How to Track Competitors' Prices. Four Approaches Presented at Pharmliga

On 1 October 2026 DSM Group held the sixth autumn Pharmliga forum, "Pharmliga: Meeting of Leaders", in Moscow. The programme had no session devoted to price monitoring. Even so, speakers came back to competitors' prices three times: at the morning panel, at the afternoon panel on AI and in the talk by the founder of Grand Capital. I retell what was said on stage and add my own view. Speakers' figures are given as they stated them: they cannot be checked against public sources.

Expensive money and a short horizon

The backdrop was set by Rodion Lomivorotov, executive director and senior economist at Sber. He expects the Bank of Russia key rate to end the year at 14% or 13.75%. Next year, in his forecast, the rate will gradually fall to about 12–13%, while new commodity and inflation shocks would keep it near 15%. The economy has barely moved since mid-2025 and grew by about 1% over the year. In a Sber client survey, businesses named a rate below 12%, ideally below 10%, as comfortable for investment. In a poll during the talk, the audience did not believe that inflation would fall to the 4% target next year.

In the evening Alexander Auzan, dean of the Faculty of Economics at Moscow State University, spoke about planning horizons. In Russia the horizon is about two years, while elites in the "Asian tigers" look 20 years ahead. Growth, he said, needs a long view, trust and the ability to reach agreements.

When working capital is expensive, a pharmacy chain finds it hard to open new stores, so it defends its margin through price. Prices then have to be reviewed often, without waiting for quarterly results.

Pharmacy chains: the online shopper compares prices

The morning panel, "There Is Profit but No Money. The Perfect Storm of Retail", was moderated by Oleg Komarov, CEO of Kvadrat-S. The participants were Sergey Korepin (Dobraya Apteka, Arkhangelsk), Vadim Ranshikov (TableTochka, ARTES Group), Emil Khubaliev (Vita, Samara), Egor Budrin (Tvoy Doktor, Zdes Apteka) and Kadimagomed Ibragimov (SOYUZ SHAKH). They agreed that consolidation continues and margins are eroding.

Price elasticity was discussed separately, that is, how much sales fall when the price goes up. The panel heard that demand online is much more elastic than in a store: online, the shopper cares only about availability and price. In a store, the pharmacist can explain why a more expensive product is worth buying, and the pharmacy charges a premium for that service. The conclusion was that price matters to some customers, while the rest pay for service. Another point made: the more expensive the product, the higher the elasticity.

I explain the difference by how shoppers see prices. In a store they compare a pharmacy's price with one or two neighbouring signboards, while on a pharmacy aggregator the price sits in a single list sorted from lowest to highest.

Yet few chains hold the lowest price. In my talk at Apteka 2026 on 28 September, I showed the share of products for which a pharmacy chain sets the lowest price near a metro station. The data came from monitoring online pharmacy prices in Moscow. Several chains hold the lowest price on almost half of the products. Large chains hold it on 8–12% of products, and two more chains on less than 2%. Everyone buys at roughly the same purchase prices, and most pharmacies set prices without looking at their neighbours. In half of the cases (52%), the gap between the most expensive and the cheapest pharmacy near a station is 20% or more. A ranking of chains by price relative to their neighbours is in my article on Moscow pharmacy prices by metro station.

Setting the lowest price across the whole range does not pay. If your gross margin is 20%, a 5% discount takes away a quarter of your gross profit, and you need to sell a third more to earn the same money. A chain does better by giving products different roles. First it identifies the products by whose price shoppers judge the whole pharmacy, and keeps their prices in line with the market. Shoppers add companion products to the same order without checking their price, and this is where the pharmacy earns.

Akrikhin: in-house agents instead of bought data

The afternoon panel, "How the Business Management Model Is Changing. Expectations and Disappointments of AI", was moderated by Boris Agatov, an expert on AI adoption. The participants were Mikhail Platonov (PepsiCo), Ilya Kadatsky (Finn Flare), Eleonora Akhmedzyanova (Askona), Andrey Rutskin (Kholodilnik.ru) and a representative of Natura Siberica's marketing analytics team. The pharmaceutical industry was represented by Sergey Isaev, director of digital business at Akrikhin.

At Akrikhin, the panel heard, dozens of AI agents are already at work, that is, programs built on a neural network that carry out a task step by step on their own. One of them replaced bought market data. A supplier asked 1.5 million roubles for data from a single source, while the agent collects the same data without new hires. Another agent watches prices. It reports that prices have changed on a given website, in a given pharmacy chain, in a given city, and suggests thinking about a response. The first version of the agent platform took three weeks. The savings were put at 50–80 million roubles: that is what the contractor proposals the company turned down would have cost.

It is natural for a manufacturer to watch the shelf price of its product, and a price-change alert is a good first step. Collecting prices today really is fast and cheap. It is harder to understand whose price it is and what to compare it with. On a pharmacy aggregator, one chain name can cover several signboards with different prices. One chain in Moscow keeps several price levels for the same product across its pharmacies. An alert that is not tied to a specific pharmacy and to the actual owner of the price tag easily turns into noise.

Grand Capital: from yesterday's prices to a forecast

Denis Remenyako, founder and CEO of the distributor Grand Capital, spoke in the "Opinions of the First" session, mostly about distributor pricing. According to him, until 2024 the company saw competitors' prices with a delay of one to two days and did not take into account their sales volume or their earnings on an item. In his example, a competitor raised its price to 107 roubles, while Grand Capital set 104.99 roubles instead of a possible 106.99 and lost about 2 roubles on every pack.

Remenyako illustrated the cost of handling a pack with two drugs of the same size. A drug priced at 50 roubles with a 10% mark-up brought 1 rouble after logistics, picking and overheads, while one priced at 1,000 roubles with a 1% mark-up brought about 6 roubles.

Now the company's model takes into account pharmacy sales over three years, each competitor's share and daily stock, and how each competitor's mark-up changes. All of this is calculated by an AI employee called Leo. According to Remenyako, in two branches Leo has already brought "tens of millions of roubles" of additional earnings. He expects that soon a pharmacy, too, will know tomorrow's sales and price from its own sales history, and the company plans to offer such services to pharmacies.

I find the admission of a one-to-two-day delay honest. For a distributor with thousands of price lists and daily data on competitors, such a model makes sense. I only partly agree with the forecast for pharmacies, because pharmacy pricing works differently.

In retail the delay is usually longer than one or two days. In practice, pharmacies rarely review their mark-up on the purchase price. The timing depends on the product, but on average the mark-up changes no more often than once every few months. Repricing of stock lags by the inventory turnover period, that is, by 30–90 days.

Sales history shows how a product sold at the pharmacy's past prices, but it knows nothing about the neighbours cutting their price yesterday. A demand model also needs a long sales history, and a single pharmacy usually does not have enough of it. The neighbours' prices, on the other hand, are visible every day on the aggregator. That is why a pharmacy makes more accurate pricing decisions from current online prices than from historical data.

Retail prices move fast. I compared online prices near one metro station on two consecutive days: 8 out of 28 pharmacies changed their price within a day. And one chain cut its price by about 12% overnight in all of its pharmacies at once. Such a move by a competitor is hard to predict from history, while monitoring will see it the next morning.

From Remenyako's talk, a pharmacy can use the example of packs at 50 and 1,000 roubles. The cost of handling a pack hardly depends on its price, so a cheap item with a high percentage mark-up can bring less money than an expensive one with a low mark-up. On expensive items the corridor for the mark-up is also narrow, and shoppers check the price more carefully. So the range has to be split by role, and a single mark-up for everything does not work. I explained how to do this in my article on dynamic pricing.

Natura Siberica: cheap data and chaos

A representative of Natura Siberica's marketing analytics team explained at the afternoon panel how the company saves on data. It buys cheap raw data on the market, structures it itself and brings it to a single format. High-quality external data costs 30 million roubles a year, the cheap kind 5 million. The difference in quality is huge, but after processing the reporting becomes error-free. The same panel also heard what I consider its main warning: if a system is in chaos, AI starts automating and analysing the chaos. Later the idea was repeated in other words: on chaotic data, chaos will be scaled many times over.

The morning panel reached the same conclusion from another side. It heard that complete sales data for the whole market costs huge money, while on its own data alone an agent will do a mediocre job. Buying an agent and "working a miracle" was called a utopia. Another observation from the panel: expensive off-the-shelf AI projects have not yet brought anyone on the market a publicly named positive result.

My experience with price monitoring matches this completely. For pharmacy prices, "getting the data into shape" means four things. First, you need to find the owner of the price tag: the name shown on the aggregator does not always reveal it. Second, compare only what can be picked up today, because a "today" reservation often turns into a "tomorrow" one by evening. I described how online booking works in my article on pharmacy online booking. Third, every item must be linked to a single product catalogue, otherwise one pack gets counted twice. Finally, prices can only be compared within a snapshot of one day and one metro station.

After such processing, much less is left of the original list. Each well-grounded price recommendation takes four to five products at the input. AI is useful after this work, and without it AI only multiplies errors faster.

Four approaches and one common question

All four approaches face the same question: how to check that a new pricing rule actually makes money. On the morning panel there was advice to run many cheap tests first and only then move to expensive rollouts. With prices it works like this. If you change a pricing rule, apply it first in 10–20% of your pharmacies for 2–4 weeks and compare the result with the other pharmacies. Anything not checked against a control group should be treated as opinion. Mine included.