Hey guys!
We know that performance plays a key role for every publisher. And the best presentation of result is practical study case with real examples. That's why we've prepared a series of our own articles for you. Here you know how you to improve your results with various affiliate features, which we'll tell you more about.
Today we will talk about the auto-optimization system implemented in IMI Smartlink and, of course, available for
In simple terms, auto-optimization is a system for redistributing traffic volumes between pre-lendings and offers, depending on their performance. The algorithm is a matrix model that takes into account historical data and predictive values to make decisions about reallocation. To put it even more simply, as soon as performance metrics start to drop on one link—the system starts to reallocate volumes to the other.
The causes of slumps can be very different from technical problems from the advertiser side to changes on side of the partner. For example, payment method or domain failed, or just a bug appeared on the landing page. Maybe, the advertiser turned shaving to level up their paybacks. Or the partner changes their funnel: changing creative, source, audience, accounts, advertising formats, etc. Anyway, we need to make sure, we strive to monetize the affiliates traffic as effectively as possible at that moment.

Pic.1: Split of passes w and w/o auto-optimization
Well, less words! For the split, I’ve chosen traffic of the partner with Android 3G on France.
Read about why we split traffic into segments here.
Copying the current pub pass, I’ve got a split of two passes, one with auto-optimization and other without. The traffic is split 50/50 between the passes, with both keeping the same distribution percentages at launch of the test.
Next we'll see not only the differences in the results, but also how the percentages changed on the pass with Auto-Optimization.

Pic.2: Results of A/B tests
Given the volume of this pub, we had just over two days to take the results and get a statistically significant deviation from the use of auto-optimization. As we see, absolutely all metrics with auto-optimization are higher, for example, visit2conversion is almost 30% higher. This way the difference in income was $120.45. That allows us to conclude that using of auto-optimization for this partner on this traffic segment brings up to 1.5k additional income per month.
And now, let's see how percentages have changed in the distribution of traffic for almost 2 days. Top pre-landing increased its advantage from 67% to 88%, and the top offer, on the contrary, lost 10%. This case does not give us a clear understanding of what has changed in traffic or on the advertiser's side. But it gives us an idea that traffic works differently at a distance, across the distance, and changes in the distribution give reason for a deeper analysis to find out the reasons. Applying auto-optimization only helps to avoid excessive losses and improve the efficiency of performance right now.

Pic.3: Changes in traffic distribution
Any questions? Comment the post and we'll try to help.
Profit to all!