Live experiment: How to lower your Facebook advertising cost in 2017

  • Reinart Bacalso

    Hey Jason,

    You can actually try out using PPE campaigns to run Ads that are made from the Web Clicks objective by taking the post ID of the web click ad and using it in the PPE campaign. 🙂

    I’ve done tests before, although mostly with local audiences, PPE performs better than Web Clicks Objective in getting link clicks when targeting smaller audiences with smaller budgets. (I used the Web Click ad format on the PPE campaign, not a link in the copy/caption. That doesn’t perform too well like your tests show.)

    Hope that info helps! 🙂

    • Hi Reinart,

      Thank you for the reminder – I completely forgot about that option 🙂

      I’ll spend $5/day for the next 5 days on this idea and update the results here. Stay tuned.


        • Reinart Bacalso

          Awesome! Looking forward to see how it works on this market 🙂

          • Reinart Bacalso

            Cool! that’s interesting indeed 🙂 are you gonna keep it running?

          • Yes, for 10 days 🙂 I set aside $50 for it so I guess we’ll get some data.

          • Hey Reinart, I just updated the post with the results – check it out and let me know what you think.

          • Reinart Bacalso

            Hey Jason,

            Awesome update! Happy to hear the engagement campaign is bringing in cheap clicks. Congrats too on getting that huge reduction on CPC. Excited to see where this is going!

            Interesting to see that you got more leads from the traffic campaign though. I think there’s more “serious” people found by that campaign, and the PPE campaign finds “casual” clickers/people. Like people who click a lot around FB and are just curious, but don’t really do much other than click then go back.

            Maybe an interesting test to conduct in the future is to use different UTM tags on the same web page/URL and run traffic to it using PPE and Web Clicks then compare bounce rate, average time on page, and other retention metrics from both Campaign Objectives.

            What do you think?

          • Hey Reinart,

            Good intentions, I think.

            I have previously done some tests using UTM but to be frank, they never come across as accurate unless you make some architectural investments – like getting SSL certified and whatnot. So not ideal at this stage.
            Besides, I don’t think bounce rate or average time on site are valuable indicators in evaluating the success of an ad or a piece of content. We need to find better metric pairs that will make more meaning for us.

          • Reinart Bacalso

            Cool! I’ve actually never done UTM tagging and GA analytics that much so I was just throwing off suggestions. Haha! Sounds good in theory, but your points are very valid.

            What metrics would you look at in GA as a valuable indicator for quality of audience?

          • Hi Reinart,

            To be very frank with you, none of my clients and I currently look into Google Analytics 🙂
            I know they have some useful reports there but our mainstay has always been a combination of Facebook reporting, Kissmetrics (if there is a need for 3rd party software), and maybe a custom built dashboard that brings everything together in one place, accounting for spend in other channels too.
            When we run paid campaigns, it’s always a product of 2 things: content and audience. Hence we sometimes need multiple steps to evaluate the quality of an audience, say – no. of repeated visits in the last 30 days, the CTR on a retargeting campaign, etc.
            If you have a concrete case study or client that you’d like help with, let me know and I can provide some suggestions.

          • Reinart Bacalso

            Hey Jason,

            Interesting! What would you say about a Campaign that does over 5.00% on LINK CTR, around $0.10 in View Contents, but gets almost ZERO add to carts. Like even if we get $0.10 per View Content, we get Add to Carts at $30+ which is waaaay over the KPI (products are around $30-$70 on average).

            Is it a low quality or bad audience, or bad landing/product page? This is an ecommerce on Shopify client.

            I’ve tried multiple audiences, and the results hover around the same place. USA audience.

          • Hi Reinart,

            It could be one of 3 things:
            1. The message.
            2. The audience.
            3. The experience. i.e. the funnel.

            For the experience, try installing hotjar and make sure that your visitors are able to find the checkout button. It’s a more common problem than most e-commerce businesses realize.

            For the audience, it’s not so much about finding the right people, but more of finding the right time. Are these people in a state where they need your product? Are there event or occasion “triggers” that may raise their urgency for your product? This may or may not be so applicable depending on what you sell.

            For the message, it’s about framing – are your products either an impulse buy or low-ticket or both?

          • Reinart Bacalso

            Hey Jason,

            Thank you so much for taking the time for this btw!

            Hotjar is a great idea. Will bring it up!

            The thing about the product is it’s luxury leather products. There’s no real “need” being addressed that can add to the urgency. It’s more of a luxury thing so I’m thinking that what would help the most is a really great backstory, or reviews that talk about the durability and quality of the product.

            Right now the product page is literally just a 100 word description, an add to cart button, and a lot of lifestyle pics (high quality) of the products. No reviews, no testimonials, no big name person using the luxury product.

            My suggestion to the company was to add those elements in – cool (authentic) branded backstory, reviews/testimonials, images with influencers in the market. Do you think that’s a worthwhile suggestion?

            For the message, it’s not an impulse buy or low-ticket at all. Luxury.

          • Hi Reinart,

            Glad you find my advice useful – and I would appreciate if you leave me a note to share about your experience after implementing any of my suggestions.

            I think you pointed out a couple of things that could be better. Key here is – which of these have the biggest leverage? Here’s an example: let’s say our ad CTR is 10% and squeeze page conversion rates are at 20%. You could get your ad CTR up to 15% or squeeze page conversion rates up to 30% in order to make your funnel more profitable. Which would you do? I would choose the latter – because 10% CTRs are pretty insane already. 15% is off the charts. On the other hand, getting from 20 to 30% conversion rates in squeeze pages is relatively easier.

            If you can identify the best levers, you can easily optimize anything.

            I think your suggestions make sense. I’d think that the company is going to “push back” on some of your suggestions, or agree to it but never do anything. You may need to hold their hands and implement.

            With luxurious products, you’ll want your message to say 3 things:
            1. Exclusivity -> think importance, meaning, sense of elevated status, presence and identity.
            2. Fear of missing out -> this is self-explanatory.
            3. Variety/novelty -> think fast-fashion and the trend towards owning more than one pair/set of any products.

            To answer your question directly, I am not in a position to judge if the idea would work, but I think the influencer approach is overblown. Unless you can do something very different from what most retailers are trying to do with “influencer marketing”, I’d suggest you to try something else.

    • Hey Reinart,

      sound really interesting 🙂

      I am in recruiting and I run Facebook ads (as a Newbee) for local job offers (insurance industry) all over Germany. Audiences and budgets are small (even smaller 😉 ).

      Would you mind explaining your strategy mentioned above in a little more detail?

      I think this would be interesting for a lot of people who are interested in Jason’s experiment (for small biz with a small budget).

      Thanks in advance 🙂 and greetings from Berlin.


      • Reinart Bacalso

        Hey Hans!

        Well, it just involves using the post created from making an Ad using the Web Clicks Objective (which gives you a headline, news feed description, etc., makes it look like a clickable link/article thing) on a Page Post Engagement Campaign objective by getting its Post ID and using it when creating the ad using the “Use Existing Post” button.

        This allows you to run Web Click Ad Types with the PPE Objective, and allows you to run the Web Click Ad Type at a lower minimum budget.

        Although, you can also actually run a Web Campaign at a budget lower than $5/day if you choose to be charged per impression instead of per Click. But in my tests, that is not optimal and you get really bad results haha so better try it with PPE. 🙂

        • Awesome 🙂 Thanks a lot, Reinart 🙂

  • Jason, this is really a great experiment and I am looking forward to follow up 🙂

    Cheers, Hans

    • Thanks Hans, let me know when you’ve got questions 🙂

  • Hi Jason, interesting experiment. Perhaps you could also try to increase the budget for a limited time so that Facebook can better optimize the ads delivery. I know this is meant to be for small businesses, but maybe you don’t have to run the ads every day in a month.

    PS Have you tried using another channel to accomplish the same goal (generaate leads), e.g. Google search? Just an idea, I don’t know what are the costs for your keywords in the US .

    • Hi Roman,

      Interesting suggestion. Are you saying that $5-6 per ad set is too little money for Facebook to optimize the spend? If so, do you mind sharing how that came about?

      If we were to do so, would you suggest that we daypart, so that we don’t run the ads throughout the month?

      It’s certainly not a regular suggestion but we may take it up in the future as an extension of this experiment.

      P.S. AdWords cost more than FB ads in the States.

      • If you optimize for clicks with a budget of $5 and one click costs you around $1, Facebook has only 5 clicks / day to optimize the delivery. That can be too low. Although Facebook used to claim that it needs at least 20 results / day to make the algorithms work properly, I heard that the number is getting lower from a FB account manager.

        What I would try, knowing what kind of budget you have, would be to run the traffic campaign for 3-5 days in a row with $25 daily budget and only 1 adset (and perhaps only 1 ad in that adset) on a limited audience of 50 – 400k people (for more than 400k it might be better to use oCPM optimized on clicks or some website goal but it would also require bigger budget).

        In the past I was using small budgets and a lot of adsets / ads per campaign and the performance went up a lot since I started using bigger budgets, larger audiences (500k – 2M in my case) and fewer adsets / ads). I usually mix 2-3 ads in 1 adset (or in 2-3 adsets if I want to AB test, i.e. 1 ad / adset and the same targeting) for acquisition campaigns for this kind of audience and it performs very good.

        • Hi Roman,

          Here we’re spending about $5/ad set but did you know that with more ad sets in the same campaign, it helps Facebook optimize our campaign? 🙂

          If you were to do a split test between say, a campaign with 2+ ad sets vs 2 campaigns with 1 ad set each, you might see that the former outperforms the latter 2. We have seen this previously in some client tests.

          When you say “performs very good”, how are you measuring it and what kind of results did you get relative to the past?

          • Hi Jason,
            I have never tried to AB test 2 campaigns, one with more and the other with less adsets.

            However what I did was that I split tested following – 1 campaign, 3 adsets, 1 ad in each adset = 3 variations. I used a split test feature in that allows me to split the target audience any way I want (in this case to 3 thirds). This way I tested 3 different creatives. Then I did the same campaign without splitting the audience, again 3 adsets with 1 ad each, every adset with the same budget (more or less same number of impressions for every ad, etc.).

            The second way produced better results for every ad, i.e. people were seeing different creatives. I did that with quite a big budgets (thousands of $) to have enough data.

            I’m not sure how was your test done, but it is quite possible that it worked as you described.

            I measure the campaigns performance based on cost of acquisition of an active user (they have to signup / download app + perform one of two key activities). Performs very good means compared to other marketing channels and also compared to the past when I used smaller audiences, very segmented and with smaller budgets / day / adset.

  • Hey Jason, awesome article, really enjoyed reading it. Good to see marketeers who are discussing their FB campaigns and help to stimulate debate! I have one question for you: Have you thought about using traffic and engagement campaigns simultaneously, but for a slightly different purpose? For example I like to use engagement ads to gain social validation, so the post-ad is just a photo (mostly of real people) where I work hard to build a lot of likes on the post. And then, when people see the traffic ad they are more likely to click it because of it. So one compliments the other kind of thing. What do you think? best, elco

    • Hey Elco,

      Thank you and I appreciate your compliment.

      I’ve done something similar before to what you’re suggesting, but in this case it wasn’t considered a necessary part of the “minimal” approach to making ads work for me.

      I’ve previously done it for bigger companies and institutions who want to portray their company in a certain way.

      The problem is that only a small number of people who see the engagement ad also see the traffic ad. We weren’t able to exactly use retargeting as well since we can’t yet retarget the people who engaged with a specific post. I think you’re referring to running them concurrently, but I haven’t tested whether my link ads work better when I run engagement ads at the same time.

      If anything, by the way, Facebook hates it when you run multiple campaigns targeting the same audience. They’ll see you as “monopolizing” the audience. I had huge ad delivery issues 1.5 years ago when I did this. I haven’t had the problem since but I’m not sure if that’s changed.

      • Jason, thanks for your quick response. I agree on the “monopolising” audience point, but couldn’t you still run engagement and traffic concurrently split out over mobile v desktop, or by days of the week [m,t,w,t] v [f,s,s] or by time am v pm, etc? Wouldn’t more people click the conversion ad, if they had just also seen an engagement post with 20 comments and 20 shares saying how awesome the blog is?

        • Hi Elco,

          My guess is as good as yours 🙂 Have you tried it?

          • Yes, I have tried splitting across days of the week, which seemed to be working well in the past. However, now I do favour setting up a funnel with showing an engagement post 0-3 days after x, and then traffic 0-7 after x for example. Where X can be retargeting, 3 sec video view, etc.

    • Hi,

      You can, technically. I can tell you that the arrangement you suggested aren’t ideal from a technical point of view. But first, may I know if you have tried it yourself?

  • Ken

    Hey Jason,

    Curious to know how things have been going since your last update in January! Any chance we can get a sneak peek at your “final” results? Leads generated, CPL, etc? Thanks for this great article. Really fun to read through your process ,”live.”

    • Hey Ken, thanks for your interests!

      A couple of things changed since the experiment, but the results were both good and bad:
      – We generated sub-$2 leads.
      – We didn’t get enough leads to make this worth it.

      Key lessons:
      1. Our main takeaway, like that of our clients’, is to have an entire content promotion, distribution and relationship building to grow the traffic to the site. Clearly we can generate leads at an affordable cost with warm traffic, but we can’t rely on just paid to grow the traffic.
      2. Play the long-term, instead of short-term game. Building traffic on site and retargeting is the mark of a long-term, evergreen launch type of marketing strategy. It’s not meant for time-limited big launches. Or it’ll be no different from running the ads to cold traffic directly from a cost perspective.

      What we’re doing now:
      1. We launched a brand new product and personally sold to 10+ people on a 1-to-1 level to understand their needs, pains, etc. at a deep level.
      2. We hired an amazing copywriter to write ARM-style emails.
      3. If all goes well, we’ll also hire the same copywriter to do the sales copy.
      4. Once done, we’ll start our lead generation and nurturing campaigns. For now, we’re focusing 100% on building traffic (with ads and other distribution) – and we’re getting pretty high quality and relevant clicks to our website from what we can see in Google Analytics. We also compared our GA data against FB and we see that the drop-offs are minimal, which means that we can also profitably bring people onto our website. We are currently looking at CPCs of $0.30 to $0.50 so we’re happy with this and want to keep this going.
      5. We just worked out a new publishing schedule and we will be publishing new blog posts once a month!

  • Jean StClair

    “brought in 702 visitors to-date at just $0.53 per click. This is just $0.03 above our original target of $0.40-0.50!
    Since that ad took off, Facebook spent 93% of our budget on it. It was also responsible for all 10 new subscribers we gained.”
    I’m new to FB advertizing, and I really don’t understand it.
    You spent around $350 to gain 10 newsletter subscribers? How does that make sense?
    If someone could explain that to me I’d be most grateful