Systems and methods for self-serve marketing pages with multi-armed bandit

A method implemented in a computer system, wherein the computer system comprises

• an interface database containing a plurality of graphical user interfaces (GUIs), the plurality of GUIs including a marketing campaign creation interface, a marketing campaign editor interface, a marketing campaign upload interface, a marketing campaign approval interface, one or more marketing campaign preview interfaces, and one or more marketing campaign save interfaces;
• a self-serve tool comprising one or more of the plurality of GUIs, the self-serve tool in data communication with the interface database to provide the one or more of the plurality of GUIs upon receipt of a selection;
• a marketing campaign huh in data communication with the self-serve tool, the marketing campaign hub configured to generate an external-facing webpage experience shown to users or potential customers visiting webpages or winning webpages selected for their consumption based upon contents to be displayed on the one or more of the plurality of GUIs;
• and
• a multi-armed bandit engine application programming interface (API) in data communication with the self-serve tool and the marketing campaign huh, the multi-armed bandit engine API being registered with the marketing campaign huh, and comprising rules for analyzing webpages published by the marketing campaign hub and for serving a winning webpage from the webpages,
• the method comprising:
• receiving, by the multi-bandit API, a plurality of creatives for a webpage, the webpage being published by the marketing campaign hub to a viewer and resulting in a viewer experience, wherein the plurality of creatives include at least one selected from a group of an advertisement and a selected landing page;
• measuring, by the multi-bandit API, a result based on the viewer experience, the result comprising conversion rates for a plurality of variants of the webpage, and the variants being categorized based on a user selection, wherein the conversion rates include at least one selected from a group of a percentage number of webpage views to related webpage views or a percentage value of website visitors that execute a threshold number of webpage clicks;
• determining, by the multi-bandit API, confidence intervals in association with the conversion rates;
• dynamically comparing, by the multi-bandit API, the received creatives and the conversion rates;
• automatically analyzing, by the multi-bandit API, the variants based on the comparison;
• automatically selecting, by the multi-bandit API based on the analysis and the categorization of the variants, a winning webpage variant, the winning webpage variant exceeding a computed threshold, wherein the winning webpage variant is selected based on a probability that maximizes an expected reward using a formula
• ∫
• ∏
• [
• E
• ⁡
• (
• τ
• ❘
• α
• *
• ,
• x
• ,
• θ
• )
• =
• max
• α
• ′
• ⁢
• ⁢
• E
• ⁡
• (
• τ
• ❘
• α
• ′
• ,
• x
• ,
• θ
• )
• ]
• ⁢
• P
• ⁡
• (
• θ
• ❘
• D
• )
• ⁢
• d
• ⁢
• ⁢
• θ
• ,
• Π being an indicator function, α being a webpage, τ being a reward, θ being a parameter of a distribution of τ, x being a context, and D being past observation triplets;
• and
• automatically adjusting, by the multi-bandit API and the marketing campaign hub, online traffic such that the selected winning webpage variant is displayed more frequently than other webpages.

A non-transitory computer-readable medium storing instructions executable by a computer system, wherein the computer system comprises:

• an interface database containing a plurality of graphical user interfaces (GUIs), the plurality of GUIs including a marketing campaign creation interface, a marketing campaign editor interface, a marketing campaign upload interface, a marketing campaign approval interface, one or more marketing campaign preview interfaces, and one or more marketing campaign save interfaces;
• a self-serve tool comprising one or more of the plurality of GUIs, the self-serve tool in data, communication with the interface database to provide the one or more of the plurality of GUIs upon receipt of a selection;
• a marketing campaign hub in data communication with the self-serve tool, the marketing campaign huh configured to generate an external-facing webpage experience shown to users or potential customers visiting webpages or winning webpages selected for their consumption based upon contents to be displayed on the one or more of the plurality of GUIs;
• and
• a multi-armed bandit engine application programming interface (API) in data communication with the self-serve tool and the marketing campaign hub, the multi-armed bandit engine API being registered with the marketing campaign hub, and comprising rules for analyzing webpages published by the marketing campaign hub and for serving a winning webpage from the webpages,
• the instructions, when executed by the computer system, cause the computer system to perform operations comprising:
• receiving, by the multi-armed bandit API, a plurality of creatives for a webpage, the webpage being published by the marketing campaign hub to a viewer and resulting in a viewer experience, wherein the plurality of creatives include at least one selected from a group of an advertisement and a selected landing page;
• measuring, by the multi-armed bandit API, a result based on the viewer experience, the result comprising including conversion rates for a plurality of variants of the webpage, and the variants being categorized based on a user selection, wherein the conversion rates include at least one selected from a group of a percentage number of webpage views to related webpage views or a percentage value of website visitors that execute a threshold number of webpage clicks;
• dynamically comparing, by the multi-armed bandit API, the received creatives and the conversion rates;
• automatically analyzing, by the multi-armed bandit API, the variants based on the comparison;
• automatically selecting, by the multi-armed bandit API, based on the analysis and categorization of the variants, a winning webpage variant exceeding a computed threshold, wherein the winning webpage is selected based on a probability that maximizes an expected reward using a formula
• ∫
• ∏
• [
• E
• ⁡
• (
• τ
• ❘
• α
• *
• ,
• x
• ,
• θ
• )
• =
• max
• α
• ′
• ⁢
• ⁢
• E
• ⁡
• (
• τ
• ❘
• α
• ′
• ,
• x
• ,
• θ
• )
• ]
• ⁢
• P
• ⁡
• (
• θ
• ❘
• D
• )
• ⁢
• d
• ⁢
• ⁢
• θ
• ,
• Π being an indicator function, α being a webpage, τ being a reward, θ being a parameter of a distribution of τ, x being a context, and D being past observation triplets;
• and
• automatically adjusting, by the multi-armed bandit API and the marketing campaign huh, online traffic such that the selected winning webpage is displayed more frequently than other webpages.