POF’s self-serve platform implements a CPM-based bidding system that does not take into account incremental bids, meaning what you bid is what you pay. If you have even run a few campaigns on POF, you must have wondered what a higher bid actually buys you.

We know that a higher bid is supposed to give you more volume and higher quality traffic (i.e. lower session depth traffic), but what is the magnitude of the effects? In this case study, we tackle the volume question. We’ll save quality for a later date.

The results were very interesting: there is a price point where major volume drop-off occurs and a high price point where further increases could be achieved again.

Here is the data, followed by our interpretation:

How the Data was Collected

The data was collected using the POF Traffic Volume Monitor over an 8 hour period from 12pm CST to 8pm CST on a Saturday. We used 4 different campaigns. These campaigns weren’t broad campaigns but they aren’t targeting micro-niches either. They all had frequency caps of 3 per visit and distributed traffic ASAP.

Bids were adjusted between $0.21 and $1.01 by increments of $0.10 every 15 minutes. When the campaigns reached $1.01, we would lower the bid down to $0.21 for the next 15 minute interval. Since we were testing bid prices over 8 hours, performing the test in this manner controlled somewhat for the variation of traffic over time of day.

Since we didn’t perform the test for days and days, there was some volatility over a smaller sample size as manifested through the non-smooth line, but all we’re looking for is a general trend.

How to Read the Data

Looking at the table, the “CPM Bid” of course indicates the CPM price points we tested.

“Avg Ims Per Hour” is the average number of impressions gained per hour at those bid levels.

“Marginal Imps Per Hour” is where the data starts getting interesting. It refers to the incremental amount of traffic gained from increasing the bid by $0.10. For instance, when bid was increased from $0.21 to $0.31, our data showed that the campaigns gained 1,936 impressions on average.

“% Traffic Increase” is the magnitude of marginal traffic received. For instance, when bid was increased from $0.21 to $0.31, traffic increased 127.5%, or more than doubled.

“Eff. CPM of Marginal Imps” refers to the effective CPM of the additional impressions received by raising the bid. For instance, when bid was increased from $0.21 to $0.31, the campaigns gained 1,936 impressions per hour on average. Those 1,936 additional impressions effectively cost us $0.39 per 1,000 impressions, since we could have gotten the first 1,519 impressions by only spending $0.21 per 1,000 impressions.

“Marginal Cost as % of Prev” refers to the increase in cost for the additional traffic gained by bidding $0.10 higher. For instance, when bid was increased from $0.21 to $0.31, the 1,936 additional impressions we received cost us 85% more than the first 1,519 impressions.

What does the data tell us?

We can see that there is a huge boost in traffic going from $0.21 to $0.31. We still see pretty good increases in volume up to $0.51. We still get small gains from $0.61 to $0.71, but we really don’t see any incremental traffic as we bid $0.71 and higher. That is, until we get to over $1. At $1.01 we see another increase in volume. We did not test any bids higher, but my hypothesis is that there is a clear divide between the average performing and broad-targeted campaigns, which bid in the $0.50’s and lower, and the niche-targeted, higher CTR campaigns, which bid over $0.70. Once you’re bidding over $1, you start to leave the majority of the pack behind, and the campaigns start to see more volume, beating out a majority of the campaigns. From this data, it looks like you shouldn’t be bidding higher than the $0.50’s unless you can bid over $1 as you start to see very little gain in traffic (only 4.9% more traffic from $0.51 to $0.61).

Let’s also take a look at how much the additional traffic cost you when you increase bids. As you can see, the cost goes up exponentially. While additional impressions received by increasing bid from $0.21 to $0.31 cost only $0.39 effective CPM, by $0.61, the extra traffic you gain from increasing from $0.51 cost us $2.64 effective CPM. By $0.81, it got as high as $30.46 (yes, that’s effectively $30.46 for 1,000 impressions), which clearly indicates to us that increasing from $0.71 to $0.81 just isn’t worth it for our campaigns.

Based on this data, the general takeaway seems to be to stay in the $0.21 to $0.50’s range unless your campaign truly justifies a high bid greater than $0.91. Remember, we have not taken into account the quality of traffic and its effect on campaign performance, so when you make decisions for specific campaigns, you need to analyze the overall and marginal profitability of those campaigns. What we’ve uncovered here, however, can be used as a general rule.

We hope you found this experiment interesting and helpful. If you’d like to download a copy of the data from this case study, please do so below. It will give you a good idea as to how to go about doing similar analyses for your own campaigns, if you so choose.

Hey guys, this is Aziz, and this is my debut post on the blog. I hope you guys enjoy it!

 

If you’ve ever had the chance to check out the ads running on POF, I’m sure you’ve noticed that 99% of the ads you see have the typical image-on-the-left, copy-on-the-right placement. Why is this the case? This widespread phenomenon in creative layout is molded by 2 commonly accepted principles in the marketing industry. The first is called the Gutenberg Rule, which states that when people scan a page of content, they scan left-to-right and top-to-bottom, making the top left area the focal starting point. The primary optical area is where the content needs to convince the viewer that what they are about to read or click on is going to be worth their while.

Along with the Gutenberg Rule, experienced marketers will tell you that a successful creative weighs largely on the effectiveness of your image to attract attention (50-80%), followed by your headline (10-30%), and lastly your body copy (5-10%). Simply put, according to studies, an ad will get more clicks when it includes a powerful image, placed on the left of your banner since it is the most effective when used in the primary optical area in order to get the viewer to invest more time in the creative.

In an endless sea of left-justified image placements in creatives, I started wondering what would happen if a few images defied tradition. So I decided to take that idea and make it into a POF case study with our little friend, the 310×110. First, I grabbed a batch of images that performed well in a previous campaign that targeted women age 18-21. Preserving the same targeting, I set up 3 templates: one with an image on the left, one with an image on the right, and one with 3 images across the banner. Finally, I ran the ads until each creative had at least 10,000 impressions and kept the top 2 performers from each template.

Case Study Creatives

The Results

It seems that contrary to widespread usage, left-justified image placement didn’t always produce the best CTR. We were actually pretty surprised at the clear divide in this case study. Images placed on the left performed the worst, while the images on the right got a slightly better CTR, leaving the 3-image placement as the victor.

Keep in mind that the above creatives were targeted toward one specific demographic and had collected less than 15,000 impressions each. Across data greater sample size and wider range of demographics, the results could certainly vary. What we can learn from this little case study is to never stop challenging the norm and thinking outside the box. While I still believe that the aforementioned principles hold, much of marketing is about differentiation and standing out from the crowd to catch the attention of the traffic that suffers from technology ADHD.

What are commonly accepted principles are rooted in logic and are time tested, but the magnitude of potency depends on many factors. In the world of POF, the principle of left-justified images suffers from overuse, which is especially noticeable in POF small ads that are shown 3 at a time side-by-side.

While the advertising industry seems to accept that left-justified image placement is always the most effective, I took the “test for yourself” mentality, and I urge you to step back and do the same wherever you’re advertising.

We used 310x110s here, but what would be interesting would be to see how mimicking 110x80s but testing image placement (left vs. right vs. middle) would fare in a similar case study.

If you’ve tested image placement or went against the grain another way, please share your results with us!

As a follow up to the post a couple of weeks ago on creating your own image variation packs for split testing, which some of you guys had a little trouble making the variations, we decided to put together some ready-to-go variation that you can apply to batches of your images using Photoshop actions. All you do is open up the actions using your copy of Photoshop and apply it to a batches of images on your hard drive.

We have a lot of these and have chosen some basic ones to get you started. These are by no means to guarantee you instant profits but to help get your thinking hats on with ideas to split test variations of images and ads that are already working. From our experience, you really never know what works, and the correct simple border could make a world’s difference. So, you can never test enough. At least for us, we are never going to wake up one day and think that we’re done testing.

If you haven’t already, you really need to watch the Photoshop Batch Processing tutorial. Not only will you totally understand how to use these actions, you will be able to edit the variations from this free pack and customize them. For example, you can edit the widths or the colors of the border variations included here.

Since this variation pack was put together by our very own Aziz Kamara, I’ll let him give you guys a quick guide to the usage of the download. Also lookout for some posts in the coming weeks from Aziz as he makes his posting debut on the blog.

Aziz’s Guide on How to Apply the Variation Pack

Hey guys, this is Aziz. In the zip file, you’ll fine Actions, Samples, and Shapes. The Samples give you a sample of all the variations that we’ve included in the pack and what they look like. The Actions folder contains all the kinds of variations you can apply. The Shapes folder you can largely ignore, but it’s just different cut outs that the Actions use on the images. This variation pack can only be applied to 310×110 or 110×80 images. Certain actions can only be applied to one of the two, and they are clearly labeled by their file names. You’ll see when you take a look in the Actions folder.

Here’s how you easily apply one of the variations in Photoshop:

1. Simply double click on any of the available variations in the Actions folder.
2. Have the batch of images you want to apply the variations to in a single folder.
3. After opening the Action file in Photoshop, Go to File > Automate > Batch.
4. For “Set”, pick the Action that you just opened.
5. Then, for “Action”, pick the sub-type of the Action that you want run.
6. For “Source”, choose the folder that you have your images prepared.
7. For “Destination”, choose the folder that you want the new image variations to be saved in.
8. Finally, hit “Ok”, and your variations should be created automatically.

Talk to you guys soon!