b\u00e1ch<\/h3>\n

      M\u1ed9t doanh nghi\u1ec7p B2B b\u00e1n m\u00e1y photo v\u00e0 thi\u1ebft b\u1ecb v\u0103n ph\u00f2ng t\u1ed5 ch\u1ee9c ch\u01b0\u01a1ng tr\u00ecnh \u01b0u \u0111\u00e3i k\u00e9o d\u00e0i 2 ng\u00e0y. B\u00ean c\u1ea1nh email th\u00f4ng b\u00e1o gi\u1ea3m gi\u00e1 th\u00f4ng th\u01b0\u1eddng, h\u1ecd th\u1eed th\u00eam m\u1ed9t phi\u00ean b\u1ea3n nh\u1ea5n m\u1ea1nh \u0111\u1ed3ng h\u1ed3 \u0111\u1ebfm ng\u01b0\u1ee3c 48 gi\u1edd.<\/p>\n

      Doanh thu t\u1eeb phi\u00ean b\u1ea3n c\u00f3 \u0111\u1ed3ng h\u1ed3 t\u0103ng t\u1edbi 136% so v\u1edbi b\u1ea3n g\u1ed1c. Trong tr\u01b0\u1eddng h\u1ee3p n\u00e0y, c\u1ea3m gi\u00e1c c\u1ea5p b\u00e1ch \u0111\u01b0\u1ee3c g\u1eafn v\u1edbi th\u1eddi h\u1ea1n th\u1ef1c t\u1ebf c\u1ee7a ch\u01b0\u01a1ng tr\u00ecnh.<\/p>\n

      \u1ea2nh l\u1edbn ch\u01b0a ch\u1eafc khi\u1ebfn s\u1ea3n ph\u1ea9m c\u00f3 gi\u00e1 h\u01a1n<\/h3>\n

      Theo nghi\u00ean c\u1ee9u c\u1ee7a CXL tr\u00ean 2.982 ng\u01b0\u1eddi tham gia, ph\u1ea3n \u1ee9ng v\u1edbi k\u00edch th\u01b0\u1edbc \u1ea3nh c\u00f2n ph\u1ee5 thu\u1ed9c v\u00e0o lo\u1ea1i s\u1ea3n ph\u1ea9m \u0111\u01b0\u1ee3c gi\u1edbi thi\u1ec7u.<\/p>\n

      V\u1edbi s\u1ea3n ph\u1ea9m mang t\u00ednh tr\u1ea3i nghi\u1ec7m nh\u01b0 \u00e1o s\u01a1 mi nam, \u1ea3nh nh\u1ecf k\u1ebft h\u1ee3p nhi\u1ec1u kho\u1ea3ng tr\u1eafng khi\u1ebfn ng\u01b0\u1eddi xem c\u1ea3m nh\u1eadn gi\u00e1 tr\u1ecb cao h\u01a1n so v\u1edbi \u1ea3nh l\u1edbn chi\u1ebfm h\u1ebft m\u00e0n h\u00ecnh. Nh\u01b0ng v\u1edbi s\u1ea3n ph\u1ea9m thi\u00ean v\u1ec1 th\u00f4ng s\u1ed1 k\u1ef9 thu\u1eadt nh\u01b0 \u1ed5 c\u1ee9ng m\u00e1y t\u00ednh, \u1ea3nh l\u1edbn l\u1ea1i l\u00e0m t\u0103ng \u0111\u1ecbnh gi\u00e1 c\u1ea3m nh\u1eadn. Ng\u01b0\u1eddi tham gia th\u1eadm ch\u00ed s\u1eb5n s\u00e0ng tr\u1ea3 th\u00eam 15 USD.<\/p>\n

      Nh\u01b0 v\u1eady, c\u00f9ng l\u00e0 \u0111i\u1ec1u ch\u1ec9nh k\u00edch th\u01b0\u1edbc \u1ea3nh, ph\u1ea3n \u1ee9ng c\u1ee7a ng\u01b0\u1eddi d\u00f9ng c\u00f3 th\u1ec3 \u0111i theo nh\u1eefng h\u01b0\u1edbng kh\u00e1c nhau.<\/p>\n

      \"\u1ea2nh
      K\u00edch th\u01b0\u1edbc \u1ea3nh v\u00e0 kho\u1ea3ng tr\u1eafng xung quanh c\u00f3 th\u1ec3 khi\u1ebfn kh\u00e1ch h\u00e0ng \u0111\u00e1nh gi\u00e1 kh\u00e1c \u0111i v\u1ec1 gi\u00e1 tr\u1ecb c\u1ee7a c\u00f9ng m\u1ed9t s\u1ea3n ph\u1ea9m.<\/figcaption><\/figure>\n<\/section>\n
      \n

      B cao h\u01a1n A, \u0111\u00e3 ch\u1eafc l\u00e0 \u201cth\u1eafng\u201d?<\/h2>\n

      M\u1ed9t phi\u00ean b\u1ea3n c\u00f3 t\u1ef7 l\u1ec7 chuy\u1ec3n \u0111\u1ed5i cao h\u01a1n ch\u01b0a \u0111\u1ee7 \u0111\u1ec3 gi\u1ea3i th\u00edch v\u00ec sao n\u00f3 th\u1eafng. \u0110\u1ec3 bi\u1ebft ch\u00eanh l\u1ec7ch c\u00f3 th\u1ef1c s\u1ef1 \u0111\u1ebfn t\u1eeb thay \u0111\u1ed5i \u0111\u01b0\u1ee3c th\u1eed nghi\u1ec7m, ch\u00fang ta c\u1ea7n \u0111\u1ebfn th\u1ed1ng k\u00ea. B\u1ea1n kh\u00f4ng nh\u1ea5t thi\u1ebft ph\u1ea3i tr\u1edf th\u00e0nh nh\u00e0 th\u1ed1ng k\u00ea h\u1ecdc, nh\u01b0ng c\u1ea7n hi\u1ec3u m\u1ed9t s\u1ed1 kh\u00e1i ni\u1ec7m \u0111\u1ec3 tr\u00e1nh \u0111\u1ecdc nh\u1ea7m k\u1ebft qu\u1ea3.<\/p>\n

      Ph\u00e2n bi\u1ec7t kh\u00e1c bi\u1ec7t th\u1ef1c s\u1ef1 v\u1edbi sai s\u1ed1 ng\u1eabu nhi\u00ean<\/h3>\n

      M\u1ed7i th\u1eed nghi\u1ec7m b\u1eaft \u0111\u1ea7u v\u1edbi 2 gi\u1ea3 thuy\u1ebft \u0111\u1ed1i l\u1eadp. Gi\u1ea3 thuy\u1ebft kh\u00f4ng (H0) cho r\u1eb1ng A v\u00e0 B kh\u00f4ng c\u00f3 kh\u00e1c bi\u1ec7t th\u1ef1c s\u1ef1; ch\u00eanh l\u1ec7ch quan s\u00e1t \u0111\u01b0\u1ee3c ch\u1ec9 \u0111\u1ebfn t\u1eeb sai s\u1ed1 ng\u1eabu nhi\u00ean. Trong khi \u0111\u00f3, gi\u1ea3 thuy\u1ebft \u0111\u1ed1i (H1) cho r\u1eb1ng c\u00f3 kh\u00e1c bi\u1ec7t th\u1ef1c s\u1ef1 do thay \u0111\u1ed5i \u0111\u01b0\u1ee3c \u0111\u01b0a v\u00e0o th\u1eed nghi\u1ec7m g\u00e2y ra.<\/p>\n

      \u0110\u1ec3 \u0111\u00e1nh gi\u00e1 k\u1ebft qu\u1ea3, ch\u00fang ta s\u1eed d\u1ee5ng p-value. \u0110\u00e2y l\u00e0 x\u00e1c su\u1ea5t nh\u1eadn \u0111\u01b0\u1ee3c m\u1ee9c ch\u00eanh l\u1ec7ch nh\u01b0 \u0111\u00e3 quan s\u00e1t, ho\u1eb7c l\u1edbn h\u01a1n, n\u1ebfu H0 \u0111\u00fang.<\/p>\n

      Ng\u01b0\u1ee1ng \u00fd ngh\u0129a th\u01b0\u1eddng \u0111\u01b0\u1ee3c \u0111\u1eb7t \u1edf 0.05, t\u01b0\u01a1ng \u1ee9ng v\u1edbi m\u1ee9c tin c\u1eady 95%. Khi m\u1ee9c c\u1ea3i thi\u1ec7n c\u00f3 p-value nh\u1ecf h\u01a1n 0.05, b\u1ea1n c\u00f3 th\u1ec3 b\u00e1c b\u1ecf H0 v\u00e0 k\u1ebft lu\u1eadn s\u1ef1 c\u1ea3i thi\u1ec7n c\u00f3 \u00fd ngh\u0129a th\u1ed1ng k\u00ea.<\/p>\n

      \u00cdt ng\u01b0\u1eddi tham gia th\u00ec kh\u00f3 nh\u00ecn th\u1ea5y kh\u00e1c bi\u1ec7t<\/h3>\n

      \u0110\u1ed9 m\u1ea1nh th\u1ed1ng k\u00ea (statistical power) l\u00e0 kh\u1ea3 n\u0103ng m\u1ed9t th\u1eed nghi\u1ec7m ph\u00e1t hi\u1ec7n \u0111\u01b0\u1ee3c kh\u00e1c bi\u1ec7t th\u1ef1c s\u1ef1 khi kh\u00e1c bi\u1ec7t \u1ea5y t\u1ed3n t\u1ea1i. M\u1ee9c th\u01b0\u1eddng \u0111\u01b0\u1ee3c s\u1eed d\u1ee5ng trong ng\u00e0nh l\u00e0 t\u1eeb 80% tr\u1edf l\u00ean.<\/p>\n

      C\u00f3 2 ki\u1ec3u sai l\u1ea7m c\u1ea7n ch\u00fa \u00fd khi \u0111\u1ecdc k\u1ebft qu\u1ea3:<\/p>\n

        \n
      • Sai l\u1ea7m lo\u1ea1i I: B\u1ea1n k\u1ebft lu\u1eadn B t\u1ed1t h\u01a1n A, trong khi th\u1ef1c t\u1ebf hai phi\u00ean b\u1ea3n kh\u00f4ng kh\u00e1c nhau.<\/li>\n
      • Sai l\u1ea7m lo\u1ea1i II: B th\u1ef1c s\u1ef1 t\u1ed1t h\u01a1n, nh\u01b0ng th\u1eed nghi\u1ec7m kh\u00f4ng ph\u00e1t hi\u1ec7n \u0111\u01b0\u1ee3c v\u00ec k\u00edch th\u01b0\u1edbc m\u1eabu ch\u01b0a \u0111\u1ee7.<\/li>\n<\/ul>\n

        \u0110i\u1ec1u n\u00e0y li\u00ean quan \u0111\u1ebfn MDE (Minimum Detectable Effect) – m\u1ee9c ch\u00eanh l\u1ec7ch nh\u1ecf nh\u1ea5t v\u1ec1 t\u1ef7 l\u1ec7 chuy\u1ec3n \u0111\u1ed5i m\u00e0 b\u1ea1n mu\u1ed1n ph\u00e1t hi\u1ec7n. MDE c\u00e0ng nh\u1ecf, l\u01b0\u1ee3ng m\u1eabu c\u1ea7n c\u00f3 c\u00e0ng l\u1edbn.<\/p>\n

        V\u00ec v\u1eady, m\u1ed9t trang c\u00f3 \u00edt l\u01b0\u1ee3t truy c\u1eadp d\u1ec5 r\u01a1i v\u00e0o t\u00ecnh tr\u1ea1ng thi\u1ebfu m\u1eabu (underpowered tests). Khi \u0111\u00f3, m\u1ee9c ch\u00eanh l\u1ec7ch c\u1ea7n thi\u1ebft \u0111\u1ec3 ph\u00e1t hi\u1ec7n t\u00edn hi\u1ec7u tr\u1edf n\u00ean qu\u00e1 l\u1edbn. B\u1ea1n c\u00f3 th\u1ec3 ch\u1ea1y th\u1eed nghi\u1ec7m nhi\u1ec1u th\u00e1ng m\u00e0 v\u1eabn ch\u01b0a \u0111\u1ea1t \u00fd ngh\u0129a th\u1ed1ng k\u00ea.<\/p>\n

        C\u00e1ch \u0111\u1ecdc k\u1ebft qu\u1ea3 c\u0169ng c\u00f3 s\u1ef1 kh\u00e1c nhau<\/h3>\n

        V\u1edbi c\u00e1ch ti\u1ebfp c\u1eadn Frequentist, b\u1ea1n c\u1ea7n thu th\u1eadp \u0111\u1ee7 l\u01b0\u1ee3ng m\u1eabu c\u1ed1 \u0111\u1ecbnh \u0111\u00e3 t\u00ednh tr\u01b0\u1edbc, sau \u0111\u00f3 m\u1edbi t\u00ednh p-value v\u00e0 \u0111\u01b0a ra k\u1ebft lu\u1eadn. C\u00e1ch n\u00e0y kh\u00f4ng cho ph\u00e9p k\u1ebft lu\u1eadn s\u1edbm.<\/p>\n

        Trong khi \u0111\u00f3, Bayesian li\u00ean t\u1ee5c c\u1eadp nh\u1eadt x\u00e1c su\u1ea5t th\u1eafng ho\u1eb7c thua d\u1ef1a tr\u00ean d\u1eef li\u1ec7u thu th\u1eadp \u0111\u01b0\u1ee3c. C\u00e1ch ti\u1ebfp c\u1eadn n\u00e0y ph\u00f9 h\u1ee3p h\u01a1n v\u1edbi nh\u1eefng quy\u1ebft \u0111\u1ecbnh kinh doanh c\u1ea7n s\u1ef1 linh ho\u1ea1t.<\/p>\n<\/section>\n

        \n

        B\u1eaft \u0111\u1ea7u m\u1ed9t th\u1eed nghi\u1ec7m A\/B nh\u01b0 th\u1ebf n\u00e0o?<\/h2>\n

        M\u1ed9t ch\u01b0\u01a1ng tr\u00ecnh A\/B testing ho\u00e0n ch\u1ec9nh c\u1ea7n 5 th\u00e0nh ph\u1ea7n: b\u1ea3n g\u1ed1c, bi\u1ebfn th\u1ec3, h\u1ec7 th\u1ed1ng ch\u1ec9 s\u1ed1, c\u00e1ch ph\u00e2n lu\u1ed3ng l\u01b0u l\u01b0\u1ee3ng v\u00e0 gi\u1ea3 thuy\u1ebft c\u1ea7n ki\u1ec3m ch\u1ee9ng. \u0110\u1ec3 chu\u1ea9n b\u1ecb v\u00e0 s\u1eed d\u1ee5ng nh\u1eefng th\u00e0nh ph\u1ea7n n\u00e0y, b\u1ea1n c\u00f3 th\u1ec3 \u0111i theo 6 b\u01b0\u1edbc sau.<\/p>\n

        T\u00ecm hi\u1ec3u ng\u01b0\u1eddi d\u00f9ng \u0111ang g\u1eb7p kh\u00f3 kh\u0103n \u1edf \u0111\u00e2u<\/h3>\n

        \u00dd t\u01b0\u1edfng th\u1eed nghi\u1ec7m c\u1ea7n b\u1eaft \u0111\u1ea7u t\u1eeb d\u1eef li\u1ec7u v\u1ec1 ng\u01b0\u1eddi d\u00f9ng. V\u1edbi d\u1eef li\u1ec7u \u0111\u1ecbnh l\u01b0\u1ee3ng, b\u1ea1n c\u00f3 th\u1ec3 d\u00f9ng Google Analytics\/GA4 \u0111\u1ec3 t\u00ecm nh\u1eefng trang c\u00f3 t\u1ef7 l\u1ec7 tho\u00e1t ho\u1eb7c b\u1ecf d\u1edf cao. Heatmap, click map v\u00e0 scroll map gi\u00fap nh\u1eadn bi\u1ebft h\u1ecd t\u01b0\u01a1ng t\u00e1c v\u1edbi ph\u1ea7n n\u00e0o, ho\u1eb7c b\u1ecf qua ph\u1ea7n n\u00e0o tr\u00ean trang.<\/p>\n

        B\u00ean c\u1ea1nh \u0111\u00f3, d\u1eef li\u1ec7u \u0111\u1ecbnh t\u00ednh gi\u00fap t\u00ecm hi\u1ec3u nh\u1eefng r\u00e0o c\u1ea3n ph\u00eda sau h\u00e0nh vi. B\u1ea1n c\u00f3 th\u1ec3 thu th\u1eadp th\u00f4ng tin t\u1eeb kh\u1ea3o s\u00e1t ng\u01b0\u1eddi d\u00f9ng, ph\u1ecfng v\u1ea5n exit-intent khi h\u1ecd chu\u1ea9n b\u1ecb r\u1eddi trang, th\u1eed nghi\u1ec7m 5 gi\u00e2y v\u00e0 ki\u1ec3m tra t\u00ednh kh\u1ea3 d\u1ee5ng. \u0110\u00e1nh gi\u00e1 c\u1ee7a chuy\u00ean gia d\u1ef1a tr\u00ean t\u00e2m l\u00fd h\u1ecdc nh\u1eadn th\u1ee9c c\u0169ng gi\u00fap l\u00e0m r\u00f5 nh\u1eefng r\u00e0o c\u1ea3n n\u00e0y.<\/p>\n

        \"B\u1ea3n
        Heatmap gi\u00fap b\u1ea1n th\u1ea5y ng\u01b0\u1eddi d\u00f9ng th\u01b0\u1eddng t\u01b0\u01a1ng t\u00e1c \u1edf \u0111\u00e2u v\u00e0 nh\u1eefng ph\u1ea7n n\u00e0o tr\u00ean trang d\u1ec5 b\u1ecb b\u1ecf qua.<\/figcaption><\/figure>\n

        T\u1eeb \u0111i\u1ec1u quan s\u00e1t \u0111\u01b0\u1ee3c, \u0111\u1eb7t ra gi\u1ea3 thuy\u1ebft<\/h3>\n

        Sau khi c\u00f3 d\u1eef li\u1ec7u, b\u1ea1n c\u1ea7n x\u00e1c \u0111\u1ecbnh thay \u0111\u1ed5i n\u00e0o c\u00f3 th\u1ec3 t\u00e1c \u0111\u1ed9ng \u0111\u1ebfn h\u00e0nh vi, v\u00e0 v\u00ec sao. \u0110\u00e2y l\u00e0 c\u00e1ch tr\u00e1nh \u201cgarbage testing\u201d – \u0111\u01b0a nh\u1eefng thay \u0111\u1ed5i ng\u1eabu nhi\u00ean v\u00e0o th\u1eed nghi\u1ec7m m\u00e0 kh\u00f4ng c\u00f3 c\u01a1 s\u1edf.<\/p>\n

        M\u1ed9t gi\u1ea3 thuy\u1ebft r\u00f5 r\u00e0ng c\u00f3 th\u1ec3 \u0111\u01b0\u1ee3c tr\u00ecnh b\u00e0y theo c\u00f4ng th\u1ee9c: \u201cN\u1ebfu \u00e1p d\u1ee5ng [thay \u0111\u1ed5i X], th\u00ec [ch\u1ec9 s\u1ed1 Y] s\u1ebd thay \u0111\u1ed5i [m\u1ee9c \u0111\u1ed9 Z], b\u1edfi v\u00ec [l\u00fd do t\u00e2m l\u00fd\/h\u00e0nh vi W]\u201d.<\/p>\n

        Ch\u1eb3ng h\u1ea1n, b\u1ea1n c\u00f3 th\u1ec3 \u0111\u1eb7t gi\u1ea3 thuy\u1ebft: \u201cNg\u01b0\u1eddi d\u00f9ng b\u1ecf d\u1edf form \u0111\u0103ng k\u00fd \u1edf b\u01b0\u1edbc 2 v\u00ec lo l\u1eafng v\u1ec1 b\u1ea3o m\u1eadt d\u1eef li\u1ec7u c\u00e1 nh\u00e2n. N\u1ebfu b\u1ed5 sung bi\u1ec3u t\u01b0\u1ee3ng ch\u1ee9ng nh\u1eadn uy t\u00edn ngay c\u1ea1nh n\u00fat Submit, t\u1ef7 l\u1ec7 ho\u00e0n th\u00e0nh form s\u1ebd t\u0103ng 15%.\u201d<\/p>\n

        Trong v\u00ed d\u1ee5 n\u00e0y, bi\u1ec3u t\u01b0\u1ee3ng ch\u1ee9ng nh\u1eadn l\u00e0 thay \u0111\u1ed5i \u0111\u01b0\u1ee3c \u0111\u01b0a v\u00e0o th\u1eed nghi\u1ec7m. C\u00f2n \u0111i\u1ec1u c\u1ea7n ki\u1ec3m ch\u1ee9ng ph\u00eda sau l\u00e0 gi\u1ea3 thuy\u1ebft v\u1ec1 n\u1ed7i lo b\u1ea3o m\u1eadt khi\u1ebfn ng\u01b0\u1eddi d\u00f9ng b\u1ecf d\u1edf form.<\/p>\n

        Ch\u1ecdn th\u1eed nghi\u1ec7m c\u1ea7n l\u00e0m tr\u01b0\u1edbc<\/h3>\n

        \u0110\u1ec3 s\u1eafp x\u1ebfp th\u1ee9 t\u1ef1 \u01b0u ti\u00ean, b\u1ea1n c\u00f3 th\u1ec3 tham kh\u1ea3o khung PIE, g\u1ed3m Potential (ti\u1ec1m n\u0103ng c\u1ea3i thi\u1ec7n), Importance (t\u1ea7m quan tr\u1ecdng c\u1ee7a trang) v\u00e0 Ease (\u0111\u1ed9 d\u1ec5 tri\u1ec3n khai). M\u1ed9t l\u1ef1a ch\u1ecdn kh\u00e1c l\u00e0 ma tr\u1eadn T\u00e1c \u0111\u1ed9ng – Chi ph\u00ed.<\/p>\n

        Nguy\u00ean t\u1eafc chung l\u00e0 \u01b0u ti\u00ean nh\u1eefng th\u1eed nghi\u1ec7m c\u00f3 t\u00e1c \u0111\u1ed9ng cao, k\u1ec3 c\u1ea3 khi c\u1ea7n nhi\u1ec1u c\u00f4ng s\u1ee9c th\u1ef1c hi\u1ec7n (bold testing). Nh\u1eefng thay \u0111\u1ed5i t\u1ed1n chi ph\u00ed nh\u01b0ng t\u00e1c \u0111\u1ed9ng th\u1ea5p c\u0169ng thu\u1ed9c nh\u00f3m \u201cgarbage testing\u201d c\u1ea7n tr\u00e1nh.<\/p>\n

        Chu\u1ea9n b\u1ecb bi\u1ebfn th\u1ec3 v\u00e0 c\u00e1ch \u0111o l\u01b0\u1eddng<\/h3>\n

        Sau khi ch\u1ecdn \u0111\u01b0\u1ee3c gi\u1ea3 thuy\u1ebft, b\u1ea1n b\u1eaft \u0111\u1ea7u thi\u1ebft k\u1ebf v\u00e0 l\u1eadp tr\u00ecnh bi\u1ebfn th\u1ec3. S\u1ef1 kh\u00e1c bi\u1ec7t c\u1ea7n \u0111\u1ee7 r\u00f5 \u0111\u1ec3 ng\u01b0\u1eddi d\u00f9ng c\u00f3 th\u1ec3 nh\u1eadn th\u1ea5y, n\u1ebfu kh\u00f4ng b\u1ea1n l\u1ea1i g\u1eb7p v\u1ea5n \u0111\u1ec1 \u201cNo Perception = No Impact\u201d \u1edf tr\u00ean.<\/p>\n

        Tr\u01b0\u1edbc khi ch\u1ea1y, c\u1ea7n ki\u1ec3m th\u1eed QA tr\u00ean m\u1ecdi tr\u00ecnh duy\u1ec7t, thi\u1ebft b\u1ecb v\u00e0 \u0111\u1ed9 ph\u00e2n gi\u1ea3i m\u00e0n h\u00ecnh. M\u1ed9t l\u1ed7i v\u1ee1 khung c\u00f3 th\u1ec3 l\u00e0m m\u00e9o d\u1eef li\u1ec7u th\u1eed nghi\u1ec7m.<\/p>\n

        H\u1ec7 th\u1ed1ng \u0111o l\u01b0\u1eddng c\u1ea7n c\u00f3 3 nh\u00f3m ch\u1ec9 s\u1ed1:<\/p>\n

          \n
        • Ch\u1ec9 s\u1ed1 ch\u00ednh (Primary Metric\/OMTM): D\u00f9ng \u0111\u1ec3 quy\u1ebft \u0111\u1ecbnh phi\u00ean b\u1ea3n th\u1eafng, ch\u1eb3ng h\u1ea1n t\u1ef7 l\u1ec7 ho\u00e0n th\u00e0nh \u0111\u01a1n h\u00e0ng.<\/li>\n
        • Ch\u1ec9 s\u1ed1 ph\u1ee5: Theo d\u00f5i nh\u1eefng b\u01b0\u1edbc trung gian, nh\u01b0 t\u1ef7 l\u1ec7 nh\u1ea5p Add-to-Cart ho\u1eb7c t\u1ef7 l\u1ec7 cu\u1ed9n trang.<\/li>\n
        • Ch\u1ec9 s\u1ed1 b\u1ea3o v\u1ec7 (Guardrail Metrics): Gi\u00fap \u0111\u1ea3m b\u1ea3o th\u1eed nghi\u1ec7m kh\u00f4ng g\u00e2y h\u1ea1i cho tr\u1ea3i nghi\u1ec7m t\u1ed5ng th\u1ec3. C\u00e1c ch\u1ec9 s\u1ed1 c\u00f3 th\u1ec3 g\u1ed3m t\u1ef7 l\u1ec7 ho\u00e0n tr\u1ea3 h\u00e0ng, t\u1ef7 l\u1ec7 h\u1ee7y \u0111\u0103ng k\u00fd, t\u1ed1c \u0111\u1ed9 t\u1ea3i trang v\u00e0 doanh thu tr\u00ean m\u1ed7i ng\u01b0\u1eddi d\u00f9ng (ARPU).<\/li>\n<\/ul>\n

          B\u1ea1n c\u0169ng c\u1ea7n chu\u1ea9n b\u1ecb c\u00e1ch ph\u00e2n lu\u1ed3ng ng\u1eabu nhi\u00ean gi\u1eefa A v\u00e0 B. Cookie ho\u1eb7c User ID gi\u00fap m\u1ed9t ng\u01b0\u1eddi khi quay l\u1ea1i v\u1eabn th\u1ea5y \u0111\u00fang phi\u00ean b\u1ea3n \u0111\u00e3 \u0111\u01b0\u1ee3c ph\u00e2n t\u1eeb \u0111\u1ea7u.<\/p>\n

          Cho th\u1eed nghi\u1ec7m \u0111\u1ee7 th\u1eddi gian<\/h3>\n

          Th\u1eddi gian ch\u1ea1y t\u1ed1i thi\u1ec3u c\u1ea7n \u1edf m\u1ee9c 14-21 ng\u00e0y, bao ph\u1ee7 c\u1ea3 ng\u00e0y th\u01b0\u1eddng l\u1eabn cu\u1ed1i tu\u1ea7n. \u0110i\u1ec1u n\u00e0y gi\u00fap h\u1ea1n ch\u1ebf nhi\u1ec5u do th\u00f3i quen mua s\u1eafm thay \u0111\u1ed5i theo ng\u00e0y.<\/p>\n

          N\u1ebfu p-value xu\u1ed1ng d\u01b0\u1edbi 0.05 trong v\u00e0i ng\u00e0y \u0111\u1ea7u, b\u1ea1n c\u0169ng \u0111\u1eebng v\u1ed9i d\u1eebng th\u1eed nghi\u1ec7m. Nh\u1eefng t\u00edn hi\u1ec7u t\u00edch c\u1ef1c ban \u0111\u1ea7u ch\u01b0a ph\u1ea3i l\u00fd do \u0111\u1ec3 b\u1ecf qua l\u01b0\u1ee3ng m\u1eabu v\u00e0 th\u1eddi gian c\u1ea7n thi\u1ebft.<\/p>\n

          \u0110\u1ecdc k\u1ebft qu\u1ea3 v\u00e0 l\u01b0u l\u1ea1i \u0111i\u1ec1u \u0111\u00e3 h\u1ecdc<\/h3>\n

          M\u1ed9t th\u1eed nghi\u1ec7m c\u00f3 th\u1ec3 cho k\u1ebft qu\u1ea3 h\u00f2a khi nh\u00ecn t\u1ed5ng th\u1ec3. Tuy nhi\u00ean, khi ph\u00e2n t\u00edch s\u00e2u h\u01a1n theo t\u1eebng ph\u00e2n \u0111o\u1ea1n, b\u1ea1n \u0111\u00f4i khi s\u1ebd th\u1ea5y bi\u1ebfn th\u1ec3 th\u1eafng r\u00f5 \u1edf nh\u00f3m d\u00f9ng \u0111i\u1ec7n tho\u1ea1i ho\u1eb7c macOS.<\/p>\n

          D\u00f9 k\u1ebft qu\u1ea3 th\u1eafng hay thua, c\u00e1c b\u00e0i h\u1ecdc c\u1ea7n \u0111\u01b0\u1ee3c l\u01b0u v\u00e0o kho tri th\u1ee9c chung \u0111\u1ec3 \u0111\u1ed9i s\u1ea3n ph\u1ea9m v\u00e0 marketing c\u00f9ng s\u1eed d\u1ee5ng. M\u1ed9t bi\u1ebfn th\u1ec3 ch\u01b0a t\u1ea1o \u0111\u01b0\u1ee3c m\u1ee9c c\u1ea3i thi\u1ec7n mong mu\u1ed1n v\u1eabn c\u00f3 th\u1ec3 gi\u00fap b\u1ea1n hi\u1ec3u th\u00eam v\u1ec1 kh\u00e1ch h\u00e0ng.<\/p>\n<\/section>\n

          \n

          Th\u1eed nghi\u1ec7m r\u1ed3i, v\u00ec sao v\u1eabn c\u00f3 th\u1ec3 hi\u1ec3u nh\u1ea7m?<\/h2>\n

          Nh\u1eefng ng\u00e0y \u0111\u1ea7u ch\u01b0a n\u00f3i l\u00ean t\u1ea5t c\u1ea3<\/h3>\n

          Vi\u1ec7c ki\u1ec3m tra k\u1ebft qu\u1ea3 m\u1ed7i ng\u00e0y v\u00e0 t\u1eaft th\u1eed nghi\u1ec7m ngay khi p-value xu\u1ed1ng d\u01b0\u1edbi 0.05 \u0111\u01b0\u1ee3c g\u1ecdi l\u00e0 peeking problem. C\u00e1ch l\u00e0m n\u00e0y khi\u1ebfn t\u1ef7 l\u1ec7 sai l\u1ea7m lo\u1ea1i I t\u0103ng l\u00ean t\u1edbi 30-50%.<\/p>\n

          B\u00ean c\u1ea1nh \u0111\u00f3, ng\u01b0\u1eddi d\u00f9ng c\u00f3 th\u1ec3 ph\u1ea3n \u1ee9ng v\u1edbi s\u1ef1 m\u1edbi l\u1ea1 c\u1ee7a giao di\u1ec7n. Kh\u00e1ch c\u0169 nh\u1ea5p v\u00e0o B v\u00ec th\u1ea5y kh\u00e1c v\u1edbi th\u01b0\u1eddng ng\u00e0y, khi\u1ebfn ch\u1ec9 s\u1ed1 t\u0103ng trong nh\u1eefng ng\u00e0y \u0111\u1ea7u r\u1ed3i gi\u1ea3m d\u1ea7n. \u0110\u00e2y l\u00e0 hi\u1ec7u \u1ee9ng m\u1edbi m\u1ebb (novelty effect).<\/p>\n

          Chi\u1ec1u ng\u01b0\u1ee3c l\u1ea1i c\u0169ng c\u00f3 th\u1ec3 x\u1ea3y ra. Nh\u1eefng kh\u00e1ch h\u00e0ng \u0111\u00e3 quen v\u1edbi giao di\u1ec7n c\u0169 c\u1ea3m th\u1ea5y kh\u00f3 ch\u1ecbu khi n\u00f3 b\u1ecb thay \u0111\u1ed5i, khi\u1ebfn ch\u1ec9 s\u1ed1 c\u1ee7a B gi\u1ea3m m\u1ea1nh l\u00fac \u0111\u1ea7u. Hi\u1ec7n t\u01b0\u1ee3ng n\u00e0y \u0111\u01b0\u1ee3c g\u1ecdi l\u00e0 hi\u1ec7u \u1ee9ng ch\u1ed1ng \u0111\u1ed1i (primacy effect).<\/p>\n

          Th\u1eafng nhi\u1ec1u l\u1ea7n nh\u1ecf, hi\u1ec7u su\u1ea5t v\u1eabn ch\u1ea1m tr\u1ea7n<\/h3>\n

          M\u1ed9t thi\u1ebft k\u1ebf \u0111\u00e3 l\u1ea1c h\u1eadu c\u00f3 th\u1ec3 ti\u1ebfp t\u1ee5c \u0111\u01b0\u1ee3c ch\u1ec9nh m\u00e0u n\u00fat, kho\u1ea3ng c\u00e1ch ch\u1eef v\u00e0 nhi\u1ec1u chi ti\u1ebft kh\u00e1c. D\u00f9 m\u1ed9t s\u1ed1 thay \u0111\u1ed5i th\u1eafng trong th\u1eed nghi\u1ec7m, hi\u1ec7u su\u1ea5t c\u1ee7a trang v\u1eabn ch\u1ea1m tr\u1ea7n.<\/p>\n

          \u0110\u00e2y ch\u00ednh l\u00e0 b\u1eaby c\u1ef1c \u0111\u1ea1i \u0111\u1ecba ph\u01b0\u01a1ng (local maxima trap). \u0110\u1ec3 tr\u00e1nh m\u1eafc l\u1ea1i \u1edf \u0111\u00f3, b\u1ea1n c\u1ea7n k\u1ebft h\u1ee3p nh\u1eefng th\u1eed nghi\u1ec7m vi m\u00f4 v\u1edbi c\u00e1c l\u1ea7n th\u1eed thay \u0111\u1ed5i l\u1edbn v\u00e0 t\u00e1o b\u1ea1o h\u01a1n.<\/p>\n

          \"\u0110\u1ed3
          Nh\u1eefng ch\u1ec9nh s\u1eeda nh\u1ecf c\u00f3 th\u1ec3 \u0111\u01b0a thi\u1ebft k\u1ebf hi\u1ec7n t\u1ea1i \u0111\u1ebfn gi\u1edbi h\u1ea1n. Mu\u1ed1n c\u1ea3i thi\u1ec7n h\u01a1n n\u1eefa, \u0111\u00f4i khi b\u1ea1n c\u1ea7n th\u1eed m\u1ed9t h\u01b0\u1edbng thi\u1ebft k\u1ebf kh\u00e1c.<\/figcaption><\/figure>\n

          Kh\u00e1ch nh\u1ea5p nhi\u1ec1u h\u01a1n ch\u01b0a ch\u1eafc mua nhi\u1ec1u h\u01a1n<\/h3>\n

          M\u1ed9t bi\u1ebfn th\u1ec3 c\u00f3 th\u1ec3 t\u0103ng 30% t\u1ef7 l\u1ec7 nh\u1ea5p n\u00fat, nh\u01b0ng \u0111\u1ed3ng th\u1eddi l\u00e0m gi\u1ea3m ch\u1ea5t l\u01b0\u1ee3ng kh\u00e1ch h\u00e0ng ti\u1ec1m n\u0103ng (lead) v\u00e0 t\u1ed5ng doanh thu th\u1ef1c t\u1ebf. N\u1ebfu ch\u1ec9 nh\u00ecn v\u00e0o b\u01b0\u1edbc t\u01b0\u01a1ng t\u00e1c ban \u0111\u1ea7u, b\u1ea1n r\u1ea5t d\u1ec5 b\u1ecf s\u00f3t nh\u1eefng g\u00ec di\u1ec5n ra \u1edf cu\u1ed1i h\u00e0nh tr\u00ecnh mua h\u00e0ng.<\/p>\n

          V\u00ec v\u1eady, b\u00ean c\u1ea1nh ch\u1ec9 s\u1ed1 nh\u1ea5p chu\u1ed9t, b\u1ea1n c\u1ea7n theo d\u00f5i song song c\u00e1c ch\u1ec9 s\u1ed1 b\u1ea3o v\u1ec7 v\u00e0 ARPU.<\/p>\n

          Kh\u00e1ch m\u00f9a sale c\u00f3 th\u1ec3 h\u00e0nh x\u1eed kh\u00e1c ng\u00e0y th\u01b0\u1eddng<\/h3>\n

          C\u00e1c \u0111\u1ee3t megasale nh\u01b0 Black Friday hay T\u1ebft c\u00f3 th\u1ec3 l\u00e0m thay \u0111\u1ed5i h\u00e0nh vi mua s\u1eafm th\u00f4ng th\u01b0\u1eddng, t\u1eeb \u0111\u00f3 g\u00e2y nhi\u1ec5u k\u1ebft qu\u1ea3 A\/B testing. M\u1ed9t phi\u00ean b\u1ea3n th\u1eafng trong \u0111\u1ee3t sale ch\u01b0a ch\u1eafc ti\u1ebfp t\u1ee5c hi\u1ec7u qu\u1ea3 khi ch\u01b0\u01a1ng tr\u00ecnh k\u1ebft th\u00fac.<\/p>\n

          B\u1ed1i c\u1ea3nh m\u00f9a v\u1ee5 v\u00ec v\u1eady c\u0169ng c\u1ea7n \u0111\u01b0\u1ee3c t\u00ednh \u0111\u1ebfn khi di\u1ec5n gi\u1ea3i k\u1ebft qu\u1ea3.<\/p>\n

          T\u00e1c \u0111\u1ed9ng v\u00e0o t\u00e2m l\u00fd c\u0169ng c\u00f3 th\u1ec3 ph\u1ea3n t\u00e1c d\u1ee5ng<\/h3>\n

          Theo nghi\u00ean c\u1ee9u t\u1eeb ti\u1ebfn s\u0129 Brian Cugelman v\u00e0 CXL, vi\u1ec7c t\u1ea1o s\u1ef1 khan hi\u1ebfm gi\u1ea3 ho\u1eb7c \u00e9p bu\u1ed9c th\u00f4ng tin trong th\u1eed nghi\u1ec7m A\/B c\u00f3 th\u1ec3 ph\u1ea3n t\u00e1c d\u1ee5ng. Khi nh\u1eadn ra m\u00ecnh b\u1ecb thao t\u00fang, ng\u01b0\u1eddi d\u00f9ng d\u1ec5 tr\u1edf n\u00ean ho\u00e0i nghi v\u00e0 r\u1eddi b\u1ecf th\u01b0\u01a1ng hi\u1ec7u v\u1ec1 l\u00e2u d\u00e0i.<\/p>\n

          \u0110i\u1ec1u kh\u00f3 nh\u1eadn bi\u1ebft l\u00e0 t\u1ef7 l\u1ec7 chuy\u1ec3n \u0111\u1ed5i ng\u1eafn h\u1ea1n v\u1eabn c\u00f3 th\u1ec3 t\u0103ng. B\u00e1o c\u00e1o cho th\u1ea5y m\u1ed9t bi\u1ebfn th\u1ec3 \u0111ang th\u1eafng, trong khi t\u00e1c \u0111\u1ed9ng ti\u00eau c\u1ef1c ph\u00eda sau l\u1ea1i b\u1ecb b\u1ecf s\u00f3t.<\/p>\n

          V\u1edbi n\u00fat \u201cMua ngay\u201d \u1edf \u0111\u1ea7u b\u00e0i, \u0111i\u1ec1u c\u1ea7n t\u00ecm hi\u1ec3u v\u1eabn l\u00e0 ng\u01b0\u1eddi d\u00f9ng c\u00f3 nh\u1eadn th\u1ea5y thay \u0111\u1ed5i hay kh\u00f4ng, v\u00e0 n\u00f3 t\u00e1c \u0111\u1ed9ng th\u1ebf n\u00e0o \u0111\u1ebfn c\u1ea3m nh\u1eadn c\u1ee7a h\u1ecd. \u0110\u00e2y c\u0169ng l\u00e0 c\u00e1ch nh\u00ecn A\/B testing nh\u01b0 m\u1ed9t c\u00f4ng c\u1ee5 \u0111o l\u01b0\u1eddng h\u00e0nh vi t\u00e2m l\u00fd, v\u01b0\u1ee3t ra ngo\u00e0i vi\u1ec7c l\u1eadp tr\u00ecnh hay thi\u1ebft k\u1ebf giao di\u1ec7n.<\/p>\n

          \u0110\u1ec3 nh\u1eefng b\u00e0i h\u1ecdc \u1ea5y c\u00f3 c\u01a1 s\u1edf, th\u1eed nghi\u1ec7m c\u1ea7n \u0111\u01b0\u1ee3c ch\u1ea1y \u0111\u1ee7 th\u1eddi gian, \u0111\u1ee7 m\u1eabu v\u00e0 ph\u00e2n t\u00edch b\u1eb1ng th\u1ed1ng k\u00ea nghi\u00eam ng\u1eb7t. Doanh thu t\u0103ng th\u00eam t\u1eeb m\u1ed9t bi\u1ebfn th\u1ec3 th\u1eafng l\u00e0 m\u1ed9t k\u1ebft qu\u1ea3, c\u00f2n m\u1ee5c ti\u00eau l\u1edbn h\u01a1n c\u1ee7a ch\u01b0\u01a1ng tr\u00ecnh v\u1eabn l\u00e0 t\u00edch l\u0169y hi\u1ec3u bi\u1ebft v\u1ec1 kh\u00e1ch h\u00e0ng, qua c\u1ea3 nh\u1eefng l\u1ea7n th\u1eafng l\u1eabn nh\u1eefng l\u1ea7n thua.<\/p>\n<\/section>\n","protected":false},"excerpt":{"rendered":"

          M\u1ed9t n\u00fat mua h\u00e0ng kh\u00e1c m\u00e0u ch\u01b0a ch\u1eafc khi\u1ebfn ng\u01b0\u1eddi d\u00f9ng mu\u1ed1n mua h\u01a1n. \u0110i\u1ec1u g\u00ec \u0111ang di\u1ec5n ra trong t\u00e2m tr\u00ed h\u1ecd? B\u1ea1n \u0111\u1ed5i n\u00fat \u201cMua ngay\u201d t\u1eeb xanh sang cam, th\u1eed 3 ki\u1ec3u font cho ti\u00eau \u0111\u1ec1 r\u1ed3i ch\u1ec9nh th\u00eam v\u00e0i pixel \u0111\u1ec3 trang ch\u1ee7 tho\u00e1ng m\u1eaft h\u01a1n. V\u1eady m\u00e0 sau 3 tu\u1ea7n […]<\/p>\n","protected":false},"author":1,"featured_media":5239,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1,36],"tags":[54],"class_list":["post-5201","post","type-post","status-publish","format-standard","has-post-thumbnail","category-cro","category-tam-ly-hoc-chuyen-doi","tag-phuong-phap-nghien-cuu"],"blocksy_meta":[],"_links":{"self":[{"href":"https:\/\/vbaseup.com\/kien-thuc\/wp-json\/wp\/v2\/posts\/5201","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/vbaseup.com\/kien-thuc\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/vbaseup.com\/kien-thuc\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/vbaseup.com\/kien-thuc\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/vbaseup.com\/kien-thuc\/wp-json\/wp\/v2\/comments?post=5201"}],"version-history":[{"count":2,"href":"https:\/\/vbaseup.com\/kien-thuc\/wp-json\/wp\/v2\/posts\/5201\/revisions"}],"predecessor-version":[{"id":5244,"href":"https:\/\/vbaseup.com\/kien-thuc\/wp-json\/wp\/v2\/posts\/5201\/revisions\/5244"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/vbaseup.com\/kien-thuc\/wp-json\/wp\/v2\/media\/5239"}],"wp:attachment":[{"href":"https:\/\/vbaseup.com\/kien-thuc\/wp-json\/wp\/v2\/media?parent=5201"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/vbaseup.com\/kien-thuc\/wp-json\/wp\/v2\/categories?post=5201"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/vbaseup.com\/kien-thuc\/wp-json\/wp\/v2\/tags?post=5201"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}