Today we will be covering an intelligent, one-of-a-kind tool that’s exclusive to Splitly users. Through machine learning and artificial intelligence, Profit Peak automates pricing for FBA sellers to maximize profitability.
Pricing optimization is quite a complex process, that’s why you need dynamic pricing software that works with Amazon’s ranking algorithm.
We’re going to break down Profit Peak’s algorithm so you can understand how and why it works, to get the most out of this intelligent pricing tool.
Webinar Replay & Slides:
As always, here’s the full webinar replay:
The Problems With AB Testing (Before Profit Peak)
Before this dynamic pricing software was developed, you would have to stick with split testing. i.e running two variants at different pricing points to see which one performs better.
AB testing the price works the same way as testing your main image, but you would look at profitability instead of conversions.
However, this process is sub-optimal for the following reasons:
1. Less Certainty Due to Amazon’s Algorithm
Amazon’s product ranking algorithm threatens to interfere with our pricing optimization process. You may already be aware of sales velocity its knock-on effect on product ranking and visibility.
Here’s an example of this mechanism in action:
- Lower pricing on day 1 results in increased sales
- Product ranking and visibility is boosted on the following days
- The next day with higher pricing receives an unfair advantage due to increased rankings
- Vice versa, higher pricing days resulting in fewer sales pull down the ranking on lower-priced days
The knock-on effect we experience gives benefit to higher prices and hurts lower prices.
In cases when the lower price still wins, in spite of bias, we can have increased certainty in the winning variant.
But if the higher price wins we must be more careful when drawing conclusions. In the long-run, the higher price could be eroding your overall ranking.
2. It’s Time-Consuming
AB testing with Splitly makes the entire process more organized and efficient. However, it still requires manual effort and attention to set up and run successive tests. Since you’ll only be testing 2 prices at a time, you will need to keep running more tests to find the most profitable price.
3. Supply & Demand Also Affect Profitability
Competitors can change their pricing, run promotions or disappear completely while they restock. These events influence the delicate balance of supply and demand within product categories.
If you’re running AB tests for pricing optimization, you’ll need to notice these changes and react accordingly by retesting. It’s going to be very difficult to keep track whilst managing all other aspects of your operation.
So instead of running successive split tests and then having to factor out these problems, we decided to create one-stop automated pricing solution- Profit Peak.
Profit Peak’s Dynamic Pricing Software
Pricing is changed every day while Profit Peak runs. As we discussed before, orders and sales velocity are going to influence future days. Since all these prices are interconnected and influencing each other, then how can we make sense of all this?
We can’t on our own.
That’s why we rely on Profit Peak’s machine learning algorithm to find relationships between pricing and profitability.
Let The Software Work, Look For the Peaks
Our solution to make sense of this “messy” data is Regression – a branch of machine learning that helps us identify patterns.
In our case, we use it to map price and profitability neatly on a graph. We named our dynamic pricing software “Profit Peak” as the peak of the output graph reveals our most profitable price.
More Factors Taken Into Account
If it wasn’t complicated enough, sales velocity affects products differently. The knock-on effect it has on ranking is not always equal.
Therefore we need to run multiple regression algorithms to find the best relationship with the least errors. Once this is set up, Profit Peak simply asks “should I increase or decrease price to increase profit?”
It’s a purposefully slow algorithm that doesn’t jump around. It slowly increases or decreases price to identify trends with low errors for greater certainty in results.
Getting Started With Profit Peak Automated Pricing
The true beauty of using Profit Peak is the ease of execution. There’s really not much to do and it’s all very straightforward.
When it comes to minimum and maximum price settings:
- You should still be profitable at your minimum price
- You should be comfortable selling at your maximum price
- Achieve a good spread, +/- 25% on your current price should be plenty
Setting your Unit Cost is important for working out the profit. Profit Peak already takes Amazon related fees and PPC into consideration. But the cost of purchase and shipping will need to be entered manually.
Dynamic Pricing Software Advanced Settings
If you use advanced settings, they can override the algorithm and take away from the price optimization process. Here’s a quick explanation of each setting:
- Child ASIN Sync – When you have multiple variations of a product at the same price (such as color variants), this is how you can group them together and test all products in one block.
- Target Daily Sales – Think of this as a “sales velocity regulator”. It ensures a minimum number of sales per day by cutting pricing optimization. Remember, you can’t find the optimal price AND maintain the highest sales velocity throughout.
- Price Schedules – This enables you to automate turning on and off of Profit Peak. This can be especially useful during lightning deals or if you switch back to AB tests.
- Price Decimal – By default, Profit Peak will lock this decimal and run through whole dollar amounts only. We auto-fill this field to whatever your current pricing chemical is for your product (e.g. $.97). This prevents “weird” decimals such as $.13 from being used. If you do remove this value it will only run through the higher end decimals.
- Stock Deplete Date – If you have a restock date this will ramp up automated pricing to reduce the chance of running out of stock before that time. Although, it is perfectly OK to run out of stock since you will maintain your sales velocity while you restock. But this optional feature can be used when you don’t want to restock and intend to sell off at maximum profitability.
Collecting Results & Analyzing Graphs
Remember we mentioned earlier that peaks indicate your most profitable pricing? Well if your chart seems irregular and doesn’t show a clear peak, it’s likely you need to wait and collect more data.
Mapping difficulty varies from product to product but data is king. Running Profit Peak for just a few days or a week is not going to be long enough to acquire data. Regression and machine learning is data hungry so expect to run for at least 30 days seeing good peaks.
Pausing Profit Peak & Switching Back to Split Testing
Due to the interference and data corruption, we don’t allow you to run Profit Peak and AB testing at the same time. Instead, you should focus on periods of running one then the other.
Once your optimum price does not seem to be moving much, you may consider pausing Profit Peak to pursue further split testing. However, it’s well worth revisiting Profit Peak regularly since marketplace supply and demand can change frequently.
Questions & Answers
Should Profit Peak be run on each child variation separately?
If they are the same price (such as a difference in color) then you should sync them using Profit Peak’s advanced settings. If they have different pricing then you must run each separately.
Does Profit Peak also target winning the buy box?
No. This is an automated pricing tool for FBA and private label sellers only, not a buy box re-pricer. This involves a different algorithm and is not the focus of this particular tool.
Does the Stock Deplete Date help you deplete stock quickly to increase sale velocity?
No. It actually does the opposite. If you want to do this we recommend using the “target daily sales” option with a reasonably high number. This will force Profit Peak to drop the price and encourage more sales.
If a competitor lowers their pricing will Profit Peak react quickly?
We designed Profit Peak to be slow learning and don’t want it to react too quickly. However, we built the following mechanisms to help solve this problem:
We implemented a half-life on all data collected. The most recent data has a higher impact than older data so it can adapt to changes in the market. This means changes to supply/demand can be taken into account more quickly.
Profit Peak monitors for “bad periods” when sales drop off out of the blue. It will start dropping the price slowly to try and encourage sales to rise. These bad periods could be a fluke or may be triggered by supply/demand changes.
When is AB testing preferred over using of Profit Peak?
When optimizing price, just stick with Profit Peak. The only exception is when you have two particular pricing sweet spots you want to investigate. Just remember to take the bias towards higher pricing into consideration when AB testing.
Can you explain more about this biased towards the higher price?
When you have a lower price and sales increase, product ranking will be boosted on proceeding days. Increased visibility could boost sales in spite of higher pricing. This is an effect of sales velocity and Amazon’s algorithm.
Will Profit Peak become available for other Amazon Market Places?
If we get enough requests we always consider adding extra features and further software developments. That’s why your feedback is so vital! It really helps us provide better tools to suit your wants and needs.
How much of a high and low price should we set from our current price?
Usually, 25% leaves enough room to thoroughly test pricing.
The maximum price is not too important because Profit Peak will detect significant drops in sales and reduce the price accordingly.
When you set a minimum price, Profit Peak will not allow you to go into negative profits (not including PPC).
We really hope we’ve helped to de-mystify our dynamic pricing software Profit Peak.
Don’t just copy others or take a shot in the dark when it comes to pricing. Collect your OWN data and draw your own conclusions using Profit Peak’s machine learning algorithm. This simple process could just uncover a whole bunch of profit you’ve left out on the table!
We’re incredibly proud to continue developing intelligent tools to help maximize your Amazon revenue. If you’ve got any questions or suggestions, please do leave us a comment below.