The pricing paper goes to the exec meeting with a willingness-to-pay chart, three proposed tiers and a recommended uplift of nine per cent. Nobody in the room has read a customer contract. A price increase is a change to what an existing customer pays for something they already have, which makes it a contractual event before it is a commercial one.
That ordering is the argument. What you can charge is bounded by what customers will bear. What you can actually implement this year is bounded by notice periods, renewal dates and clauses somebody signed before you joined. Most pricing work fails on the second constraint while studying the first.
How do you raise prices on existing customers?
Read the contracts before you model the price. Extract each account's renewal date, notice period, any cap on annual uplift and any most-favoured-nation clause, then group accounts by the freedom you actually have. Set the increase inside those constraints and sequence it by renewal date.
This inverts how the work usually runs. The standard sequence is research, then number, then a rollout plan that collides with reality in week three. Starting from the contracts costs a fortnight of unglamorous extraction and tells you something the survey cannot: how many accounts you are permitted to touch this financial year, and when.
The answer is often smaller than expected. A programme covering three hundred accounts frequently turns out to reach forty of them in the first twelve months, and the other two hundred and sixty roll at their existing rate whatever the board approves. Better to know that before the number is announced than after.
Your contracts have already made most of the decision
Four clauses do the work. The renewal date sets when a change can take effect. The notice period sets the deadline for telling the customer. An uplift cap limits how much you may raise, often to a published index. A most-favoured-nation clause can force your best available terms across a whole cohort the moment you concede them to one account.
The uplift cap is the one that quietly rewrites strategy. A contract tying increases to the Consumer Prices Index does not care that your costs rose faster than the index, or that the product now does three times as much. You are capped, and the only lever left is changing what the customer is buying — a repackaging exercise with a much longer lead time than the finance plan assumes.
Auto-renewal plus a ninety-day notice window means the decision deadline sits three months before the date in your CRM. Miss it and the old price locks for another full term. Check the notice mechanism as well as the period: some contracts require written notice to a registered address, and an email to the day-to-day contact does not count.
Then there is the archaeology. The order form references Schedule B, Schedule B references a price list version nobody has located since the finance system migration, and the account manager who negotiated it left in 2023. Budget for this. It is usually three weeks of somebody's time and it cannot be parallelised.
Grandfathering is a permanent decision disguised as a temporary one
Grandfathering means leaving existing customers on their previous terms after a change. It is the concession everyone reaches for when the increase gets uncomfortable, because it converts an argument into a decision that feels reversible. It is not reversible. Without an expiry date agreed at the same moment, the protection becomes permanent by default.
The mechanism is straightforward. Nobody is ever incentivised to take a protection away, the customers who hold it become your longest-tenured and best-referenced accounts, and each year the gap between the legacy rate and the current one widens enough to make correction harder than the year before. Three years on, a meaningful share of revenue sits on a rate card your billing system supports only through exceptions.
If you grandfather, write the expiry date, the transition mechanism and the name of the person who can extend it into the same paragraph as the concession itself. A protection without an end date is a second product, and you are now maintaining two.
The methods for finding a number, and how each one misleads you
Every method for arriving at a price tells you something narrower than it appears to. Surveys measure stated preference, not spending. Benchmarks measure list prices, not transaction prices. Cost models measure your floor, which no customer has ever cared about. Pick a method knowing what it cannot see.
| Method | What it actually tells you | How it fails |
|---|---|---|
| Van Westendorp Price Sensitivity Meter | The range buyers describe as too cheap or too expensive | Asks people to price something in the abstract, with nothing at stake and no budget holder in the room. Peter van Westendorp designed it in 1976 for consumer goods, not for negotiated enterprise contracts. |
| Conjoint analysis | Relative value of features and price levels under forced trade-off | Needs a large, representative sample and careful design. Across three hundred accounts you cannot field it properly, and a small sample produces precise-looking output that is mostly design artefact. |
| Competitor benchmarking | What comparable products publish as list price | Published prices are close to fiction in enterprise sales. The discount is the price, and you cannot see anyone else's. |
| Cost-plus | The floor below which the product loses money | Says nothing about willingness to pay, and anchors the internal conversation on your cost base, which is the one input the customer has no interest in. |
| Win/loss analysis | Where price killed a deal you were actually in | Only sees deals you entered, and never the buyers who filtered you out on the pricing page. Sales records "price" as the loss reason because it is the least awkward field to fill in. |
| Price experiment | The causal effect of a price on conversion | Not available to you here. Charging two contracted customers different amounts for the same thing is a commercial problem and, under a most-favoured-nation clause, a contractual one. |
The last row is worth sitting with, because it removes the tool product managers reach for by reflex. The reasoning in the guide to running experiments without the traffic to prove anything applies with an extra constraint: even where you have the volume, you may not have the right to randomise. What remains is qualitative, and interviews structured to produce decisions rather than quotes do more here than any survey instrument.
The discount is the price. The rate card is marketing.
Run a price increase in this order
The sequence matters more than the number. Most programmes are designed as a communications exercise and then discover the contractual constraints during rollout, at which point the announcement has already gone out and the concessions get made under time pressure. Doing the unglamorous work first removes that.
- Pull the contracts rather than the customer relationship management (CRM) record. Extract renewal date, notice period, uplift cap and any most-favoured-nation clause into one sheet, and record where the source document could not be found.
- Segment by contractual freedom, not by account size. The relevant groups are "can change this year", "capped", "locked until the following renewal" and "unknown".
- Decide the value metric before the number. Moving from per-seat to per-transaction is a different negotiation from raising a per-seat rate, and it has a longer lead time.
- Set the grandfathering rule and its expiry in the same sentence, or do not grandfather.
- Name the churn figure that would pause the programme and the person with authority to pause it. Do this before any customer is told, while it is still a number rather than a specific account.
- Sequence by renewal date, taking the smallest exposure first, so you learn on accounts you can afford to lose.
- Brief account managers with the real reasoning and a scripted answer to "what if they say no". The answer they invent under pressure becomes your actual pricing policy.
Step five is the one that gets skipped, and it is the only step that protects you from your own commercial team. Watch net revenue retention and downgrade requests rather than headline churn, because the first response to an increase is usually a reduction in seats rather than a cancellation — a pattern the guide to numbers that quietly mislead covers in more general terms.
Regulated products make a price increase an evidenced decision
In United Kingdom retail financial services, pricing is a documented exercise rather than a commercial judgement. The Financial Conduct Authority (FCA) Consumer Duty includes a price and value outcome, set out in PRIN 2A.4 of the FCA Handbook, requiring firms to carry out and regularly review a value assessment showing that the amount paid is reasonable relative to the benefits.
Two details bear directly on the work above. First, the rule applies to existing and closed products, not only to new ones, so a legacy cohort is inside scope rather than outside it. Second, where a pricing structure means different groups pay different prices, firms are expected to consider whether the product delivers fair value for each group. Grandfathering creates exactly that structure.
The FCA has also said that fair value assessments work best when they reflect the analysis done at the time decisions were made, including how risks of poor value were identified. As of 2026 that remains the position. You cannot reconstruct the reasoning afterwards from a spreadsheet, which means the assessment has to be produced alongside the pricing decision rather than in the quarter that follows it.
The operational failure is the familiar one. The commercial decision gets made in a Tuesday meeting with no minutes, the fair value assessment needs discussion at a governance forum that sits monthly, and the notice deadline for the first cohort falls between the two. If you sell to institutions rather than retail customers the Duty does not apply to you, and the discipline of writing down the reasoning at the moment of decision is still worth borrowing — the same argument as working with a compliance function that can veto a release.
The terms worth agreeing before the meeting
Pricing arguments run long because participants use the same words for different things. Sales means list price, finance means realised price, product means the metric being charged for. Agreeing six definitions at the start of the paper saves an hour every time the paper is discussed, and all six belong in writing.
- Value metric — the unit you charge by, such as a seat, a transaction, an account or a gigabyte; changing it is a larger decision than changing the number attached to it.
- Grandfathering — leaving existing customers on their previous terms after a change, with or without a stated end date.
- Most-favoured-nation (MFN) clause — a commitment that no comparable customer receives better terms, which can propagate a single concession across an entire cohort.
- Uplift cap — a contractual ceiling on annual increases, frequently tied to a published index such as the Consumer Prices Index (CPI).
- Notice period — the window before renewal within which a change must be communicated for it to take effect at that renewal.
- Net revenue retention (NRR) — revenue from an existing cohort after expansion, downgrade and churn, measured against the same cohort a year earlier.
Frequently asked questions
How much can I raise prices without losing customers?
There is no general figure, and any source quoting one is guessing about your market. The answer turns on switching cost, contract terms and how visible your price is inside the customer's budget. What you can control is deciding in advance which accounts you are willing to lose, naming the churn number that would stop the programme, and sequencing renewals so you learn before you are exposed.
Should I grandfather existing customers on their current price?
Sometimes, but only with an expiry date agreed in the same decision. Open-ended grandfathering converts a temporary concession into a permanent second product carrying its own billing exceptions, reporting problems and roadmap questions. If you grandfather, state when the protection ends, what happens to those accounts at that point, and who holds the authority to extend it.
How much notice do I need to give before a price increase?
Your contracts tell you, and the answer usually differs by account. Terms commonly sit between thirty and ninety days before renewal, and missing the window normally means the existing price rolls for another full term. Check the notice mechanism as well as the period, because some agreements require written notice to a registered address rather than an email to your usual contact.
Can I charge different customers different prices for the same product?
Commercially yes, and most enterprise software already does through discounting. The constraints are contractual and regulatory rather than ethical. A most-favoured-nation clause can force your best terms across a cohort, and in United Kingdom retail financial services the price and value rules expect fair value to be assessed for each customer group, including where different groups pay different amounts.
Should I raise prices for new customers only?
It is the lowest-risk option and it postpones the problem rather than solving it. A new-only increase widens the gap between cohorts every year, which makes the eventual correction larger and harder to justify to the customers who have been loyal longest. It also distorts reporting, because average revenue per account moves for reasons unconnected to any decision you made.
How do I know whether the price increase worked?
Not from an experiment, because you cannot randomly assign contracted customers to different prices. Track net revenue retention against a threshold you set beforehand, watch downgrade requests and renewal-cycle length rather than headline cancellation, and allow at least two full renewal cycles. Revenue rises on the day of the change and the cost of it appears much later.
Once the first cohort has moved, you are running two populations paying different amounts for the same product, and every subsequent roadmap decision acquires a second question: does this go to the legacy tier as well. That question is harder than the pricing one, because the honest answer is often no, and no is what turns a temporary rate into a visibly worse product. The frameworks people reach for to settle it each mislead in their own way.