Need a process to go by..

8 replies
  • |
I've setup a campaign, and I have separated keywords into relevant ad groups. But what is the process to optimize the campaign?

Is the goal now to determine the highest CTR ad for each ad group? And once I've determined that I start split testing landing pages to determine the highest CTR?

  • Profile picture of the author gprialde
    What do you cover and what is it?
    {{ DiscussionBoard.errors[11042519].message }}
  • Profile picture of the author dburk
    Hi karmadog,

    Start With A Goal

    The process starts with setting a goal, or goals.

    To use an analogy, think of your campaign as a journey across the sea. As a captain of a sailing vessel you need to set a goal, in the form of a destination port that you would like to arrive at once you complete the next leg of your voyage.

    You need that goal to set your heading and to chart your course, without it you will likely end up lost at sea, not knowing where you are heading or when you are likely to reach it.

    In marketing you also need to have a goal, without that goal your campaign metrics have little meaning and the direction of adjustments can become ambivalent. You need a goal to measure your progress against and to chart a course to get there.

    Your goal needs to be a real value based goal. Don't use diagnostic metrics like CTR as a goal. Use something that has bankable value, like sales volume, growth rate, or profits. Those are goals that have real bankable value.

    Identify Key Metrics

    Your primary metrics will be those that indicate your progress towards a goal. Most other metrics are diagnostic and should not be considered objectives. Use diagnostic metrics to gain a better understanding of what is happening and don't target a diagnostic metric as a goal for optimization. Your goal should always be something that has bankable value.

    Once you have set a meaningful goal you can work backwards from there to set objectives for number of conversions, cost per acquisition, cost, conversion rates, Clicks, CTR, and impressions. Keep in mind these second level of metrics are diagnostic and not actual goals. You can seek to make changes that improve a metric, but it's only useful if it effects the primary objective of your value based goal.


    The goal of marketing data analysis is to gain an insight.

    Things like CTR help you understand how people are reacting to whatever changes you made, but what truly matters is why they behave the way they do, and how that insight can be applied to the next round of changes, all in an attempt to move toward your ultimate goal.

    Make changes that are designed to impact how your visitors behave. Seek to understand why people behave as they do and use those insights in your next round of adjustments.

    There are many things you can adjust to optimize your campaign, but you need to have firmly established goals to know which direction to make adjustments. Too, little, or too much is not optimal, you need to have those goals established to see when you have adjusted too far, or too little.

    Now, to the core of your question.

    You should be looking at CTRs, but not in a vacuum. There are many things that effect CTR, including your ad position. So by simply raising your bids will improve your ad rank score, which may improve your ad position, which in turn will improve CTR. However, it is possible that none of those things will move you closer to a goal based on real value. It might, or it might not, you need to be measuring the impact of bid adjustments on your value based goal, not on a diagnostic metric like CTR.

    Having said that, once you have isolated variables, like ad position, you will often see a correlation between CTR and ad relevance. So CTR can be quite useful in helping you ascertain which ad message your audience finds more appealing. Making adjustments to your marketing message is one of the areas that tends to have big payoffs in campaign performance.

    Test Value Propositions

    If you setup your campaign correctly, you should have written ads for at least 2 or more different value propositions and included at least one ad text variant for each value proposition within each ad group to measure the relative difference between those competing ad variants.

    Finding the optimal value proposition for your ads and landing pages are key to a successful campaign.

    Get Granular

    In addition to testing ad variations you can also make budget and bid adjustments. You need to make budget and bid adjustments on the most granular level that is practical. Not just at the campaign, or ad group level. You need to adjust down to the individual keyword, or target audience segments. And you need to make adjustments based on relative impact on achieving progress toward your primary goal.

    Here are a few things that you can adjust during the process of optimizing your campaigns:
    • Value Propositions
    • Headlines
    • CTAs
    • Max CPC bis Adjustments
    • Device Bid Adjustments
    • Location bid Adjustments
    • Hour Of Day Bid Adjustments
    • Day Of Week Bid Adjustments
    • And more...

    Optimize Search Terms Not Keywords

    Keywords are what we use to trigger ads, but it is the actual search terms that we need to identify, review, add to, or exclude from, the campaign so that we can optimize the performance of search terms. not merely the keyword that triggers a group of search terms.

    You should be generating a Search terms report on a frequent basis and analyzing that report to discover irrelevant terms to exclude and relevant terms to add to the campaign for more specific targeting.

    Be Strategic

    You cannot make adjustments until you have enough data to be actionable, and you might not have enough time, or data, to make every possible adjustment all at once, so be selective in what you focus on at first. Focus first and foremost on where you are spending the most and where you are likely to have the biggest impact on campaign performance. Save the smaller less important adjustments for later, after you have tackled the ones that will make the biggest impact.

    Make sure that your account has been structured to provide the data you need as quickly and efficiently as possible. Things like testing value propositions are so important that you cannot afford to put them off. Make sure that you have granular ad groups, and multiple ads, testing specifically researched value propositions. Focus on finding optimal value propositions right from the beginning, as there is no point in trying to optimize a poor performing value proposition, you won't get anywhere very fast that way. So start there, find the best value propositions and then get on with the other adjustments.

    This should help you get started in the right direction, there is really so much you have to learn, don't worry about knowing it all at once, just get started.


    Don Burk
    {{ DiscussionBoard.errors[11043380].message }}
    • Profile picture of the author karmadog
      How do I know whether it's the keywords, the ad or the LP that has to be tweaked? For instance, if my conversion rate is very low, I wouldn't want to remove that keyword, because it could easily be the ad, and/or the landing page.
      {{ DiscussionBoard.errors[11043698].message }}
  • Profile picture of the author dburk
    Hi karmadog,

    The answer is easy, test multiple ads, using multiple value propositions. A value proposition is copy that answers the question "why should I buy from your site?"

    At first, you might not be good at writing effective ads, so if needed, copy the value propositions from the top competitors' ads and include one or two of those in the ad group as a baseline for testing your own ads against.

    A big part of optimization begins before you even write the first ad, it starts with value proposition research. Study your own unique value propositions, study your target audience's comments, reviews and online discussions, and study your competitors ads and websites to identify the most compelling value propositions and utilize those in ad copy of at least 2 or more distinctly different ads, each using different selling points, for each of your ad groups.

    The idea is to split test the best candidates with traffic from the same source, using the same keywords/search terms, to see which distinctly different ad message your target audience finds the most compelling. Make the winner of each of your tests the new baseline and test any other good value proposition candidates until you are sure you have found the most effective value proposition (selling points) for you ads.

    The same principle applies to your landing page's sales copy. The intent of user that searched a particular search term must be aligned with the message of your ad, and the landing page needs to share that same consistent message to achieve the highest level of message match. When they are all synchronized you tend to get the maximum effectiveness.

    So keep your ad group keyword list tightly focused on a single intent, craft an ad text message that perfectly aligns with that intent, and make sure the same messages is carried through onto the landing page.

    Split test various value propositions to find the most effective, and always use the most specific and relevant keywords, avoid overly general or ambiguous search terms, they are almost never effective.

    Value Proposition: 3 steps for laying your value prop testing groundwork | MarketingExperiments Blog: Research-driven optimization, testing, and marketing ideas


    Don Burk
    {{ DiscussionBoard.errors[11043859].message }}
    • Profile picture of the author karmadog
      So, basically, I'm supposed to be testing the ad copy and the landing page together as one, as opposed to randomly testing ads with random landing pages?
      {{ DiscussionBoard.errors[11043869].message }}
  • Profile picture of the author karmadog
    Also, If my goal in my current campaign is to get a sale (no opt in with this campaign) I would optimize for conversion?

    But to get the cheapest clicks and highest quality score, wouldn't that require optimizing for ad CTR?

    So my question is, how can you optimize for both conversion and at the same time CTR, if your "goal" is a sale?
    {{ DiscussionBoard.errors[11043963].message }}
    • Profile picture of the author dburk
      Originally Posted by karmadog View Post

      So, basically, I'm supposed to be testing the ad copy and the landing page together as one, as opposed to randomly testing ads with random landing pages?
      Hi karmadog,


      You do not want to just sling mud against a wall to see what sticks. Each of your tests should be designed to answer a specific question that will help you improve your campaign results, nothing random about it.

      All of your marketing experiments should be by design, purposefully testing a hypothesis that is likely to improve your campaign performance, always by design, never random. You should attempt to control all variables so that you do not end up with random results.

      Think of your ad as an opening line of a conversation, and the landing page is your reply to a "please, go on, tell me more" response. Ideally, ads and landing pages should be carefully crafted to go together.

      To be clear, your tests are designed with a specific purpose in mind, the only thing that should be random is the random selection of visitors in your A/B split testing application.

      Originally Posted by karmadog View Post

      Also, If my goal in my current campaign is to get a sale (no opt in with this campaign) I would optimize for conversion?

      But to get the cheapest clicks and highest quality score, wouldn't that require optimizing for ad CTR?

      So my question is, how can you optimize for both conversion and at the same time CTR, if your "goal" is a sale?
      Pick one, just one.

      It all starts with a well defined goal. Hopefully your goal is not merely to get "a sale", because a singular sale seems to be a meager goal. It would be better to call that an objective for each ad impression, but wholly inadequate for a meaningful goal. I'm sure that you are hoping, and wishing for more than a singular sale.

      To be clear, the primary objective of a single ad impression is to get a sale, however the goal of a campaign should be more meaningful, time-bound, and specific (i.e. $90,000 in 30 days, 30% increase in sales in 90 days, $20,000+ in profits by May 31st., etc.)

      By setting a meaningful goal, that targets a specific bankable value, you will instantly remove the question you have about which metric to base your optimization efforts on. At some point you will realize that metrics like CTR are diagnostic in nature, they inform you of conditions, which is useful, but not your primary campaign objective.

      I realize that you are probably referring to the campaign setting that allows you to choose automatic optimization of ads within the ad group based on conversions, or clicks. It really depends upon you goal and the strategy you have chosen to reach your goal. If you have not yet set a meaningful, time-bound, specific and measurable goal, it is doubtful that you have selected a strategy for that goal.

      It starts with a goal. Every decision is driven by and measured against that goal. Without a destination, how can you know what course to chart? Start with a goal. And no, to "get a sale" is not a goal, at least not a meaningful one.

      So... tell me your goal, if it is a good one then I will answer your question about your optimization strategy. I suspect, that once you have set a meaningful goal selecting an optimization strategy will be easy.

      Since your question seems to imply that you are concerned with both achieving a sale, as well as controlling costs, perhaps you should set a goal based on maximizing profits? That way CTR and conversions become diagnostic metrics rather than primary objectives. There can be only one "primary objective" per campaign. So pick one.


      Don Burk
      {{ DiscussionBoard.errors[11044370].message }}
  • What DBurk said and research ads and landing pages of your direct competitors.

    That's called "lazy" method but it's time-proven and worth every penny:
    • Using competitors research tool, find all the keywords your competitor is displayed for in contextual advertising. It's a ready-to-use list of keywords for launching your PPC campaign. Or...
    • you can choose another tactic and exclude these keywords from your list to avoid competition.

    Also you can find what landing pages does your competitor align with PPC ads. When you export a list of ad keywords of your competitors, find URL column in the report:

    Use this to modify and tweak your PPC campaign and you landings and their CTAs.

    Those are proven methods as you'll be dealing only with verified keywords, ads' examples, and landings. Competitors already spending their ad budgets researching these so you'll be kinda like a parasite

    Informational keywords

    Riskier method if you don't focus on your content but works GREAT with our clients who are very meticulous about the quality of their content:
    • If you align your PPC ad with relevant informational page on your website (something like a blog article), it will work well for your brand awareness and may improve your conversion rate. By answering the questions of your users, you'll encourage them to purchase on your website.
    {{ DiscussionBoard.errors[11046656].message }}

Trending Topics