Make Your Performance Checks Count
As SEM data becomes more complex, it’s become harder and harder to keep an eye on things. While it’s easy to spot overall performance trends, there are many underlying factors and segments that don’t get the attention they deserve. Between understanding splits between brand and non-brand, search and display, images ads and text, search and search partners, different match types, ad extensions and placements – it’s hard to know everything that’s going on at all times. Today we’ll discuss some of the most common performance issues and ways to make regular performance checks more impactful. Over the next few weeks, we’ll discuss different segments and figure what to look for in those instances.
It’s not uncommon for one major segment of an account to grow while another struggles. One of the more common instances involves overall search and content performance. As Google’s pushed GDN aggressively, content now takes up a large share of spend and can easily mask downturns in search performance. The opposite has always been true, as search was the first network to really dominate spend in AdWords. As Google adds complexity, it adds tools for measuring performance. Daily checks should include a trip to the Networks tab. Its biggest limitation is the lack of segmentation options, so both the main campaigns tab and dimensions tab become critical.
I like to do as much work as possible in the UI, but one of its bigger limiting factors is that you can only use one segment at a time. This can be fixed by simply exporting your view into Excel and applying a second segment on download. I highly recommend regularly reviewing a combination of networks as well as five or more week-over-week trends and day-over-day trends for the last 10 business days. Having multiple segments makes the data more convoluted, so limiting the number of rows is key. Given Google’s reliance on Mon.-Sun. ranges for WoW, adding in a DoD component is critical to catch trends early. I prefer to use 10 days to capture the last three days and their counterparts from the previous week. Most accounts experience consistent weekday vs. weekend trends, so comparing Tuesday to Saturday probably won’t help much. If anything, it might cause you to think your account is being affected by some adverse market conditions when, in fact, this is just normal day-of-week behavior.
During these performance checks, it’s important to stay focused and know what you are looking for. It’s easy to get distracted when dealing with these large data sets. With two segments, the number of factors increases substantially. Assuming you’re starting with a basic set of 10 KPIs, then looking at network data (search, search partners, managed placements and auto placements), and finally looking at the data for five distinct weeks – you’ve suddenly turned 10 KPIs into 200 data points. Narrowing the focus on critical components will help you stay on track.
First order of business – is this a volume or margin problem? If dealing with volume, work from the top down from impressions, to clicks, to total conversions, finishing with conversion rate. On the margin side, start with your ROI metric of choice, then go to CVR, and finish with CPC. There you will quickly identify where the margin issue exists and which of the two major factors is influencing it: conversion side or price of the auction.
Working your way from big numbers down makes it easier to spot overall trends. While a drop from 100 conversions to 90 represents a 10% dip, the trend is a bit more powerful if you spot it at the impression level – e.g. 1mm impressions vs. 900k. Ultimately, how one processes data is a personal preference, so try it out and see what works.
Short of having some piece of technology or enough savvy to know how to leverage the Google API, you need to make do with what you’ve got. Automated reports, saved filters, and performance alerts are your first line of defense against performance fluctuations. Use these tools to create a simple top-down analysis process that starts with the overall account and breaks it down into smaller and smaller chunks. Each layer of analysis should get you closer to your answer and, with a little luck, you might just figure out what’s wrong (or right) with your account in minutes rather than after hours of wild analysis.
– Sean Marshall, Director of Search Engine Marketing