Menu
Improvements Explained

Improvements explained: excluding search partner sites

Guillaume Devinat.

Guillaume Devinat

June 15th 2021
5 minute read

In this post we share an analysis of search partner performance across 10,000+ accounts, how to manually optimise search partners, and an insight into the more advanced optimisation logic in Opteo.

Opteo blog image.
Opteo blog image.

Search partner targeting is a classic example of an on-by-default setting which warrants scrutiny. While we might expect similar performance between search partners and Google search, a recent analysis by our data science team tells a different story.

Before we jump into the research and some useful nuances in optimising search partners, let’s quickly recap the basics.

What are search partners?

Searches that happen on sites other than Google itself fall into the “search partner” category. There are more than we might imagine!

First, there are sites such as YouTube and Amazon that embed ads from Google after a visitor searches for a video or product. Then there are large websites like The Guardian or W3Schools that use Google Custom Search Engine (CSE) to power their site searches.

Google Custom Search Engine (CSE) to power site searches.
Google Custom Search Engine (CSE) to power site searches.

Currently, there are 273,396 live websites using Google Custom Search. That’s a lot of potential search partners! Given the scale of search partners, we wanted to find out to what extent they are a worthwhile investment.

Search partner data analysis

We looked at a large dataset of accounts that have not used any Opteo Improvements to make sure that the data is a true representation of real-world Google ads accounts. We then filtered the dataset so that we were left with over 10,000 Google Ads accounts with 10 or more conversions in both Google Search and search partners.

CPA delta between Google Search and search partners
CPA delta between Google Search and search partners

Looking at the difference in CPA between Search and search partners, the median difference puts search partner conversions at ~7% more expensive than the equivalent search conversion for the same account.

We plotted the differences in CPA on the histogram above. The distribution has a slight positive skew and a central tendency but also some surprisingly long tails. In other words, search partner performance can vary wildly, as such, we should keep an eye on it. Let’s look at how to evaluate and optimise search partners for your accounts.

How to manually analyse and exclude search partners

The basic workflow of optimising Google search partner performance involves selecting the relevant campaigns, breaking performance out by network, and comparing results. For campaigns where search partner performance is unacceptable, you'll find the option to opt out of search partners in campaign settings.

Here's a step by step guide for reference:

  • First, filter your account for Search and standard Shopping campaigns (only select enabled campaigns).
  • Second, pick a date range where both Search and Search Partners were running alongside each other to ensure a fair, apples-to-apples comparison.
  • Third, isolate Search Partner performance by selecting “Network (with search partners)” as a segment.
Segmenting by network to break out search partner performance
Segmenting by network to break out search partner performance
  • Fourth, compare search partner CPA vs. Google Search CPA. You can do this for individual campaigns, or as a total at the bottom of the campaign view.
Comparing search partner CPA vs. Google Search CPA
Comparing search partner CPA vs. Google Search CPA
  • Fifth, if Search Partner performance is unacceptable, you can disable them under the “Networks” heading in an individual campaign's settings.
Turning search partners off in campaign settings
Turning search partners off in campaign settings
  • Finally, watch out for auto-applied recommendations! Many advertisers don’t realise Google may be making changes to their accounts behind the scenes. There's no use in switching off search partners if Google is going to switch them back on automatically!
Switching off auto-applied recommendations in Google Ads
Switching off auto-applied recommendations in Google Ads

Taking search partner optimisation a step further with Opteo

When ads perform poorly on search partner sites, Opteo will recommend disabling search partners for the relevant campaigns. Our algorithms carry out the analyses for you, taking into account a few additional nuances like carefully comparing performance to targets and ensuring there's enough data to make decisions.

Search Partner optimisation with Opteo
Search Partner optimisation with Opteo

Using campaign groups to aggregate performance data

One of the most common roadblocks to making decisions about optimisations, like leaving search partners on or turning them off, is having enough data. If we're to simply compare search partner performance to Google Search performance campaign by campaign, the likelihood is, most campaigns wouldn't have enough data to make a reliable decision.

To solve this, we can group campaigns together and look at their aggregate performance. But in order to do that, we need to understand which campaigns should and shouldn't be grouped together. For example, you probably wouldn't want to mix your brand and generic campaign performance data together because the acceptable performance is so vastly different for each of them. They may even use different conversion types.

In Opteo, setting your campaign groups tells us which campaigns belong together, which conversion types are relevant, and what the target is. Once campaign groups are in place, we're able to correctly aggregate campaign performance data and compare it to the relevant target. We're also able to understand a particular campaign's performance relative to the other campaigns in its group (useful when there is no target for instance).

Creating campaign groups enables us to aggregate data and make better decisions
Creating campaign groups enables us to aggregate data and make better decisions

Campaign groups and performance targets set the groundwork for our decision making. If search partners are performing worse than their target, switch them off. If they are performing at or better than their target, leave them on. But it’s not that simple, let’s layer in some refinements.

Imagine a scenario where within a campaign group, search partners are performing worse than your target, but better than Google Search.

 A scenario where search partners performs worse than the target but better than Google Search
A scenario where search partners performs worse than the target but better than Google Search

In this case, it'd be a mistake to cut out search partners. We would be better served cutting costs elsewhere. So we can refine our decision making to only suggest excluding search partners where the CPA is above (worse than) both Google Search CPA, and the target CPA.

Breaking out high volume campaigns

It’s reasonable to assume that even in campaign groups where search partners perform poorly, there may be individual campaigns where they do work. Instead of switching off search partners for all campaigns inside a campaign group, we should pick out campaigns which have enough volume to be evaluated by themselves.

To figure out which campaigns have enough volume to warrant a campaign level evaluation of search partners, one method we can use is to compare its clicks to the minimum click threshold of its campaign group as below:

minClicks = (1/AvgConvRate)*X

If a campaign’s search partner clicks is above the minimum clicks threshold (ie. it has accumulated enough clicks to on average produce X conversions) then we can evaluate that campaign by itself. We take a deeper look into minimum clicks in Improvements explained: adjusting keyword bids in Opteo.

This way, important high volume campaigns can be treated individually, and lower volume campaigns (which do not have enough data to make a per campaign decision) can be grouped together and treated as a collective.

In the example below, campaigns one, two and three each have more clicks than the minimum click threshold and so are treated individually. Campaign one and three have search partner CPAs above their campaign group's target, so search partners should be turned off. Campaigns 4-8 don't have enough clicks to pass the minimum clicks threshold, so their performance data is aggregated. The aggregate search partner CPA is above the campaign group's target, so search partners should be switched off for campaigns 4-8 as well.

In this example, we would disable search partners for Campaigns One, Three and Four to Eight, while leaving them enabled for Campaign Two.
In this example, we would disable search partners for Campaigns One, Three and Four to Eight, while leaving them enabled for Campaign Two.

Wrap up and takeaways

I hope you picked up something useful from this post. Here’s a quick recap of the key points:

  • Search partner performance varies wildly campaign by campaign and account by account so it's certainly worth monitoring regularly.
  • Evaluate search partner performance by comparing their CPA (or ROAS) to both Google Search and your target for that campaign.
  • Use the “minimum clicks” method to identify campaigns that have a high volume of search partner clicks. Make a decision about search partners campaign by campaign for high volume campaigns.
  • For campaigns that don’t pass the minimum clicks test, consider grouping their performance data together (only if they belong to the same campaign group) so that you have enough data to make a decision.
  • Look out for auto-applied recommendations — amongst other potentially undesirable changes, auto-apply might well switch search partners back on.
  • Given the large variance in search partner performance, it could be worth testing them again if they’ve been turned off as standard, especially if you're after more volume. You can rely on a tool like Opteo to flag any inefficiencies that need to be dealt with.

Explore more articles