How much data is enough: three rules for working with information in decisions

The report took three months. It ran to hundreds of pages, with dozens of charts and references to everything that could possibly be studied. The decision built on it lasted three weeks. The data was not too thin. The trouble was that nobody checked whether the data answered the real question, or whether the conclusions contradicted each other.

Foreign companies entering Russia produce reports like this more often than anyone. The market is unfamiliar, the stakes feel high, and the natural reaction to uncertainty is to collect more. Collecting more feels like work. It rarely feels like what it often is: a way to postpone the decision.

There are three rules that keep work with information inside reasonable limits when any strategy is being built. I will call them sufficiency, coherence and objectivity. Below is what each one guards against, and what each one looks like when the market in question is Russia and the people deciding sit somewhere else.

Why three rules and not one

Sufficiency. The volume and depth of research match the goals, the time and the resources. They do not match an abstract ideal of completeness.

Coherence. Conclusions, decisions and their consequences stay logical in relation to each other.

Objectivity. There is a visible path along which information turns into facts and facts turn into conclusions, and every step on that path gets checked.

It has to be these three, and not one general slogan like “do good research”, because each closes its own weak spot. Without sufficiency a project drowns in data collection that never turns into a decision, because there is always a little more to learn. Without coherence decisions get made in pieces, and sooner or later two neighbouring decisions start to contradict each other and undermine trust in the whole strategy. Without objectivity any convenient assumption can pass as an established fact, and the decision rests on sand that looks like stone until someone puts weight on it.

Sufficiency: the temptation to collect everything

A team without a clear threshold of “this is enough” tends to stretch the research indefinitely. There is always one more source, one more variable it would be nice to account for. In the end the budget set aside for analysis eats the budget needed for action.

The rule asks you to decide in advance what minimum of data you need to make the decision with acceptable risk, and to stop there. You stop at that minimum and do not chase an illusion of completeness.

For a company selling into Russia from abroad, the temptation has a specific shape. Head office wants the Russian market described in the same format as every other market: size, segments, competitor shares, the full template. Some of those cells are hard to fill reliably for Russia, and a lot of time goes into filling them anyway. Meanwhile the questions that actually decide the launch get less attention. Is there search demand for this product in Russian, and how is it phrased? Who already answers that demand in Yandex results and on the marketplaces? What does a buyer expect to see before they write to you?

Those questions have practical local answers. Wordstat shows how Russian buyers phrase the need, and the phrasing often differs from a translated keyword list. A look at Yandex search results and marketplace listings shows who already holds the demand. A few weeks of a modest test campaign with Metrica and call tracking in place tells you more about real behaviour than a desk study of the same length.

So the sufficiency question for Russia sounds like this: what is the smallest set of facts that lets us decide whether to launch, where, and with what offer? Everything outside that set can wait until there is money flowing and real data to look at.

Coherence: decisions that cancel each other out

Coherence catches a different error: fragmented thinking, where each decision is made in isolation, without checking it against the ones already made.

The classic example is a decision to cut the price to attract one group of customers, made without regard to the fact that the same company presents itself as premium to another group. Each decision on its own is logical. Together they destroy each other. Coherence means holding the whole chain of decisions in your head at once, instead of solving tasks one by one as if the earlier ones did not exist.

In a market entry this error has a structural cause. Decisions about Russia are usually split between people who rarely sit in one room. The global brand team owns positioning. A local distributor or partner owns pricing and the marketplace listings. An agency owns the advertising account. Someone translates the site. Each of them makes reasonable choices inside their own area, and nobody checks the sum.

The result can look like this. The brand is sold as premium in the global deck. On the marketplace it competes on discounts, because that is how the partner moves volume. The ads promise expertise and personal service, while the translated site has no Russian phone number and the form answers in English two days later. No single decision here is wrong. The combination tells the Russian buyer three different stories about one company.

The fix is unglamorous. Keep one written chain of decisions for the market: who the buyer is, what the offer is, at what price level, through which channels, with what promise. Every new decision, whether it is a promotion, a new keyword cluster or a change on the site, gets checked against that chain before it ships. If it contradicts the chain, either the decision changes or the chain changes openly, with everyone told.

Objectivity: where facts turn into wishes

Objectivity catches the substitution of what we want for what is. It asks you to trace explicitly: here is the data, here is the fact derived from it, here is the conclusion drawn from the fact. At each step you ask whether an assumption slipped through dressed as something established.

Most often the error happens on the step from fact to conclusion. The fact “sales grew after we changed the packaging” easily becomes the conclusion “the packaging caused the growth”, although dozens of other factors could have been at work and nobody checked them.

Companies working in Russia from a distance are especially exposed here, for two reasons.

The first is that many of the factors moving Russian demand are invisible from outside. Seasonality differs by region and category. Marketplaces run their own promotions that shift sales between sellers. Competitors enter and leave. The exchange rate changes what imported goods cost on the shelf. A head office looking at a monthly sales line sees the movement and none of its causes.

The second is measurement. Western teams often bring their analytics habits with them, and those habits assume Google Analytics, Google Ads attribution and email as the main contact channel. In Russia the search traffic largely comes from Yandex, the native analytics tool is Yandex Metrica, and a large share of real conversations happens by phone and in messengers. If your tracking misses calls and messenger chats, you are building conclusions on a partial picture. If you compare numbers from Metrica with numbers from GA4 without understanding how each one counts sessions and sources, you can produce a “fact” that is really a difference between two tools.

The working habit is simple to state and dull to practise. Write the fact and the conclusion as separate lines. Next to each conclusion, write what else could explain the same fact and whether you checked it. If the answer is “we did not check”, label the conclusion as a hypothesis and treat it as one. If you need help building that measurement layer before you trust the numbers, that is what customer research for the Russian market is for.

Why modest decisions outlast impressive ones

All three rules work by cutting things away, and so they are invisible in the finished decision. What you see is what is missing: tons of data collected just in case; a conclusion that argues with a decision made a month earlier; a convenient assumption put where an unchecked fact should be. The modest look of such a decision is misleading. Most of the work went into removing what did not belong, and nobody shows the removed parts.

Everyday decisions work the same way, only without documents: a conversation with a partner about the future, a choice between two job offers, judging someone’s advice. Nobody runs a formal analysis there, and it is not needed. It is enough to notice that the decision you want to make right now usually appeals through its presentation, while what will hold it up is whatever is left after the cutting.

This also explains why confidently presented decisions sometimes fall apart at the first check, while modest, less impressive options last for years. Confident delivery and durability are different things. Durability comes from three dull rules that nobody sees from outside. The persuasive tone in which a decision is announced adds nothing to it.

For a foreign company this is worth keeping in mind when the Russian plan is presented to head office. The thick report with the full market template will look more serious than a short document saying “here is what we know, here is what we decided, here is what we assume and will check in the first month”. The short one is usually the one that survives contact with the market.

What to do before the next decision about Russia

1. Name the decision first. Write down what the research has to decide, and what level of risk is acceptable. Stop collecting when more data would no longer change the answer.
2. Use Russian sources for Russian questions. Demand phrasing from Wordstat, competition from Yandex results and the marketplaces, behaviour from Metrica and your own calls and messenger chats.
3. Keep one chain of decisions for the market. Buyer, offer, price level, channels, promise. Test every new step against it.
4. Separate facts from conclusions on paper. For each conclusion, list the alternative explanations and mark which ones were checked.
5. Prefer the short plan that holds over the thick one that impresses.

A good decision rarely looks impressive at the moment it is made. It simply keeps working months later, when the impressive ones have collapsed under their own weight.

Frequently asked questions

Sources

• Russian version of this article on belokrylovo.ru: Три принципа работы с информацией в решениях