CoRe Hackathon Results

[ashammi228]

First place - team BI47: Alexander Naletov, Nikita Golubtsov, Maxim Zelensky and Alexey Markov.

Their solution helps to quickly assess how things are going with advertising in Yandex.Direct. For example, the system warns about "expensive conversions" in a campaign — too high a CPA — and about low conversion for specific ads. At the same time, the system calculates the average number of clicks required for conversion, CPA for first-click and last-click attribution models, and also recommends raising or lowering the bid based on the specified CPA value. And for each campaign and each type of call ("primary", "targeted", "commercial real estate", etc.), the average number of visits before the phone call is determined.

Second place - FirstPartyData team: Alexander Morin, Dmitry Voronin, Denis Sobolev and Pavel Maksimov.

The team has developed a universal set of telegram gc philippines dashboards that allow tracking not only the effectiveness of advertising campaigns, but also the quality of the sales department. Three attribution models help to evaluate CPA for different channels that provide calls: U-shaped (the first and last clicks receive 40% of the value, and the remaining 20% ​​is distributed among other sources), a model based on the duration of the visit and on the depth of viewing.

There is also a conversion forecast for each key phrase and campaign, a detailed dashboard for tracking missed calls broken down by month and day of the week, and a visualization of the ratio of different types of calls. A separate dashboard displays data on deferred calls — those that occurred not immediately after the first visit to the site, but later: the system calculates the average number of days before a deferred call and shows such calls broken down by traffic sources.

Third place – UraAnalytics: Sergey Losev, Yuri Batievsky, Anton Astakhov and Ivan Ivanov.

The team members focused on the groundwork for end-to-end analytics - a table where visitor actions in Metrica and calls from CoMagic are combined into a single timeline.

The team also developed useful dashboards — visualization of the cost of a lead, a graph of changes in advertising costs in conjunction with the volume of purchased traffic, and a display of the dynamics of campaign efficiency in different attribution models. And a dashboard with data on the quality of call processing helped to identify an interesting dependency: most missed calls occur at the beginning of the working day and immediately after the lunch break.