PPC Ads Series

Post modified on

This is the first post from the PPC Ads series. You are about to learn my approach to PPC ads. How you can create them automatically via Excel or PPC Bee, how to effectively test multivariant ad copies against each other and the best way to automate the analysis. All with free tools included.

Why Should I Click on Your Ad?

The ad creation is based on keywords and search terms. Firstly, find out what are the customers’ needs and how they search.

Stand out from the crowd

What is the unique selling proposition (USP) – the main reason your customers use your services? That is the question you should ask. The USP should be reflected in each of your ads. But it needs to be unique. Have a look at your competitors’ ads in SERP (search engine result page).

TIP: Check the auction insight first. If there are any competitors that are frequently shown with your ad – these are the competitors you want to differentiate from. Read more in Brad Geddes’s article.

The most important pros and cons of your products or services are listed at price comparison sites or review pages. I prefer to use the internal data from the companies’ feedback questionnaires, which contains many individual reviews. To speed up your work, you can read the instructions at the end of this post.

Search term/ Search intent

“Every search is a story. Find it.”

Chris Nkurunziza

The ad copywriting is based on the keyword selection. You should already know what are the customers’ needs (what is the intent behind the search term) and ideally, show in the ad how you can help them. Understand the story.

Let me give you an example from the bookstore world. People searching for the authors usually want to buy the book even when they do not use the book title in the search term. I tested adding the title of the authors’ bestseller.

Ad template “original” contained just the author’s name and the client’s USPs. The second version “bestseller” contained also the title of the bestselling book of the author. As you can see bellow by the CTR and the volume impressions – the bestseller version was better (even though the search terms did not contain the title at all.

Improvement of CTR and winning more ad auctions
desktop and Google.com segment only

This is usually the same case with categories. Include the best selling product from a category and people are usually more likely to click and purchase.

‘Usually’ does not mean ‘always’. Test it yourself.

Please, be aware that common recommendations or suggestions may not work in all cases. For example, many Google case studies show that using price, discounts or countdowns in your ads works better. But from most of my analysis in 2018, the ad template with price included always lost.  For example, analysis of the Philippines fashion market promoting the lowest price on the market didn’t bring the attention. Out of 5 different ad variants, it had the worse CTR.

The predicted winner had the worst performance in CTR and won the least ad auctions


Don’t be afraid of using emotions. Your ad doesn’t need to be a list of benefits. You would be surprised how well perform ads spiced up with emotions – a sentence that could be used by any of my competitors. Think of using positive emotions – like an appraisal, using the association group.

Example – Courses that change your High schooler into a college student.

Or using the negative emotions – fear, anger or disgust

Example – 15% of students fail their entrance exams. Are you ready? Check more examples by Helen Edwards in the old WordStream article to get more inspiration. But don’t forget that you should still keep within the brand scope.

Goals of Your Ads

Questions you need to ask yourself before running the ad copies test:

  • What is the objective of this experiment and which metrics will be the core to look at?
  • How many ad templates will I test simultaneously and in one or more ad groups?
  • What is going to be the ad rotation set-up?


The ad copies experiment is just another marketing test that you do. So treat it similarly. Describe what is the test about, what is your hypothesis, which metrics are the core for analysis and timing + results.

Keep track of your learnings

I am trying to write down every meaningful idea I would love to test and also all experiments I do with the results.

The most common objectives regarding ad testing:

  • Long-term testing to get better ROAS (from continuous ad testing)
  • Low quality score and I need quickly increase CTR
  • Which selling point resonates the most
  • Create the best brand campaigns – so the CPCs are decreasing
  • Picking the best ad template that could be used for SEO metadata
  • Which ad template brings the most relevant traffic (B2B case)

If you want to get a more general overview of what matters for people who are already searching for your products, you more ad templates. The ad templates could be very different from each other. When you define the hypothesis and your ad copy is clear. You can get some interesting market data. Such as

  • Does matter having a location in the ad?
  • Does work having there shipping fee (shipping from X €) – or does it discourage the users
  • Using the benefits I am the strongest or use the benefits and I have the same with my competitors – which resonates more?
  • Should I stress quality or quantity?
  • Which CTA should I use? Are the users already ready to buy? (using the buy now) or do they need still consider the options (learn more)?
  • How strong is my brand – should I have it in the ad or not?

“If you are using the wrong success metrics in your test then your time and efforts will be wasted.”

Brad Geddes


Different objective means different metrics to check. However, the best is to analyze the ad results among many metrics. Most of the time it is CTR, the volume of impressions and conversion rate. Look at the pros and cons of the metrics:

CTR – Click-through Rate


  • The volume of clicks is always more than the conversions, therefore it is the fastest way to get a reasonable volume for test analysis.
  • It has the biggest impact on Quality score and therefore on Ad Rank as well, which usually correlates to more successful ad auctions and lower CPC.


  • At the first sight, it might be clear which ad copy is better. However, it is hard to pick the better ad because there are many traps you might fall in. CTR doesn’t only reflect the ad copy itself, but also the ad position, network, and used device.
  • The better CTR is, the better Ad Rank is. The better Ad Rank, the more ad auctions you win and you get an impression. However, you can win ad auctions for less relevant search terms (since the search intent is less relevant, the CTR gets lower) So in the end, the better ad could have lower CTR when you aggregate all the search terms. I’m explaining this issue in my last post from this series.
  • Better CTR doesn’t mean the same conversion rate. Better CTR just means more clicks. But you might just buy more clicks from people who are not interested in your product – you should be aware of it especially in the B2B segment (check this article to learn more)

The Volume of Impressions


  • The more impressions = more successful ad auctions


  • Any new ad has a disadvantage compared to an old ad. Old means another ad that has been already in the ad group for some time. It has “established” some quality score and Google can expect the CTR. The new ad is in a worse position. It needs to prove that it has higher CTR but at the beginning, it might lose some ad auctions because Google doesn’t trust it yet.
  • This metrics else doesn’t say anything about the quality of the traffic.

Conversion Rate


  • It finally reveals the quality of the traffic and business importance


  • It doesn’t tell you much about CPC and CTR (which might cause a drop in a quality score)
  • Conversion rate influences way more factors than CTR.
  • To overcome this con you need to have a lot of data and it could take some time till you can really tell which ad copy is the best based on business metrics.

Other metrics

Bounce rate – it often reveals “fake” winners in CTR. The easiest check of the traffic quality.

CPC – it is good to check CPC while you are deciding based on CR.

Quality score – It would be great to see the quality score of your ad templates. But for that you would need to have ACE (AdWords Campaign Experiments). I have never used ACE just for ad testing. It is way too time-consuming when you can check the CTR in a segmented way and estimate the impact on QS anyway. So I just compare the quality score before and after I change the ads.

Simple or multi ad group testing?

Multi-Ad Group testing

You aggregate data from all ad groups and you can find the results faster than test each ad group individually. Plus, it is the most common way to test ads. The main obstacle here is to pick only the “homogenous” ad groups. Don’t compare apples and oranges.

Just specific categories

“Running shoes” and “Adidas running shoes” might sound similar, but that you would be surprised how different are “just” categories compared to categories + brand. It’s not only in the conversion rate but selling points used in ad copy differ as well.

Just product with similar USPs

Discounted products require specific ad templates then the non-discounted. The same applies to products in stocks on in preorder. Therefore I recommend separating the specific products in the PPC Bee ad generator.  

Single Ad Group testing

This is way easier. But you need to wait longer. This strategy pays off only for the most important ad groups that generate the biggest volume.

“5% of Ads Account for 85% of Impressions”

Larry Kim 2018

It is most common for the brand campaign (promoting your company brand). You spend a lot of money and the conversion rate is the highest. Why not boost it with better ads? You might get lower CPC or just use some nudge to increase the conversion rate.

How many ad templates should I use?

It depends on how different they are. The more different the ad templates the better. So if you have a chance to change the H2 do it there. If you just change the description it will take a while since you get enough data so it is better to use fewer ad variants in this case.

How to track the ad templates?

The best way is to use labels. However, tools like PPC Bee won’t let you add a label to the ad. But it allows you to have URL parameter appended to the URL www.exmaple.com/product&ad-template

Watch out if you have auto-tagging enabled. Because this solution might ruin your GA data if you are using a manual tagging set-up. Some people use manual tagging for specific campaigns. The set-up is in GA.

Advanced options with ad testing

Ad customizers

When I need to add price, discount, number of products, etc. the best way is to use Ad customizers.


The COUNTDOWN, is a must-have for short period promotions. It is also great for events – like courses, conferences etc. You can even use COUNTDOWN within IF function.

The IF function is great when you are serious about your “audience management”. Do you have different target groups with different needs but using the same search queries? Do you have a huge user base and you want to show them different ad copy but you don’t want to duplicate the campaigns.

There are some examples:


  • Ad  1: {=IF(device=mobile, Create your bank account without leaving your phone): Bank account with the best interest rate!}
  • Ad  2: Bank account with the best interest rate!}

Take both ads – segment device mobile and compare the performance of Ad 1 and Ad 2.

When it works, you might test 2 different ad copies for mobile devices

  • Ad 1: {=IF(device=mobile, Create your bank account without leaving your phone): Bank account with the best interest rate!}
  • Ad 2: {=IF(device=mobile, Check the new internet banking app): Bank account with the best interest rate!}

Or test just the desktop versions

  • Ad 1: {=IF(device=mobile, Check the new internet banking app): Bank account with the best interest rate}
  • Ad 2: {=IF(device=mobile, Check the new internet banking app): The Bank of the year 2019}


Testing different ads for different audiences is more complicated. Let’s stay with the bank example, you want to have different ad based on the audience – current users vs. users of another bank

  • Ad 1: {=IF(audience IN (Current Users),Lovely to see you again.):Change your bank and get better interest rate!}
  • Ad 2: Change your bank and get a better interest rate!

However, in the Google Ads is nothing like “audience” segmentation. So you can’t use this method of having 2 ads. What can You do? Have only 1 ad with the audience IF function and compare the performance of the specific audience before and after the ad change. Try to avoid any major changes in the campaigns.

Or you theoretically can test it simultaneously (to avoid external factors) with ACE (AdWords Campaign Experiments).

The good thing is that you can group the performance among selected ad groups already in the Google Ads UI – just select Segment > Group rows by Audience

DKI – Dynamic keyword insertion

Watch out, some PPC specialists believe that DKI adds the search term into the ad, however, it is your predefined keyword that you use in the campaign. So remember DKI means “dynamic keyword insertion” not “dynamic query insertion.”

  • I personally do not use DKI anymore. I have either generated campaigns and ads exactly as I need for each product (either using PPC Bee or Power Query) or just the DSA campaigns.
  • A usual mistake in DKI is that the keywords are often longer than the left space in the ads and therefore the keywords in the ads are not used at all.

10 rules for PPC ad copies:

  1. Do not use exclamation marks or dots in the headlines. Just don’t.
  2. When you want to use some delimiter use vertical bar (|)
  3. Do not write the headlines (1,2,3) as one sentence (or even worse – do not split any words into 2 headlines). Bear in mind that the third headline is featured only in some cases.
  4. Do not use all caps. Google hasn’t specified how many letters in a row you could write in capitals until the system disapproves your ads. I would guess only a maximum of 7 letters in capitals.
  5. Watch out for declension (changing the wordform) in ad customizers. This is not really a problem with English, but in other languages the declension changes a lot and the ad looks too artificial.
  6. Set up 2-3 ad copies already at the beginning when you are creating the ad groups
  7. Do not forget the visible path in the URL
  8. Do not use words like “click here” that directly call the user to click on the ad. Google’s algorithm doesn’t like it.
  9. Your ads should attract only relevant users. So try to even discourage the irrelevant users from click (this matters for B2B advertisers in search terms which might be relevant for both B2C and B2B clients)
  10. Use only the most relevant URL – if you miss the landing page, ask for it your SEO team and do not make URL from your internal search

Now you know what to use in your ads. Let’s create them – The easiest is to use PPC Bee which will be described in the following post.

+ There is the promised bonus

Bonus: How to make unigrams, bigrams and trigrams in Open Refine

You can split the sentences into groups of one, two or three words (unigram, bigram, trigram) and check the most frequent groups. Later use the same wording in your ads.

Example of unigrams, bigrams and trigrams (source: Wikipedia)

The unigram is already in Open Refine under word facet. But how to create bigrams or trigrams?

The first step is to change capital letters to lowercase: Edit cells > Common transforms > To lowercase

From the reviews column create bigrams:

forEachIndex(value.split(" "), i,v,value.split(" ")[i-1] + " " + v).join("¤")

or trigrams forEachIndex(value.split(" "), i, v, value.split(" ")[i-2] + " " + value.split(" ")[i-1] + " " + v).join("¤")

Split the result by delimiter ¤

You got it. You can also cluster the most similar words by using Cluster and edit function

And create text facet to have it sorted 🙂

If you want to learn how to make the most of Open Refine – I must recommend a workshop from Filip Podstavec, founder of Marketing Miner.

Recommended Posts