At this point, you should be a wizard (or wizardess) when it comes to troubleshooting PPC performance issues. Identifying root cause and pinpointing dimensions that are dropping your numbers should be second nature. However, you might find that there are no large-scale problems and yet you still aren’t meeting your business goals. You’re starting to get pressure from your superiors to figure out a way to turn things around. So what now?
If you find that root cause and dimensional analyses aren’t leading to actionable insights, the issue might be bigger than you realize – you might not be optimizing to the correct metric altogether.
Optimization Metrics are the Be-all and End-all
What is an Optimization Metric?
An optimization metric is a metric you use in order to determine what bid you would like to place on a keyword, which bid adjustment you would place on a device, etc… Optimizations will always, to some degree, be related to your business goal. As an example, if your business goal is to achieve a monthly ROAS of 150%, revenue should be a part of your optimization metric, meaning you would increase bids on keywords generating more revenue, and decrease bids on keywords generating less revenue. If your goal is to hit a CPA of $20, conversions would be a part of your optimization metric. Taking the same principles as ROAS, you would bid higher on keywords generating more conversions, and down on keywords generating fewer conversions.
These are low-funnel metrics, which sometimes aren’t enough to get you where you need to be. For example, if your goal is CPA, and you only sell a small portion of conversions per month but have tens of thousands of keywords, you would only be able to effectively bid on a handful of keywords, when realistically, there are a plethora of keywords that might give you similar, or better, opportunities to capture cheaper conversions. Utilizing higher-funnel metrics would help uncover these keywords, which could potentially be cheaper in the long run, increasing your efficiency.
Different Types of Optimization Metrics to Optimize Towards
Revenue and conversions. These are the metrics that matter. In a perfect world, these should be the only numbers worth optimizing towards, but in reality, your program might have a very low-volume, low-funnel metric, such as a low volume of conversions. In this case, you need to dig a bit deeper.
Hybrid metrics are a mix of high-funnel and low-funnel metrics. These are extremely valuable when your lower-funnel metric is what you ultimately care about yet too low volume for it to make sense to optimize toward.
For example, let’s say you’re selling a computer game, and your goal is to hit a certain number of conversions efficiently each month (i.e. you want to spend x amount monthly for someone to purchase your game). However, you have tens of thousands of keywords you are bidding on, and historically you only sell 100 computer games each month. This means that at most, only 100 keywords in a given month will lead to someone buying this computer game. It’s more likely that there are only a handful of keywords driving conversions, and they might be doing so inefficiently.
Now let’s say your click to conversion path is the following:
From this, you can see there are actually two ways a user can make the journey from a click to conversion:
- By clicking directly on the button that leads to
a purchasefrom the ad.
- By clicking on the button that lets you learn more about the game, and then clicking the button that leads to the purchase from the ad.
In this case, the higher-funnel metric is the button that lets you learn more about the game (let’s refer to this as a lead) and there is a lower-funnel button that is directly responsible for your purchase (let’s refer to this as a sale).
If you were to collect data on ad clicks over a long period of time, you would likely see that a subset of all sales came from leads, and a large proportion of your leads drove zero sales. This information should be enough to give you an indication of what ratio of leads to sales you would expect, and this should help you determine a hybrid metric that you can apply to all the other keywords in your account.
To illustrate: you might find on average that for every 40 leads you generate, you make a sale, and there are a great many keywords that are generating leads for a cheaper sum than some of the keywords you are currently bidding higher on that are generating little to no sales. With this information, you could do two things: first, start bidding higher on cheaper keywords that historically generate high lead volume and have a good chance of leading to a sale; and second,
If you set up the hybrid metric correctly, you should still be bidding high enough on keywords that have solid sales volume, to a point that sales themselves weigh more in your optimization strategy than leads, while also preventing you from spending too much money on keywords that led to a sales by random chance.
When looking at the users who interact with your website, some will be inherently more valuable than others. Users who buy certain products might be more likely to come back and make further purchases in the future. As an example, if you are selling a subscription-based service, and someone buys a certain subscription that lasts three months with certain features. This user might be likely to come back three months later and buy a
Ideally, you’d like to spend the majority of your budget targeting people more likely to consistently buy products from your website. If you collect this type of information in your CRM, you can create a Lifetime Value (LTV) Model to determine what types of users are more likely to buy a product from your website again in the
In terms of keyword-level bidding, if your CRM attributes a user’s purchase to the last click, you can tie values of your LTV model back to the last click of the keyword and subsequently use the LTV metric as a tool for bidding.
The only caveat to this type of bidding is when assessing the value of a keyword, you are more prone to bid too high or too low on a keyword if the LTV model isn’t accurate, so you would have to recalibrate your LTV model on a consistent cadence, especially if seasonality plays a role in the LTV model.
Your Business Model Will Impact Your Optimization Strategy
Understand how your website and business goals relate to your website
What type of metric you optimize to is dependent on how much data you have readily available, how your click to conversion operates, behavior of your users, and so on. Questions you should ask yourself when determining which type of metric to optimize towards are:
- What is my overall business goal? (Profit Maximization, CPA, Brand Presence, etc?)
- Do I have enough data directly related to my business goal that I can optimize my bids towards?
- If not, are there other aspects of my website, or user behavior, that I can use to generate a proxy metric to optimize towards?
Asking these types of questions will guide you towards the best optimization metric you can utilize in terms of giving yourself the best shot of reaching your overall business goals for PPC.
Why is My PPC Performance Bad?
This article brings to a close our mini-series looking at the intricacies of why your PPC performance may not be at the desired levels. In Part I, we touched upon the need to understand all aspects of your program including your weaknesses and how, with the right approach and the right tools, you can effectively identify the root cause and begin making effective changes to ensure your numbers are back trending in the right direction. Part II delved into the specific dimensions (device, location, audience, etc…) that might be causing problems and how to troubleshoot them.
Addressing the weaknesses in your program by understanding which campaigns, ad groups, keywords, and product groups are driving bad performance can help you not only resolve your performance issues in the short-term, but help you determine which segments need more attention in the long-term to help achieve – and exceed – business objectives year over year.
The post Why is My PPC Performance Bad? Part III: How to Ensure You’re Optimizing Towards the Correct Metric appeared first on QuanticMind.