Artificial Intelligence Calculated Optimal Prices For E-commerce, Increased Gross Profit By 20.1%, And Reignited Demand

Gold-Standart.com was in an advantageous position until the onset of the coronavirus in the spring.

From June 2019 onwards, we employed an automated real-time repricing strategy in response to our competitors' prices. Whenever competitors raised the prices of specific SKUs, our price automatically escalated to maintain the required position. If competitors lowered prices, our price adjusted accordingly.

The Imprice cloud platform performed this strategy fully automatically, including the gathering of competitors' prices and updating the current prices across our brick-and-mortar shops and our online platform. We only periodically fine-tuned the pricing rules: altering margin constraints and shifting the target price position — be it the second, third, or fourth price in the market.

Automation of repricing based on competitors' data increased our gross profit by 16% in several months. We experienced a Like-for-Like (LFL) growth of about 30%, with a satisfactory margin level.

March 2020 marked a significant downturn.
With the closure of gyms, most people ceased strength training, and sales of proteins and gainers took a dramatic hit. Moreover, we had to temporarily close our brick-and-mortar stores. This resulted in a 60% drop in turnover.

Due to fluctuating currency rates, the cost of procuring goods escalated, leading to a sharp decline in margins. We experienced significant losses. By the time April rolled around, it felt like the company's demise was imminent.

We continued to utilize competitive dynamic pricing, but sales had dropped close to zero. It seemed nobody at the market understood what prices could be optimal.

This prompted us to employ a novel pricing strategy, AI-powered price optimization based on demand, rather than competitors' offerings.

We significantly improved identifying KVIs already in 2019.

First, we leveraged Imprice ML-module for recognizing all KVIs in our assortment. The algorithm identified 150 KVIs, whereas our employees only knew about 50 of these items.

Second, we set into the Imprice platform list of stores we considered our competitors. Imprice kept our price attractive compared with these retailers: every time, it took the lowest price among set competitors and made our price slightly higher, considering our margin restrictions.
That strategy helped us grow in both orders and profits.

In April 2020, sales of KVIs fell, like sales of other goods. Imprice experts offered us to check which stores were affecting, in fact, the sales of our KVIs: to whom the Gold-Standart.com consumers went if they offered a better price. We launched an AI-driven Imprice module, "Automatic identification of key competitors.

Imprice identified our key competitors. Surprisingly those were bad reputation stores with extremely low prices.

We provided consumers good-level consultations about assortment; we posted high-quality content on social networks; we sponsored sports events. It seemed that consumers had a high loyalty and recognized Gold-Standart.com as a confident store with good service; we supposed our competitors were well-known large stores.

But in fact, our key competitors were online stores that:
✓ processed 100 orders per day or less,
✓ had a very narrow assortment and low-quality sites,
✓ set the lowest price on the market and used price aggregators for advertising.
These stores "stole" sales of our KVIs.

At this point, artificial intelligence had started to manage prices of about 2,000 SKUs. We monitored every step of optimization algorithms and tried to understand the reason for every price changing made by AI.

We got fantastic insights: Imprice recognized short-term surges in demand and immediately increased prices for the corresponding items, such as Vitamin D3, Melatonin, Zinc. Only later we learned that articles had appeared in the media stating these vitamins and supplements improve resistance to COVID-19. People rushed to buy the recommended, and sales of these items began to rise. Artificial intelligence detected the increased demand and raised the price of necessary SKUs. For example, we sold vitamin D3 at the highest price on the market: it started to be out of stock in other stores, and our sales grew at a price 1.5 times higher than usual. In July, the over-demand ended, and the algorithm lowered the price again.

As a result, the category of vitamins hit the second place in gross profit, while before it wasn't even in the top ten.

Another surprise was consumers were willing to pay more than we assumpted for many items from our "long tail" segment. Imprice identified these goods and increased prices. Sales in pieces remained about the same level as before, but now we earned more on each item sold.

Finally, there were many goods which prices automatically followed competitors' ones before, and consumers stopped buying those from the start of lockdown.
Artificial intelligence reduced these items' prices below the market level, and surprisingly demand reignited.
What exactly happened? Suppliers' prices went up, all stores raised shelf prices, but people had no willingness to pay so much, with a new lower living standard. All competitors made the same pricing mistake, and we just followed the market and lost our sales. Imprice switched our prices back to the pre-COVID-19 level, so sales returned immediately, with a still good gross profit.

What have we learned after launching price optimization?

First output:
Even top competitors are not gods and can make mistakes. Now is a time of turbulence, and the business environment changes rapidly every day. In our case, all stores made sports nutrition too expensive for consumers, which led to sales falling.

When sales dropped down, it better not to repeat: "This is a crisis; people stopped buying." Test different prices, try to put them below the market level if margins allow. We made this step, and our disappeared demand reignited.

Second output:
The company's experts are not gods either. Our specialists assumed as main competitors large stores, but in fact, low-priced stores with a bad reputation "stole" sales and consumers from us. We switched from assumptions to real data and improved KVIs' sales.

Third output:
It's possible to find underpriced goods in the assortment, even during an economic crisis. We increased such items' prices and improved gross profit, selling the same number of pieces.

Before, we lost profits because we missed short-time shortages of goods at competitors' stores. There are only relatively small chains and stores in our market, not large chains with perfect supply management. Stores have a short-time lack of some supplements and vitamins often. With Imprice, if some popular items disappear from competitors' stocks, the AI-module immediately detects a surge in conversions, rapidly raises our prices of such goods, and our company gains more profits.

Another example is "background", "long tail" goods. Since July, the demand for sports nutrition has revived: gyms have opened, and sportspeople have started to buy protein. Protein for strength training is an expensive product, and its price reaches 50-60 euro. Consumers buy something less costly with protein often, for example, dietary supplements. Imprice records which products are often purchased with protein and raises prices for these complementary goods; so Gold-Standart.com improves profits per order.