This account is a mess!
A structure that makes no sense. Thousands or even hundreds of thousands of keywords. A mix of SKAGs and huge ad groups with loads of unrelated keywords.
Gibberish naming convention. What on earth does “skg+bm+s+sp+new -final-copy” mean? How is it different to “alpha_exp_2018_mcnv_lb”?
22 conversion actions. Some broken, some with no recent conversions. And nobody who knows for sure which ones are important.
But, you're going to turn this account around - more conversions, lower costs, 10 quality scores etc.
There’s a lot riding on getting this right. Maybe it's a brand new client and you want to prove that you’re better than their previous agency. Maybe you've just been hired or promoted.
You're going to be a hero when you get this account rocketing. But where do you start?
Even though it's a mess it is generating some leads. You know you need to prune, restructure, and introduce some sanity, but how do you fix the account without killing it? It’s like sticking a screwdriver into an airplane engine, at 20 000 feet.
Here is a collection of principles, tips and ideas that’ll help you figure out where to start, what to do next and how to avoid a disaster.
I took over an account from a large agency. At first glance it looked like the account was built sensibly and managed well. (The client had fired the agency for reasons unrelated to performance.)
The client repaired mobile phones.
There was a campaign for each brand - iPhone, Samsung, Huawei etc. Each brand had a landing page with a repair price calculator. You started calculating the repair cost by choosing the phone model. The iPhone landing page had a button for iPhone SE, a button for iPhone 6, a button for iPhone 8 and so on.
If the visitor accepted the price and entered credit card details it counted as a conversion.
I knew nothing about the economics of the phone repair industry so I asked the client for a lesson. He told me three important things:
As a phone got older the cost of repairing it approached the cost of buying a better second-hand phone. This meant that the older the phone, the less likely they were to complete the sale.
They had to lower the repair prices as phones got older. They’d charge, and make, much less for an iPhone SE screen replacement than they would for an iPhone X.
They didn't repair phones older than five years.
None of this vital business knowledge made it into the Google Ads account.
The client was paying for clicks for phones they didn't repair. For instance, the iPhone 3G or Samsung Galaxy SII. The repair price calculator didn’t even have an iPhone 3G or SII button. There was no way to convert those clicks.
The account ignored the difference in value between a repair of a one year old phone and a four year old phone. They treated every conversion as if it had the same value.
The account matched the business at the time it was set up, but it had never been updated. Nobody at the agency had asked “Should we still be advertising this?” Or “Are these conversions still worth the same?”
A new account almost always looks like a mess at first because reading an account is hard. Unless you had a brilliant handover, you don't have much to go on. Everybody has their own naming convention. The change history tells us what was done, but not why it was done.
It is true that the last person who ran the account might have been a cretin, but it’s not always the case. They may have had a sound basis for how they built and managed the account.
It's tempting to demolish and rebuild from scratch, but that's risky. A complete ground-up rebuild will cause a period of erratic performance. This has real-life consequences if the business behind the advertising depends on the leads.
And sometimes - for no obvious reason - your new campaigns won't perform as well as the old junk. An slower iterative rebuild will save you from trying to explain that your structure is better even though the lead count has plummeted.
Your instinct is going to be to make changes. I understand that, you’ve got to impress a boss or client. You want to do something - that’s what you’ve been hired for.
But, changes cause turbulence. Even changes that look good on paper can cause short-term performance drops.
You need to allow time after making a big change for the turbulence to settle and to accumulate data.
Some changes have low impact. Pruning dead wood, implementing a coherent naming structure or arranging negative keyword into shared lists won't rock the boat. You can do these kind of changes as fast as you like.
Other changes like fixing targeting mistakes, choosing a different bidding strategy or restructuring an account have a much bigger impact. You're in for a tough time if you stack a bunch of high impact changes on top of each other and the account tanks. Finding the root cause is almost impossible when it could be anything you've just done.
The speed at which you make changes should be guided by how much of an impact you expect, and how long it'll take to get enough data to confirm if you're on the right path.
Generally you’ll find that the 80/20 rule applies. 80% of your impressions come from 20% of your keywords. 80% of your clicks come from 20% of your ads. 80% of your conversions come from 20% of your search terms and so on. In some cases the ratio is skewed even more - 90/10, 95/5 or 99/1.
Start by finding the entities - keywords, ads etc - that have lead to conversions. I label them so they’re easy to find because you’ll build on them when it comes time to make changes.
Targeting mistakes are when your ads are shown to the wrong people:
The geographic report and user location report show where people are. The search terms report shows what words and phrases they used.
There are three other targeting dimensions worth looking at for mistakes. They are:
For instance, if you’ve spent a load of money showing birth control ads to 65+ males, and never had a conversion, it might be a targeting mistake.
If you paying for clicks at 3am but haven’t had a conversion in 10 years it might be a mistake.
If you’re getting wildly high cost per conversion on search partners, it might be a mistake to continue to advertise there.
Fixing targeting mistakes gives you fewer opportunities to show your ads. This may mean fewer impressions, clicks and conversions, but they should be better impressions, clicks and conversions.
We’ve already agreed that you’re going to be a hero when you get this account to take off. But before you put on your cape, remember that it is possible to suffocate a struggling campaign by mistake while you're giving it mouth-to-mouth.
I saw this when I consulted with a local client, Nathan (not his real name). Nathan runs a successful ecommerce store. He sells the products he makes here on our tiny island to customers from all over the country. (We live on the Isle of Wight at the bottom end of England.)
Nathan contacted me after sales from his Google Ads dried up. He built his own campaign and saw promising results. He hired a freelancer to improve his early efforts.
The freelancer scrapped Nathan’s work and rebuilt the account from scratch.
After two nail-biting weeks with no sales Nathan turned off the freelancer’s campaigns. He crossed his fingers and turned the campaign he'd built back on. The sales started flowing again.
Nathan asked me to audit the freelancer’s work and see what went wrong.
Her decisions were mostly sound. She’d restructured a mess of keywords into tighter ad groups. She’d written extra ads to split test the copy. She’d chosen a sensible bidding strategy, and a load of other changes. On the surface everything she'd done was right. This was textbook stuff.
If there was one thing she was guilty of, it was doing too much too soon. It looked like she’d read a list of the top 25 ways to optimise Google Ads and decided to do them all. At once.
But there was a bigger problem…
She hadn’t thought about how Nathan felt. She was so focussed on the technical aspects that she’d ignored the person behind the ads. You don’t get called in to take over an account when it's rocking. You're there because it isn't working. The person behind the ads - your boss or client - is unhappy, distrustful, anxious or plain not convinced that Google Ads works.
This is not the time for radio silence but that’s exactly what happened. She got her head down and started work.
She made an enormous number of changes in two weeks. But she didn’t let Nathan know what she was going to do, why she’d do it and what the likely outcome was. The only hint he had was an order book full of empty.
If she'd had a lighter touch and better human skills she'd have kept Nathan as a client forever.
The entities - keywords, ads, extensions etc - that clutter your account without bringing conversions are dead wood.
Look for entities with:
You'll have to decide what 'enough' means. How many impressions without a click, how many clicks without a conversion, before you pause.
I've seen many rules for deciding when to pause a keyword:
I'm not convinced that any of these approaches is founded on enough data to pass a rigorous statistical validity test. But, that probably doesn't matter. Pausing these entities is about as safe as anything you can do to a campaign.
If they've had a long history of being useless it's unlikely that pausing them will cause you to miss a load of cheap conversions. If you've made a horrible mistake you can always enable the keywords.
If you pause keywords that got clicked but didn't convert you'll save some money. Other keywords might absorb that budget. If that doesn’t happen your ad spend will be down - not a problem unless spending the budget every month is a priority.
I prefer to pause dead wood rather than remove it. I might circle back later and see if any of the dead wood can be brought back to life. I label dead wood before pausing it so it's easy to find again.
Verify that conversion tracking is a) working and b) tracking what you think it is.
It’s easy to get conversion tracking wrong. It's often in the hands of a third party - a web developer, landing page service or call tracking provider. Even if it's done in-house something as simple as updating a website theme can break it.
You must be able to trust your conversion data. The only way to do that is to verify that it's working as expected.
Does the number of enquiries you get match the number of conversions? More enquiries than conversions is OK. Some of those enquiries might have come from other traffic other sources. Fewer enquiries than conversions means that something is wrong.
It's especially easy to over measure if you're tracking a click-to-call button or link. Conversion tracking can only tell you the click happened. It can't tell you if the visitor made the call and spoke to someone.
It's somewhat easier, but not perfect, if you're using Google's call forwarding or a third party service.
If you're not confident that phone conversions are telling the truth it's better to exclude them from the total conversion count.
If you're using a live chat service that includes conversion tracking, check what they count as a conversion. Does it record a conversion if someone starts a chat, but there is no one at the office to respond?
Google defaults to counting every conversion. It’s probably fine for ecommerce but it over measures when the conversion is a lead. Conversion tracking should be set to count one lead as one conversion.
I have a client in tourism. Clicks cost an arm and a leg. Conversion rates are low across the industry. He hired me to nurture his in-house PPC team.
Our first meeting was brutal. I audited their conversion tracking to find out if we could trust the numbers.
They tracked enquiry forms, calls and live chat. Nobody on the PPC team could tell me if a lead ever used more than one of these ways to enquire. They called the CRM lady into our meeting and asked her.
She told us that about 30% of people in their database had used more than one way of getting in touch. They'd fill in the enquiry form and phone, or they'd fill in the form and then use the live chat. They were measuring each of these actions as a conversion. One lead could potentially be counted as three conversions!
Are you including intermediate - a.k.a. micro - conversions in your total conversions count?
Sometimes you might want to track a series steps towards a final goal. Something like tracking how many people click from a landing page to a contact page, then how many start filling in the form, and finally, how many submit the form. This makes sense if you're trying to find bottlenecks in a complex conversion process.
Check if these conversions are included in the total conversion count. If they are, is there still a good reason for their inclusion?
And finally, are any values assigned to conversion still valid?
Test the contact form and other conversion paths.
Fill in the enquiry form, click the call button and start a live chat.
Sending traffic to a site with a broken form is useless but it happens. I've tested click to call buttons that ring an old phone number. I've seen live chat systems that are dead. I’ve seen fantastic sales leads end up in the spam folder.
It’ll take a minute or two to check and you’ll look like you really know what you're doing if you find something that doesn't work.
I like to keep a list of opportunities I spot while working on the account. I’ll often find things like old text ads instead of expanded text ads, expanded text ads without a third headline and second description and ads without extensions.
I list them so I don't forget but I don't do anything about them right away because not all opportunities are worth pursuing. For instance, testing a third advert in an ad group that gets 100 impressions a year isn’t going to give meaningful results anytime soon.
The value versus complexity model can be used to prioritise opportunities.
Every opportunity is evaluated based on it’s business value and how complex it is to put in place. Opportunities are placed on a simple matrix. The one axis has business value and the other complexity.
Opportunities that have high business value and low complexity get first priority.
The middle groups - low value and low complexity and high value and high complexity are next.
Opportunities with low business value and high complexity are lowest on the list.