The impact of the dreaded C-bomb (Covid) is pretty clear. But what are the longer term impacts of a lockdown and tepid market conditions?
- Cost inflation perhaps
- maybe stifled consumer spending
- or more competition as off’ goes online
To get a handle on sentiment, we probed Marketing Managers from start-ups to established companies to share with us their biggest PPC headaches and concerns.
The top 5 issues were:
- How to stop your costs going up and ROI going down
- How to increase your conversion rate
- What to do about poor quality traffic
- How to boost performance of low performing products
- What to do when the weather affects your performance
🔥 Cost going up – 🥶 ROI going down
One thing that cropped up is that, “AdWords was consistently profitable for us 2 years ago, but now it’s getting more expensive and we want to know what to do about it”.
Because of all of the external factors that can affect performance in your Google Ads accounts – what worked last year won’t necessarily work this year.
If Google increases CPCs in your industry, or more competitors decide to compete, you cannot stop that (unless you are lastminute.com).
The key to improving your ROI is to firstly cut wasted spend, and then to maximise on strong intent.
Lets’ go:
1. Cut wasted spend with a close variant culling
How do I say this politely… close variant performance is a mixed bag.
Not great, not necessarily bad…
But, they can be dire.
They will often perform worse than searches that actually match with your keywords. This is particularly true for [exact match] close variants.
The cynic in me suggests close variant matching for exact match was only introduced to increase Google’s revenue.
Fortunately, there’s a close variant matching script that negatives close variants from your campaigns.
You will still be able to see in search term reports which close variants you matched, so you can make the decision to add them as keywords in their own right should you wish.
Use the script to preserve your exacts, let your phrase and broad match catch the rest.
In doing so you keep a cleaner view of how your exact terms are really performing.
Goodbye Phrase, hello BMM.
Match types are also another area where accounts are typically wasting money without realising.
In my experience, phrase match should be avoided in all but the highest of spend and volume accounts.
Compared to broad modified, Phrase consistently results in higher CPCs and works to lower ROI.
By avoiding Phrase match you also simplify your account.
Even in circumstances where phrase match has been used over BMM because the advertiser is trying to preserve word order, this can be mitigated by clever use of negatives.
For example: “london to paris flights” has a distinctly different meaning to “paris to london flights”.
Phrase match would allow you to preserve this order and the different meaning, whereas with BMM word order does not matter.
Negative use case: if you wanted to appear for “paris to london flights” but not “london to paris flights”, you can add the latter as a phrase negative. You’d add +paris +to +london +flights as your BMM keyword.
It takes a little more work and care, and foresight, but negative phrases can clean-up any mishaps with your broads.
Pause low quality score keywords.
Low Quality Score keywords are often lurking in accounts and driving up costs, lowering ROI. As you can see from the chart below, a Quality Score of 4 is 75% more expensive for your CPCs than a score of 7.
If shaving cost is your priority, then looking for quality keywords should be a priority.
You can analyse each component of Quality Score to determine exactly how you can fix your lower score keywords:
- ad relevance
- landing page experience
- and expected CTR
Ad relevance and expected CTR.
Ad relevence is why many advocate so strongly for single keyword ad groups aka SKAGs.
Having one keyword in each ad group means that every ad for that keyword can be tailored to the keyword exactly.
You can have the exact keyword in the first headline, meaning your ad relevance goes up.
This improves your expected CTR and ultimately your quality score.
Easy.
Creating single keyword ad groups may seem daunting, especially in an account with thousands of keywords.
That’s why you don’t do that.
You cherry-pick your top spending keywords, perhaps your top 10, 20, 50 or whatever. Focus on those.
Want an easy life? Use my automated SKAG script.
SKAG’ing an entire account is a fools errand.
You slice data thin, you create a gazillion ads, hurt the algorithms and you create low value work in maintaining the mess.
If someone advocates SKAG for an entire account they’ve either got automation doing it… or they’ve been smoking too many pancakes.
Just keep it simple and focus on the important areas.
Your landing page experience score.
Landing page experience is commonly believed to be a reflection of site speed.
You can check how Google rates your speed here. With more traffic moving to mobile every year, upgrading your user experience on both mobile and desktop is not optional.
But that shouldn’t be news to you.
Even big budget websites can get the little things wrong, such as uploading jpeg and png images that haven’t been appropriately compressed or optimised.
Want to squeeze every last kb out of your images? Use tinypng.
Other factors in landing page experience include the relevance of the landing page to the query, and how easily navigable it is.
This includes an appropriately sized font for the screen width, no more 12px content – please.
If you’re using an off the shelf solution for your website, such as WordPress, there are plenty of free and paid caching options.
Don’t skimp on web hosting either. Speed matters and saving a few pennies is a false economy.
Focus on indicators of strong intent.
The graph above shows the amount of possible daily data points in 10,000 keywords over time.
Let that sink in.
Now breathe…
Clearly, this is getting harder and harder to manage manually.
The most effective and efficient way to do this is either through a bidding script, or using smart bidding strategies. This includes ‘Enhanced’ CPC too.
Manual bidding is so 2019.
Caveat: I’d still recommend using scripts on lower spend/volume accounts and brand campaigns. If you ask smart bidding to jump through small hoops (low budget and/or low conversion volume), it will stumble, trip and headbutt the hoop.
The hoop being your budget.
Why? Because automated bidding rely on Google’s machine learning algorithms which are great good at making smart decisions when they’ve got enough data.
We’ve actually created an automated bidding script for manual bid campaigns, follow the link to grab the code and install it.
There’s also an instructional video.
By now you’ve probably tried or are using some form of smart bidding. Our own data at AdEvolver shows that less than a third of all campaigns remain on manual bidding.
*sad manual bidding noises*
Anyway:
Simple smart bidding rules to follow + avoid failure.
- Get some data – a dozen or so conversions is just about enough, the recommended is 30 in a month.
- Start with either target CPA (tCPA) or target ROAS (tROAS).
- I wouldn’t use tCPA on campaigns that have revenue reporting.
- To start set a realistic target, one that is at or below your ideal ROAS number, at or above ideal CPA.
- Don’t try and ‘cheap’ the system with unreasonably low tCPA or high tROAS.
- For CPA, ensure your campaign budget isn’t smaller than your target (where possible).
- For ROAS, start at just above break-even and work up, you’ll get more sales and data on which to optimise.
- To be safe only increase/decrease smart bidding targets by 5-10% and after at least one additional conversion has accrued, or a week has passed.
That’s about the long and short of it, but here are some more things to consider.
Do you need some data to use smart bidding, or can you use it on a brand new campaign?
Pro-tip: Google advocates you hit a minimum of 30 conversions in a month i.e. an average of 1 conversion per day, before jumping on the smart bidding bandwagon.
However, the impatient can use smart bidding with no data. Things will get a little choppy but it generally works.
Expect peaks and troughs much greater than normal.
Pro-tip: If you take this approach, just be sure to set strict budgets and watch them like a hawk. Or use Budget Groups (an AdEvolver™ feature) to automatically protect your campaigns and ad spend:
Product placement over…
Optionally: you can move on to maximise conversions for campaigns that have tCPA performance in the bag, and now you want to stretch your legs going for volume not efficiency.
Use maximise revenue on strong shopping campaigns that have proven their value, not before you’ve given tROAS a chance to perform.
But expect CPCs to hit the roof on some queries. If you have a lot of budget overhead, consider lowering it first otherwise Google will see a big green, money print light turn on!
CPA vs ROAS bidding?
tROAS takes into account the revenue generated by a conversion, whereas tCPA does not.
For example, tROAS appreciates that it’s worth spending more to acquire a customer that will spend more, but tCPA is suitable for accounts where leads are considered equal.
The learning period for either bid strategy can be be up to 4-6 weeks, so expect some volatility to start with.
However, on anything but smaller accounts, smart bidding should come good in the first week.
You’re generally safe to make incremental changes weekly where required.
Debunking the BIGGEST smart bidding myth.
Changing bids and budgets ‘resets’ the smart bidding, right?
Erm… no.
Caveat: I think the Google UI is to blame for this perception, because – to save face – Google do highlight a new learning period after you make changes.
That might be where the “resetting learning” myth came from, but it doesn’t work like that.
At least, not in the way many people describe.
Machine learning is essentially a prediction engine, it knows what it knows.
If you change a parameter like bids or budget it doesn’t just forget what came before and reset everything.
For Google Ads, the query level impressions are the foundation.
- You’re bidding against a keyword.
- The machine is matching a query and collecting an impression.
- For each time a (query triggered) auction runs.
The ML is predicting how many impressions it can afford to hit a target. Your target. The more data, in theory, the better it is at this.
It gets more complex than that of course, because this is at a query level and you have to consider device, audience, geo etc.
But let the ML worry about that for now.
Where people fall fowl is they make BIG changes which means it recalculates past learnings or has to find new/more data to create a reliable prediction going forwards.
This further fuels the ‘resetting learning’ theory.
An example – you’re tCPA is 50 and you change it to 60.
The ML knows how to get conversions at 50, it also knows which areas of the account/campaign/ad group isn’t able to achieve that.
It will first try and exhaust (impressions) the areas that get tCPA at 50, because it’s highly probable they will work at 60, and then move on to previously weaker areas.
The same is true conversely; exhaust the known good before tackling anything else.
A BIG change makes the known good, turn to… crud.
To reiterate, when changing automated bidding, keep it to 5-10% increments. 15-20% is the very maximum and more risky, especially if you’re closer to impression limit.
Rant over.
Next:
3. How to increase your conversion rate
Maximise strong intent and make sure that you are using all the available signals, in a sensible way, to be present in front of the right people.
Reducing wasted spend will give you a higher conversion rate on paper, but really that’s a false increase – all you’ll be doing is getting rid of people who were never interested.
Do that, of course, but also you want to do is make the people who were on the edge of conversion cross that line.
Improve your ads.
That can be either through urgency, personalisation, upfront pre-qualification or landing page testing.
Making your ads more urgent in this case means placing limited time offers right in your ad copy. One very effective way of doing this is to use countdown ads – a form of ad customiser.
If you are running offers only on specific items, consider using ad customisers pulling from a business feed.
You can then easily change the offer for one set of items, without having to manually change hundreds of ads.
You can find more information on how set up ad customisers here.
Personalisation can also be a very powerful tool.
Again this can be achieved through customisers, and means you can include very tailored messaging in the ad copy.
For example, “50% off delivery within Manchester” or “25% off for returning customers”.
Both sets of this information you can include using customisers, for both location and audience membership.
Pre-qualification is an important step to take in ads.
Especially if you have lower limits for customers to be eligible for your products.
It might feel odd putting anything in your ad copy that could deter customers, but avoiding pre-qualifying traffic in order to keep visitors up, is a false economy.
If you have minimum requirements, it’s not like up-selling. Be honest about them and deter those wasted clicks.
Don’t waste your ad spend on people who do not fit your criteria.
The wrong (or right) landing page selection.
Consider choosing less traditional landing pages in order to improve your conversion rate.
You might be surprised about which landing pages actually lead to the best conversion rate.
Homepages for example, despite requiring the longest user journey from visit to conversion, tend to tell more of a brand story.
That can be a powerful influence.
Pro-tip: try ranking all the possible landing pages on your site from the least to the most specific (so for example homepage, category page, sub-category page, then product page).
That should give you a list of what to test against your assumptions.
For example, you could test the least specific (i.e. the homepage in this example) against the most specific (product landing page).
Depending on the outcome, you can move the test further towards one end of the scale until you find the best converting page.
On-site improvements cannot be overlooked.
Sometimes in order to increase conversion rates, you have to strip and repaint the walls.
If resource is more limited, you can take some fairly common sense steps to improve your user experience:
- Showcase available inventory, not sold out items
- Have a prominent CTA
- Highlight your USPs up front
- Include ratings, reviews, and testimonials
Pro-tip: if you’re in a competitive market, visit your competitor landing pages and take a note of what they’re not saying.
That’s your opportunity to stand out by being different.
3. What to do about poor quality traffic
Reinvigorate low quality traffic by creating RLSA audiences of unprofitable visitors.
The most common way to use RLSA audiences is to bid up on recent visitors, or visitors who have shown an interest in your products.
However, you can use RLSAs to identify people who are not interested.
Pro-tip: if someone has visited several times and not converted, you may decide that you no longer want to pay for their traffic. If they want to convert, let them come direct or via organic channels.
Caveat: if you are using a last-click attribution model, then you may see a negative impact on the volume of your conversions by doing this.
Organic and paid search should always be viewed holistically (fancy word for ‘together’ or as a whole, one supports t’other).
Pitting the two channels against each other will often result in one cannibalising the other, and you ultimately paying more per sale.
Secondly, run an analysis on customer visits per conversion before deciding on a cut-off number of visits for blacklisting someone.
The search query report is still valuable.
Due to quarterly revenue being down ‘privacy’ concerns Google recently hid a chunk of data in the search term report; even so, you need to make sure you are not consistently appearing for irrelevant searches.
It can be easy to overlook long term trends in searches if you only look at search terms on a weekly or bi-weekly basis.
I suggest analysing the last 3-6 months’ worth of data in your account. Rank your search terms from highest to lowest, in terms of cost, impressions, CPA, and conversions in turn.
Cost and cost/conversion or conv.value/cost are obvious enough, but you also want to keep an eye on high impression, low click-through terms.
Those queries drag your quality score down and your cost will go up.
That will give you a thorough understanding of not only what is costing you a lot, but also which searches you are appearing for often with little intention of conversion.
Anything you do not want to appear for, you can then add as negatives.
Pro-tip: using phrase negatives, even if it is just for one word, means you can negate any searches including that word or phrase – very handy for not having to add hundreds of combinations of negatives.
If the search terms you are pulling in that are irrelevant are from close match variants, you can use this script to negative those from your account.
Auction analysis is easy.
Often overlooked, auction insights allow you to see who you’re competing against in the auction.
Furthermore, they will show you if you’re bidding against unlikely competitors or advertisers that are simply not in your market.
A quick Google search of the keyword and related terms will highlight any problems.
Pro-tip: check the organic results too, if they don’t look appropriate then you have a problem – Google thinks differently about your keyword than you do!
If this is the case, you need to rethink and rework your keyword selection.
Go a step further and run an n-gram analysis.
Ngrams are search queries on steroids. The determine which combinations of words do not perform well for you.
Unfortunately, you’ll have to wait a few weeks for me to finish our ngram tool.
Boo.
Just sign-up to this site and you’ll be on the list for when we launch the feature.
Double check ad copy accuracy.
Make sure you are not running sales ads for a sale that doesn’t exist, or that any offers you are advertising are still valid.
This sounds basic – but in large accounts this is surprisingly easy to miss.
For future ref: Labels are your friend.
If you are running thousands of ads across products and/or territories for example, it is very possible for automated rules in Google Ads to have not run correctly, or for some human error to have occurred.
Pro-tip: look at ad groups where you are seeing a particularly good CTR but poor conversion rate – that would suggest there is something in your ads that isn’t being realised once they land on your site.
Alternatively, you can download your account into Google Ads Editor.
Filter for enabled ads, and then search for text relating to sales, discounts, or whatever your latest offer was.
Exclude app traffic from GDN.
Unless you have a specific reason not to, then excluding app traffic is usually an easy win.
App traffic will often have a disproportionately high CTR, which is not reflected in the conversion rate.
The reason is that it is very easy to accidentally click on app ads. Many apps are intentionally designed to trick fingers into clicking ads – because, that’s how the ‘free’ apps make money.
Worse still – if you are running remarketing, you can end up repeatedly chasing those random ‘finger visitors’ who really had no interest in the first place.
You can remove all app traffic in Google Ads Editor, by selecting “All apps” from the negative mobile app categories menu.
Caveat: many apps will slip through the net.
Why? Because until they are categorised by Google, they’re not going to get excluded.
Do no evil Google, do no evil. 📉😆
4. How to boost performance of lower performing products
Retailer or managing retail ads?
Then you know it can be tricky to marry up your ecommerce priorities with your digital marketing activity.
Here are some steps you can take with your Google Ads account:
1. If you have specific keywords for those products, check the impression share for the keywords.
If they are losing more than 5-10% to rank, then it would be worth bidding up on them to increase traffic initially.
If they are losing impression share to budget, but you don’t want to increase the spend on the campaign they are in as a whole, move them into their own campaign, and allocate appropriate budget there.
2. If impression share isn’t the issue, test different landing pages.
If it is a subsection of products you are struggling with, try sending searches directly to the product page for the most popular product within that subsection.
Or, try sending other related keywords to the page for your less popular products.
Bear in mind, it may just be that those products are not as popular, so monitor this carefully. You could jeopardise the performance of your account as a whole if conversion rates drop.
If you tend to run last minute offers on products to boost their performance, then you should use ad customisers.
They run on a feed enabling you to change the offers quickly and easily, without having to change hundreds of ads.
You can find a guide on how to set ad customisers up here.
5. Is performance under the weather?
What do you mean… it’s raining outside?
This is Great Britain – we don’t get rain!
*intense stare at summer flip-flops*
Well, some slippers might be nice right about now…
Determine whether the the weather is a problem by correlating whatever your main KPI is with the weather over the last year – the more data the better.
It’s a bit of a manual job, but you should be able to find a regional chart of weather patterns for your business.
1. Blame the weatherman.
If you’re sure that weather is an issue, then consider testing a weather bidding script.
This script from the Google Developers allows you to set the temperature boundaries for each weather type (i.e. what is “hot”, what is “cold”), as well as set the bid modifier you want to use.
This means your account will automatically react to unexpected changes in weather that could impact your results.
One issue you will face when trying to set up weather bidding, is that it is extremely hard to find downloadable historic weather data on which to base your analysis in the first place.
There are some providers which you can pay a fee to receive data (like meteoblue.com), but the majority of sources require you to be able to use their API, like openweathermap.org.
2. Weather or different signals to use.
Is it really the weather, or are there other more pertinent signals to use?
For example, a t-shirt store may see interest peak when there is a bout of hot weather.
However, increased traffic from more impressions and a better CTR will mean budget is depleted quicker.
The potential result is lost impression share and revenue potential.
In this case, it isn’t actually necessary to increase budgets depending on the weather – not bids.
You don’t need a fancy script for that. You merely increase budgets when impressions and CTR increase, regardless of why that may be.
Google Ads rules can be used to adjust budgets on specific campaigns if there is a spike in interest, to ensure that a minimal amount of search intent is being lost.
Consider those Google Ad Performance issues, squished
We’ve covered some decent ground here, I hope you’ll agree.
- How to stop your costs going up and your ROI going down
- How to increase your conversion rate
- What to do about poor quality traffic
- How to boost performance of lower performing products
- What to do when the weather affects your performance
Nice.
