An invitation to bid arrives feeling like a starting gun. On the corporation's calendar it is the finish line. By the time the documents go out, the money for the task has been allocated, the task itself has been written up in somebody else's words, the timeline has been reconciled with the work plan, and the person who will answer for the result has already been named and has already spent political capital getting that money approved. Between the moment all of this happened and the moment you saw the announcement, six to fourteen months went by.
This article is about those months, and about what a supplier who arrived in Kazakhstan recently can actually do inside them.
How money appears inside a corporation
A large Kazakhstani company runs on an annual budget, and forming it takes roughly a quarter. In September, departments start collecting requests for the following year. By October those requests move to the finance function, where they get cut, consolidated, and sent back for rework. November and December go to defending them before the management board or the board of directors, and the budget is approved by year end. From January, the money exists as line items with wording attached.
The wording matters more than the amount. The manager who defended that line came to the committee with a description of the problem, a cost estimate, and an explanation of why the money had to be spent now. For capital expenditure, requests usually carry a business case with a payback period, and two to three years is the horizon most committees accept. A system that pays back in five years on paper never reaches approval, even when the business wants it.
Approval thresholds work in steps, and every step costs time. Below a certain amount the head of the function decides, above it the management board comes in, above that the board of directors. The exact figures differ by company, but the gap between steps is measured in months of approvals, and knowing those thresholds lets you shape a deal deliberately: a pilot priced under the lowest threshold starts in two or three weeks, while the same scope submitted whole disappears into committees for six months.
Mid-year budget revision deserves its own note. It exists almost everywhere, usually runs once, in June or July, and passes only emergencies: something broke, a contract expired, a regulatory requirement changed. Selling a planned initiative through a revision is close to impossible, because the person inside would have to explain to the committee why they failed to foresee it in the autumn.
The shape of this cycle says something plain about timing. A company that starts working the Kazakhstani market in March and expects a deal before year end is planning to land in a budget that does not exist. The realistic horizon for a first large deal is to make it into the autumn request and close the contract in the first half of the following year. Anything faster comes either through pilot amounts below threshold, or through unspent fourth-quarter money.
Buyers choose what they can defend
«Nobody ever got fired for buying IBM» is a line from the seventies that outlived IBM in that role. Gerd Gigerenzer, the German decision researcher, studied the mechanism behind it and called it defensive decision making: a manager picks the option that serves the company worse but protects them personally if things go wrong. In surveys of managers at large companies, the share who admit to such decisions comes out high.
«Nobody ever got fired for buying IBM.»
Industry saying, 1970sThe procurement procedure serves exactly this need. It turns a personal choice into a documented process, and a year later, when the project stalls, the person produces the protocol instead of explaining their preferences. There is nothing corrupt in that. It is ordinary risk management in a company where the shareholder cannot verify the quality of a decision directly, but can verify how the decision was made.
This construction has a second layer that shapes every price conversation. The quality of a system becomes visible in about two years, and by then half the people in the room will work somewhere else. Price and comparability can be measured today, on paper, in front of witnesses. An organisation measures what can be measured at the moment of choice, so price carries weight out of proportion to its real importance, and arguing with procurement about it goes nowhere: they have no instrument for weighing your total-cost-of-ownership case.
For a seller this turns into a fairly hard requirement. A year from now the person inside will defend this choice without your deck in front of them, in their own words, to people who are irritated that day. Everything you hand them today gets judged on one criterion: can they repeat it in that room.
Everything below applies to commercial companies — private banks, telecom, insurance, retail, industrial holdings. The quasi-state sector is regulated separately, the mechanics there differ, and that is a separate conversation.
Your main competitor sits inside
Sellers arriving in Kazakhstan with a finished product usually build their competitive analysis around vendors and compare themselves to international and Russian alternatives. In practice the budget request most often goes to the customer's own IT team.
Large Kazakhstani companies have had in-house development for years, and in some cases it is strong. Banks, telecom operators and big holdings keep product teams that built internal systems around their own processes and know how to run them. When the business shows up with a task, it goes to those people first, and they answer that they can build the same thing for a fraction of your price.
At committee level this sounds convincing, because the internal team already sits in the budget. Their salaries are approved on a separate line, so building in-house looks nearly free from where the committee sits. Your proposal gets compared against the delta the company would pay on top of salaries it already committed. Nobody puts the full cost of the internal project into that comparison, and you lose it almost every time.
Then the internal project starts, runs for eighteen months, eats the time of a team that is also carrying operational work, and ends with a system that covers part of the original task. Two years later the cycle repeats. An external vendor wins precisely in that window, and finding it requires knowing the history: which internal projects the company started, which of them reached production, who owned them, and what happened to that person.
What works here is a conversation about splitting scope. The part your product covers out of the box goes to you. The part tied to the company's specifics stays with the internal team as integration and customisation. The IT director gets a project where their people do interesting work instead of rewriting somebody else's functionality, and at the committee they talk about joint architecture rather than giving up their own capability. That framing is defensible. «We are cheaper than building it yourselves» is not.
Six filters that fire before the functionality conversation
This is where deals stop quietly. The refusal sounds soft: we are not ready yet, let's come back next year, let's revisit after the reorganisation. Every item below removes the approver's ability to cover their own «yes».
Data residency. Personal data of Kazakhstani citizens is collected and stored inside the country. A cloud product without local hosting fails the security review no matter how much the business liked it, and the answer is hosting with a local provider. Sort it out before your first meeting with the security function, because after a negative assessment there is almost no way back into the same deal: the assessment is filed, and the next person reads it instead of re-examining the question.
Sector regime. Banks work under the National Bank's information security requirements, telecom and energy fall under the regime for critical information infrastructure. These requirements bring items into the project that never came up in the commercial conversation: certification, access restrictions, requirements for the supplier's personnel. Each one adds weeks, and they all surface at the stage where commercial terms are already agreed and changing them is expensive.
Withholding tax. Payments to a foreign supplier carry withholding tax, so the finance department sees your price as higher than a local competitor's by exactly that amount, even when the numbers in the two proposals match. Double taxation treaties change the picture, but they require residency certificates and correspondence the buyer has no motivation to handle. A local legal entity removes the question entirely, and this is one of the few reasons to register in Kazakhstan before the first deal rather than after it.
Cost classification. A perpetual licence lands in capital expenditure, a subscription in operating expenditure, and the two have different approvers, different thresholds, and often different budget years. The same product moves easily in one form and gets stuck in the other, so ask about CAPEX or OPEX in the first meeting rather than while drafting the contract. The answer changes both the structure of your proposal and who you talk to next.
Currency. The budget is approved in tenge while your price may be in dollars. An exchange rate is baked into the line item at defence, and if the tenge weakens beyond that assumption, the approved amount no longer covers the contract. The deal goes back for re-approval, and re-approval lands in the next budget cycle. A one-year contract with renewal clears far more easily than a three-year one, and the currency clause is better discussed before the autumn defence than after signature.
Payment terms. Large companies often pay no advance at all, or cap it at thirty percent, and payment is tied to a signed acceptance certificate with thirty to sixty days of deferral that stretches further in practice. For a supplier without a local entity and without working capital that is a cash gap of a full quarter. Agree separately on who signs the acceptance certificate and against what criteria: if the criteria are unwritten, signature stalls with someone who has no reason to hurry.
Who is in the room and what each can defend
Sorting participants by job title helps less than sorting them by which word they can say and what covers them when they say it.
The pilot lives on the budget calendar
In Kazakhstan a pilot is usually offered as a way to speed a deal up, and in that role it works poorly. A pilot initiated by the vendor and run for free converts rarely, because nobody inside the company is attached to it: the person who approved it risked nothing and spent nothing.
A different construction works. The pilot is paid for out of a department's operating budget, at an amount below the approval threshold, with a named sponsor, written success criteria and a deadline. Money spent creates a person inside who needs a result, and their pilot report becomes an attachment to the budget request.
That produces a hard date constraint almost nobody accounts for. The pilot has to finish by the end of August for its results to reach the September request cycle. A pilot that starts in October loses a year: it finishes in December or January, when the budget is already approved, and the next opportunity arrives eleven months later.
Signs the decision is already made
Requirements are written as versions, release numbers and product names where functions would have done. The delivery deadline is shorter than the real implementation time for systems of that class. The experience requirement is expressed as a number of projects in a specific industry, and exactly one company on the market has that number. Answers to clarification questions are formal, and the document does not change by a single line afterwards. None of the future users has met you, and every contact runs through procurement.
On its own each sign means nothing. Three of them together mean the cycle is closed. From there the choice is yours: stay in for the sake of appearing in the protocol, which sometimes buys useful introductions, or save a quarter of your team's work for the next window.
One note on challenging outcomes. The instruments exist and they sometimes work, but pricing them against a single tender is the wrong calculation. Executives in Kazakhstan rotate inside a narrow circle of organisations, and the person you fought this year turns up on a procurement committee somewhere else two years later.
A calendar instead of a funnel
From January to June, look for problem owners rather than published tenders. The job in those months is to be familiar by the time somebody sits down to write a request, and to understand their problem better than they can express it.
July, August and September decide the year. The person inside has to defend a line item in front of finance, and three things from you help: the problem described in their language, a cost estimate they can take to committee that survives scrutiny, and a reference from a company of comparable size. References in Kazakhstan get checked by phone call, not by a logo on a slide, and a case from another country weighs noticeably less than one from the same circle of twenty or thirty organisations.
October, November and December go to defence and approval. You can no longer influence the outcome, you can learn it, and that knowledge shapes your next year.
The fourth quarter opens a separate door. Unspent money gets cut from next year's budget, so short purchases below threshold appear in November and December. A supplier with a tight scope and the ability to start within two weeks can get in, and for a vendor new to the market this is the fastest route to a first signed acceptance certificate with a Kazakhstani company.
A separate multi-year goal is getting onto a framework agreement. A company inside a framework receives subsequent orders without a new procurement procedure, and that changes the economics of being on the market more than any single deal.
How many attempts you get
The number of companies in Kazakhstan buying enterprise systems at several hundred thousand dollars and up is somewhere around twenty to thirty. Multiply that by one budget cycle a year, subtract the companies where the task is already locked with an incumbent for three years, and you have the real number of attempts available to you.
The narrowness cuts both ways. Executives move between those same organisations, a CIO from a bank shows up at a telecom, a procurement director moves to a holding, and a supplier's reputation travels with them faster than case studies accumulate. How you behaved in a deal you lost gets discussed longer than the deal itself.
What to do about it tomorrow
Recalculate your sales cycle from the moment the budget request is written rather than from the tender announcement. For most products that adds six to fourteen months to the funnel and changes who you look for in January and which meetings you count as progress.
Check the six filters before you build the demo: data residency, sector regime, withholding tax, cost classification, currency, payment terms. Five of the six are solved by deal structure and legal preparation, and nearly all of them are cheaper to solve while you have promised nothing yet.
Find out what the customer's internal IT team already tried to build in this area and how it ended. That determines whether you are selling a system or a share of somebody else's project.
And rebuild your materials for a different reader. Everything you hand to the person inside gets retold in their words, in a room where you are absent, to people having a bad day. Arguments that do not survive that retelling never reach a decision.
