A subscription migration is not only a product decision. It is the single largest pricing opportunity a software vendor gets in a decade, and it only comes once.
No other pricing initiative moves as much money in one program. That is also why it is unforgiving: the size of the prize is set before the project starts, by the gap between what customers currently pay and what they would be willing to pay. Nothing in the program changes that gap. What the program decides is how much of it is actually captured, and how many customers are lost on the way.
What follows is the structure we use, drawn from more than 25 migrations. It is deliberately written as a program, not as a price decision, because that is where migrations succeed or fail.
Part 1 · What is at stake
The size of the prize is fixed. The capture rate is not.
Two things are worth separating before any planning starts. The monetary potential of a migration, the aggregate willingness to pay across the installed base, is largely predetermined. It follows from the product, the market position and the years of pricing decisions that came before. A program cannot meaningfully enlarge it.
The capture rate is entirely a function of how the program is run. And the single biggest lever on it is granularity: how differentiated the approach is across the customer base.
Customers do not have one willingness to pay. They have a distribution, and a uniform offer prices against its lower end.
A migration letter with identical conditions for every customer is fast, cheap and defensible. It is also the most expensive choice available, because it takes two risks at once: it prices far below what the upper half of the base would have accepted, and it prices above what the lower half tolerates, which turns into churn exactly where margins were already thin.
Everything that follows is machinery for avoiding that trade-off, for treating customers differently at scale, without turning the program into hundreds of individual negotiations.
Part 2 · The program
Ten work packages
A migration program breaks into ten work packages across three phases. The sequence matters: every package downstream depends on decisions taken upstream, and reopening an early decision late is what turns a nine-month program into an eighteen-month one.
01Setting the frame
The migration strategy fixes the guardrails: which customers, contracts and products are in scope, the timing and cadence of the migration, and how much force the program is allowed to apply. These are management decisions, and they belong at the start because everything else is designed against them.
The competitor benchmark is done through mystery shopping rather than desk research. Some customers will look for alternatives during a migration, that is a given. The question is whether the vendor knows what those alternatives actually quote, or is guessing.
The migration offering defines what customers are offered, excluding price at this stage: the price model, the offer structure, the packaging of the new generation. Deliberately separating this from the price conversation keeps the product logic from being distorted by individual negotiations.
The migration tactics answer the hardest question of the program: how much force. Everything between a purely voluntary migration and a fully forced one is available, and the choice drives both the take rate and the churn rate.
02Offer and price
Clustering groups the base into homogeneous cohorts along criteria such as size, cloud affinity and churn risk. This is what makes an individualized approach scalable, the alternative is either one offer for everyone or one negotiation per customer.
The customer-specific price is the core of the program. A regression model translates each customer's characteristics into an estimated willingness to pay and a churn probability, and from those two the target price follows.
The migration material is the part most often underestimated: letters, argumentation, objection handling, escalation paths. Without it, the differentiated pricing that took months to derive gets flattened in the first difficult call.
03Execution
Organisation and process come before the first letter goes out. The team needs to be set up, enabled and trained, with a process that can carry the volume. Business case and reporting quantify the expected effect and make the actual effect visible early enough to correct. And the execution itself is iterative by design, sticking to the plan on the substance, pivoting on the tactics.
Part 3 · What decides the outcome
Four success factors
01Clustering is craft, not analytics
Especially in B2B, customer rationale and decision-making are diverse. The objective of clustering is to build cohorts that are homogeneous in the way they will react to the migration, not in the way they look in the CRM. Five criteria are usually enough, and each one has to demonstrably change something about the approach. A criterion that does not change the offer, the price or the tactics is a reporting dimension, not a clustering criterion.
Cluster performance is not driven by the size of the price increase. Cohorts with the same increase perform very differently, and cohorts with very different increases perform the same.
That observation is worth sitting with, because it contradicts the instinct that drives most migration planning. The lever is fit, not level.
02Offer design and force
Just as much as it is about the price, it is about what exactly is offered and how much pressure sits behind it. Three questions decide the shape of the offer.
- What is the initial target offering?
Typically the full target setup, new product generation, cloud delivery, subscription, offered to everyone first. Fallbacks exist, but they are not on the table in the first letter.
- How much force?
A voluntary migration with strong incentives usually beats a forced one on net effect, because forced migrations buy take rate with churn. The incentives have to be real enough to move a decision, and time-bound enough to create urgency.
- Is there an unattractive alternative?
Leaving the old model available at an unchanged price guarantees the migration stalls. The old model needs a deliberately weaker economic position, the "ugly brother", or there is no reason to move.
Termination is the strongest instrument available and the one with the most collateral damage. There are several gradations between never mentioning it and terminating the existing contract with the migration letter itself. The gradation is a strategic choice, not a legal detail.
03Willingness to pay without market research
In B2B enterprise software there is no live testing and no meaningful quantitative market research. Willingness to pay has to be estimated from what the vendor already knows.
The method is a price driver analysis followed by a conjoint-like assessment of those drivers. A long list of potential drivers is built from data analysis, expert interviews and sales experience, then weighted and reduced to the handful that actually explain differences in what customers pay today. A regression model translates each customer's characteristics into an estimated willingness to pay and a churn probability.
The two curves are the output that matters. The price-revenue function gives the revenue-optimizing point, the price-churn function gives the risk-reducing one. Between them lies a band, and the vendor's positioning decides where inside that band the target price sits. Companies in consolidation mode sit high in the band; companies protecting a fragile base sit low.
04Execution beats the plan
No plan survives contact with the base. In our experience the last five to ten percentage points of the total effect are captured through iteration during rollout, adjusting argumentation, resequencing cohorts, refining objection handling as real responses come in.
That is only possible if the reporting is set up before the first letter goes out. A migration measured quarterly cannot be corrected; a migration measured weekly can.
The bottom line
The money in a migration is already there before the program starts. What the program decides is how much of it arrives, and how much of the base is still there afterwards.
Both come down to the same thing: treating customers differently, deliberately and at scale, instead of sending one letter to everyone and calling the response rate a result.
Has guided more than 25 software companies through subscription transformations, from licence and maintenance contracts to recurring models.