没有重定向的Google Analytics实验：会话的分布不均

Question

抱歉我的英文。

我在Google Analytics中创建了新实验，并使用了method without redirects。我使用此内容在Google跟踪代码管理器中创建了代码

#script with api
<script src="//www.google-analytics.com/cx/api.js?experiment=1EN3yB34T0WE-0WxSqvsKw"></script>
<script>
    # item ids to launch test
    var itemIds= ['898324', '914734', '904249', '936082', '904275', '951077', '977818', '942227', '933907'];
    if( $.inArray({{productId}}, itemIds) !== -1 ) {
        # get chosed variation
        var chosenVariation = cxApi.chooseVariation();
        $(function(){   
            if (chosenVariation == 1) {
                $('div.orange.service').hide();
                $('div.orange.cardif').hide();
            }
        });

        # make non-interaction hit to send data. Maybe here i get mistake 
        #(hit making by send data to Google Tag Manager, and all events with event === OWOX - will send to Google Analytics)
        dataLayer.push({"event":"OWOX","eventCategory":"Non-Interactions","eventAction":"show","eventLabel":"products","products":[]});
    }
</script>

经过7天的实验，我们得到： big gap between original variant and variant with alternative design

我想问为什么原始页面与具有交替设计的页面之间的差距如此之大？