Andrey Eremichev

The KPI tells you what happened.
The commercial mechanics tell you why.

I work with telecom and MVNO businesses to find where acquisition, pricing, customer behaviour and channel economics actually create — or destroy — contribution margin. My starting point is usually the economics of real customer cohorts, not the reported aggregate.

ACQUISITION → USAGE → REVENUE → DIRECT COST → RETENTION → CONTRIBUTION MARGIN

01

Selected problems I've worked on

Telecom · MVNO · Fintech-adjacent

CASE 01 — ACQUISITION ECONOMICS

High gross adds. Negative contribution.

Acquisition and payback were monitored primarily at partner and channel level, and at that level of aggregation the numbers looked defensible.

What I changed
Moved the unit of analysis from the partner to the individual seller, and to the customer cohorts each seller generated. A simple model followed those cohorts through revenue, direct costs and CAC over three- and six-month windows.
What appeared
Several sellers producing substantial gross additions were generating negative contribution margin. Their apparent sales performance concealed value destruction.
What changed
Acquisition terms were redesigned around the underlying economics. Several partners unwilling to operate under the revised terms were discontinued. Overall acquisition payback improved.
Principle
When an aggregate looks acceptable, decompose it until you reach the actor or mechanism actually producing the economics.

CASE 02 — SEGMENTATION & CROSS-SELL

20% of customers. 60% of revenue.

The business was managed and reported through three separate product streams — telco, payments and transfers, and bank cards. Each had its own numbers and its own owner.

What I changed
Segmented customers by their actual product combinations rather than by product: single-product, multi-product, and users of the complete ecosystem.
What appeared
Approximately 20% of customers generated around 60% of total revenue and an even greater share of margin. That segment was invisible in every product-level report because it existed across all three.
What we asked next
We asked those customers why they used multiple services. A consistent journey appeared: a good telco experience produced trust in the brand, and trust produced exploration and adoption of everything else.
What changed
Rather than searching for an abstract umbrella proposition covering every service, the commercial logic became explicit: win through telco → establish trust → systematically cross-sell.
Principle
The organisational structure of a company is not necessarily the economically meaningful way to segment its customers.

CASE 03 — PRICING & PACKAGING

Raise the price without increasing churn.

An unlimited voice and data proposition was the cheapest comparable offer in its market. The business needed to raise the price — but a straightforward increase would have turned the proposition into same product, higher price.

What the data showed
Paid SMS and tethering already existed as separate charges, but only small percentages of customers actually used them. Their expected incremental economic cost, if bundled, was therefore limited.
What changed
SMS and tethering were included and the proposition was repackaged as a more complete all-included offer alongside the price increase.
Result
ARPU increased. There was no material increase in churn and no deterioration in gross additions attributable to the change.
Principle
Look for things that carry meaningful perceived customer value but low expected marginal economic cost.
Two more cases — where things went wrong

CASE 04 — CHANNEL INCENTIVES

A better customer offer that the channel would not sell.

Second- and third-month subscriber activity was weak. A three-month proposition priced at the equivalent of two months was introduced to improve customer value and early retention. The expectation was that it would become the majority of new sales.

What happened
Adoption approached roughly 50% in own retail. Through dealers and partners it stayed below about 20%.
What we learned
A proposition being economically better for the customer did not mean an independent dealer would prefer to sell it. Dealer economics and incentives dominated proposition quality.
Principle
Never assume intermediaries optimise for the customer's economics — or the company's. Understand what the actor actually making the decision is optimising.

CASE 05 — MVNO POSITIONING & PRODUCT MECHANICS

A transparency brand making a promise its billing could not keep.

An MVNO built around transparency and simplicity. Instead of a conventional tariff portfolio it ran a single adaptive plan: no usage, no payment, then progressively larger usage tiers up to a high maximum allowance. The proposition promised the customer would be notified before moving into the next pricing tier.

The structural problem
The service was postpaid. The underlying mechanics could not reliably guarantee that advance notification while usage was occurring. With a small base the contradiction had not yet become visible — but growth would raise the probability of customers meeting it, and the product promise risked contradicting the master brand promise of transparency itself.
Proposed change
Remove dependence on a technically fragile promise and reposition the mechanic around something the product could actually deliver: customer control of the spending boundary. The customer chooses the maximum tier while retaining the possibility of paying less when usage is lower.
Two testable framings
Bottom-up: start from zero. No use, no pay. Set your own maximum.
Ceiling: choose the maximum bill you are prepared to pay. Most months you will pay less.
Principle
A positioning claim is not merely communication. The underlying product mechanics must be capable of keeping the promise.
02

What I look at

Four views before the first hypothesis
CustomersCohorts · segments · usage · retention · contribution margin
AcquisitionChannels · sellers · acquisition source · CAC · payback
EconomicsRevenue · direct costs · contribution · cross-sell · lifetime behaviour
MarketTariff architecture · competitor propositions · packaging · positioning
Product mechanicsWhat the customer actually buys · how charging works · what happens as usage changes

I prefer to see the evidence before discussing solutions.

03

How I think

Method, not framework
01See the evidence.Customer economics, market, product mechanics and positioning — before forming the first commercial hypothesis.
02Find the mechanism.Which customers, sellers, channels, products and behaviours actually produced the reported result.
03Search for leverage.Deliberately look past the obvious intervention for the smaller change capable of a disproportionate effect.
04Test what matters.Behavioural assumptions are hypotheses. Find the cheapest credible way to let reality answer.
Human judgment, augmented by agentic search

Once the commercial mechanism is understood, I use a specialised agentic process to widen the search for possible interventions, built around one narrow question: what is the smallest plausible change capable of producing a disproportionately large change in the outcome?

It searches systematically across perception, remarkability, friction, signal, choice architecture, timing and sequence, incentives, unexamined assumptions, peak experience, and compounding effects. Each surviving candidate must state what exactly would change, why behaviour or economics might change, where the underlying pattern has worked before, and the cheapest credible test.

No scoring is produced and the agent makes no commercial decision. Its role is to expand the search space — including possibilities that experience and established industry practice might cause a human expert to overlook. Candidates then return to human judgment. AI can increase the number and diversity of hypotheses examined; it does not turn those hypotheses into facts.

04

Who is behind this

Andrey Eremichev

My strongest work starts with an existing commercial system, not a blank sheet.

Senior commercial and strategy experience primarily in mobile telecom, with additional experience in fintech and B2B banking. I examine how a business currently works, identify the mechanisms producing the observed economics, and determine what is worth changing.

I am most useful when an important commercial decision needs an independent second view, and I am comfortable disagreeing with prevailing management assumptions when the evidence points elsewhere. The objective is not to produce another strategy document. It is to improve the decision.

I distinguish between what the data demonstrates and what management believes customers, partners or employees will do. Behavioural hypotheses should be tested — and the same discipline applies to AI-generated recommendations.

05

A second opinion

Sometimes you don't need another strategy.
You need another way of looking at it.

You may already have an answer.

The numbers may look right. The explanation may make sense. The decision may even have been made.

But something still feels worth another look.

I like those situations.

Give me the commercial question, decision or assumption. I'll look at the mechanics independently and tell you what I see.

If that produces a hypothesis worth testing, even better. I'll make the test as concrete as useful — proposition, copy, page or visual if needed — so you can put it in front of reality rather than leave it in a recommendation deck.

In practice, that has ranged from a second look at pricing, customer and acquisition economics to MVNO strategy, launch decisions and commercial due diligence.

A second opinion is $1,500 fixed — but only when I believe I can add something useful.

Have something worth looking at?

Maybe the numbers look good but you're not convinced by the economics underneath them.
Maybe you're about to change a price, proposition or acquisition mechanic and want another view before you do.
Maybe the strategy makes sense on paper, but something about it doesn't quite add up.
  1. Start by telling me what you're looking at and why you're uncertain. Nothing more is needed at this point.
  2. If I think a second opinion could be useful, I'll tell you what context or evidence I'd want to see.
  3. If, after seeing it, I don't think I can add enough value to justify the work, I'll tell you that too.
  4. Only if I think there is something useful to deliver do we go ahead at $1,500 fixed.
Share what you're looking at → LinkedIn →
No calendar. No discovery call.