Cleaning the mess after a bad campaign management [case study]

Recently, I helped one business in the US to get their campaign on track using just simple manual bidding: no AI, no other bidding strategies, pure basics. But let’s start at the beginning…

 

A few months ago, around 7 pm, I was waiting to get Tera (my German Shepherd) for a walk. We usually go around 8 pm. So I had an hour to kill. I decided to go to my Facebook group, approve some members, review the posts, etc. I’ve checked my messages and saw this message from Gaby.

 

 

I think I haven’t replied that day. I actually though it’s a scam or just regular spam offering something, since I noticed the underlined link. Since I glanced at it for a second. Later on, I’ve read it and got intrigued. Not sure if it was the fact that couple of people already tried, or the the fact that it was an old business about to close. But I felt like this is my hero moment. I don’t have a cape, nor do I fly, but I know Google Ads.

Just think how that classic phrase about Superman would sound “Look! Up in the office building! It’s a guy! It’s a guy with laptop! It’s Google Ads guy!”. OK, moving on.

 

I kinda felt that I can help. It did not sound like a big account that would require a lot of my time. It sounded like a local business, that probably have a wrong set up, maybe other basic mistakes. So I agreed to help Gaby out and not charge anything.

 

In this post I will lay it all out. What I found and what I changed and what were the results. Honestly, there are many similar accounts that don’t need too much work and can turn around their performance.

 

Everything worked fine until…

 

This is something Gaby told me. They had a guy who worked on the account several years and everything looked fine. They were getting leads, the business was doing great.

 

Then they started getting less and less leads and the cost per lead started to increase. This is the screenshot of two campaigns that were running on October, 2022. About a year ago.

 

Everything looks great. They have two conversions: a call conversion and a lead form. Desired price per conversion is about $20, so they were not far off. Also, you can see that they are getting more lead submits than phone calls. They used to get around 100 conversions per month.

 

According to Gaby, everything started going down around spring. Since the account is quite old it’s hard to dig in and review what was changed. But there were few things that stood out.

 

As you can see here, as soon as call conversions went into the mix, the lead submit conversion went down (the blue line).

 

Let’s dissect everything further.

 

 

Finding what else could be wrong

I’m sure there are smaller things that I could have found, but we needed fast actions and turn things around. The easiest way is to look at a longer period of time and check what campaigns performed, what was changed and how it impacted results.

These stood out the most.

 

 

Auto suggestions

Google always suggest things to you, that doesn’t mean it’s a good suggestion. Most assume that if Google says its good, they should apply it. But it’s wrong. Actually, you should not apply any suggestions unless you reviewed them carefully.

 

Look at how many auto suggestions were applied. 540 new keywords were added, that is a lot.

 

You should turn these off. Go to Recommendations in your account. Find Auto-apply and click it and unselect everything.

 

 

Unless, you review every suggestion, you have no idea how it will affect your performance. And after a year, it might be a daunting task to untangle everything.

 

 

Search and Display together

When creating a Search campaign there is a check box that allows you to show your ads in Display network as well. Don’t ever use that.

 

 

You can notice the Search and Display campaign together indicated by an icon. These are two different networks and should be separated in different campaigns. The screenshot above is from this years campaign, when one of the guys from Fiverr tried to fix it. To me, combining Search and Display, indicated lack of experience.

 

Don’t be fooled by a general cost per conversion. Some of them are duplicate, others are call conversions. The actual cost is closer to a $100.

 

Smart campaigns

I hate these. This is Google’s attempt at handling everything. Using AI to replace humans. And so far, they fail miserably. Thankfully, there wasn;t a lot of spend, and whoever managed this realized that it’s s**t.

 

 

It still managed to spend $180, which is not a lot, but that could be turned into at least 6 leads. Instead, it created somekind of Call and Directions conversion and optimised towards it. WTF.

 

At least, the icon is nice, magic wand. Even if you are a Harry Potter fan, please never use Smart campaign. There is nothing magic about it.

 

 

Locations

One location per campaign. Period. I don’t know the though behind some of the campaigns. But this campaign seems to be directed to Puerto Rico. Which is great.

 

 

Then why, when you go into locations you see this:

 

 

This is not only US states, you can see cities as well. It’s great to segment one country by state or city in one campaign. Sometimes even have a separate campaign for state or city.

 

But not in one campaign. Sure, you can see the statistics by location. But can you easily see it next to each keyword or ad? And should you show same ads for Puerto Rico and USA? The answer is no.

 

I might be separating a specific state into a different campaign. I’ve noticed that Fort Lauderdale, a city in Florida is performing better. So I might want not to limit the budget there and maybe increase CPC. For that I need my separate campaign.

 

 

One primary conversion

When you create conversion in Google Ads account, by default it’s a primary conversion. I talked a out it in my last post about micro conversions.

 

Primary conversion is used for bidding and optimization, that’s why it should be your main or macro conversion. It is reported in the Conversion column.

 

Secondary conversion is only displayed near your campaigns and keywords. You have to add All conversion column and use Segments to see your other conversions. I will come back to that.

 

I’ve noticed that this account had a lot of Primary conversions. Some imported from GA4, some created in Google Ads. Those conversions will report different results. I also wrote about it here.

 

You have to either have one primary conversion. Or select a conversion when creating a campaign. By default you campaign uses account Primary conversions. And if there a lot of them, it might get confusing as in Gaby’s account.

 

 

Go to Campaigns settings, find Goals and change it to whatever conversion you want this campaign to be optimized for. Obviously, stick to one conversion per campaign. Its easier for the algorithm to understand what is important for your business.

 

 

Automated bidding strategies

When reviewing changes I also noticed, that one of the freelancers changed the bidding strategy from Maximize conversions to Maximise clicks. Probably wanting to salvage the campaign, thinking that more clicks might increase conversions. Nope.

 

 

Automated  bidding strategies work great if you have a lot of conversions or can get a lot of conversions in a short time (few weeks). Then Googles machine learning algorithm catches on quickly and starts optimizing your campaigns.

 

When you don’t have a lot of data (conversions), it takes time for it to learn. During that time Google will try lot’s of different stuff, like increasing bids by a lot, which usually causes high cost increase with 0 conversions. This is the most often heard complain by people switching to automated bidding.

 

Maximise clicks is the worst thing you could try. This strategy does what it says, tries to get as many clicks as possible with your given budget. Does it care about the quality of those clicks? No. Sure, it will use your keywords, it won’t stray too far. But with the broad match about several years back. Google’s understanding of broad keyword became very broad.

You might end up getting traffic that is not ready to convert at all. Even though it might be your target audience. One keyword does not equal to another.

 

I’m not saying this strategy can’t work. If you’re know what you’re doing, you could make it work. If you website already converts great, you have you funnel ready and all you need to do is get traffic fast. You might start with that strategy and then change to CPA bidding. I’ve heard people say it works, but I have never seen it. The question is what they define “works”, how they measure it and if the claims are true.

 

 

What happened after cleaning up everything?

I started working towards the end of October. Not the best time as the holiday season crazines inflates the prices a lot. My approach was to find what keywords worked and start fresh.

 

Obviously, I fixed what I mentioned above. But slowly conversion priced dropped to about $30, that’s from around $100 in September. And they got around 60 conversions. Which is also great since the budget is around $2000.

 

 

Everything on manual CPC. The goal was to look at each keyword and understand it’s impact on the business. At the same time, scale slowly to not exceed the budget, but try to get those sales coming in.

 

I still have some work to do, but December is not the best time to do it. Also, I’ve noticed that everything drops around November-December, by looking at the last 2 years. And it makes sense since people start to concentrate on the holidays, gifts, parties and not transporting cars.

 

I chat to Gaby often as I need to understand how things perform. She mentioned that they started getting better leads, the sales increased. That’s a good indication that the changes work.

 

We also made few changes on the website. Nothing major, since it works I don’t want to ruin it. Can it be update, of course. But let’s not for get that this is a small business and working on a website requires time and money.

 

At the same time working on qualifying the leads better. There a solution already built-in, but we’re having some problems passing the right information. But that’s details.

 

 

 

Why should you care?

This is one example of many. You don’t need fancy strategies and AI to make it work. Sure, they can help, but not all the time. Even though Google is moving towards AI and more and more features will be based on AI, it is still too early. Even their latest Performance Max campaign is doing great, but not all the time and not in all industries and all accounts.

 

What you should take from this:

• Basics are still there. Google hasn’t went full AI mode. Hope that day never comes.
• Don’t blindly apply Google recommendations in the account. You know better.
• Less using best practices from gurus, more testing and finding it on your own.
• Outsmart not outbid. You can make the business work even with smaller budgets. You just have to be smart about it.
• Choose freelancers carefully. Cheaper is not better