Grandfathering Is Dead: Legacy SaaS Plans Repriced 18% to 40%

Atlassian raised list prices on its Data Center products by 15% in February 2026. Customers sitting on legacy Advantaged plans were raised 18% to 40%. That gap is the whole story of SaaS renewal increases this year. Grandfathered pricing is being retired across the software market, and the longest-standing customers are absorbing the largest increases.

What grandfathered pricing actually was

If your renewal quote arrived with a number you did not recognise, the cause is usually not a single price rise. It is the end of an arrangement you were never contractually promised.

A discount with no contract behind it

Grandfathering holds an existing customer at an older rate while new customers pay the current one. In most standard software paper, it is a commercial policy rather than a contract right.

Read your master agreement before you argue. The price you signed is fixed for the term you signed, not for the relationship. Renewal is a fresh term at the vendor's then-current rates unless somebody negotiated a cap into the paper.

The plans being retired across 2026 are mostly of that second kind. They were policy, not paper. That is why they can be withdrawn with a month's notice and no breach.

Why vendors kept it for a decade

Grandfathering was cheap when growth came from new logos. A company adding 40% new customers a year has no need to reprice its 2018 cohort. Repricing loyal accounts creates churn, support load and awkward reference calls, all for a small gain against a large new-business number.

Three things changed that arithmetic at once. New-logo growth slowed. Seat counts stopped compounding, a shift covered in detail in the piece on seat compression and what it does to per-user pricing. And the marginal cost of serving a customer stopped being close to zero, because inference is a genuine variable cost in a way that hosting a form builder never was.

When expansion revenue gets scarce, the installed base becomes the growth plan. That is not cynicism. It is the only lever left once net revenue retention stops doing the work on its own.

The three repricings that define 2026

Three vendors published enough detail this year to see the mechanism working. They are not the only ones doing it, but they are the ones you can check.

Atlassian: 15% for list, 18% to 40% for legacy

Atlassian notified Data Center customers in mid January 2026 that new pricing would take effect on 17 February 2026. Standard list plans for Jira, Confluence and Jira Service Management Data Center rose 15%.

Customers on legacy Advantaged pricing rose 18% to 40%, varying by user tier, as those rates were brought into line with standard list. Advantaged plans generally date from before October 2019. The longer you had been a customer, the larger your increase.

The timing detail matters more than the percentage. Any renewal completed after 17 February reflected the new price. Roughly one month separated the notice from the effective date, which is shorter than most annual budget cycles can absorb.

Microsoft: the increase lands at your next renewal

Microsoft published pricing and packaging updates effective 1 July 2026, with packaging rolling out from June and completing by 1 August. The published position is explicit: existing customers remain on current pricing until renewal, then transition to the new rates.

The list changes are uneven in a way worth noticing. Microsoft 365 E5 rose from $57.00 to $60.00 per user per month, about 5%. Microsoft 365 Apps rose from $12.00 to $14.00, about 17%. Business Basic rose from $6.00 to $7.00.

The cheapest seats took the largest percentage increases. If your estate is mostly frontline or light-use licences, your blended increase is worse than the headline suite numbers suggest.

Salesforce: 6% on list, and a swapped add-on underneath

Salesforce raised list prices by an average of 6% on 1 August 2025, across Enterprise and Unlimited Editions of Sales Cloud, Service Cloud, Field Service and selected Industries Clouds. Foundations, Starter and Pro Editions were left alone.

The 6% is the part everyone quoted. The structural change sat in the same announcement. Agentforce add-ons and Agentforce 1 Editions became generally available and replaced the existing Einstein add-ons and Einstein 1 Editions.

Replacement is a stronger word than increase. Agentforce add-ons start at $125 per user per month and Agentforce 1 Editions at $550. A customer whose Einstein entitlement disappears is not choosing whether to buy AI. The choice was made by the packaging. The wider platform question is picked apart in the analysis of whether Salesforce is the platform of the agent era or its roadkill.

The third column is not comparable across rows. Atlassian's range measures the same product moving to list. Microsoft's range measures different SKUs. Salesforce's figure measures list, not what a given customer's bill did. Do not average them.

Why long-standing customers absorb the biggest increase

The pattern looks punitive. It is closer to arithmetic.

A legacy plan is defined by its distance below current list. That distance grows every year the vendor raises list and holds the legacy cohort still. After six or seven years the gap is large, and closing it in one move produces a number that looks like a penalty on loyalty.

Atlassian's own range shows this cleanly. Standard customers moved 15%, because they were already at list. Advantaged customers moved 18% to 40%, because they were not. Same product, same date, same policy, four times the range.

There is a second reason, and it is the one vendors do not say out loud. Long-tenured accounts are the least likely to leave. Switching costs compound with tenure: integrations, custom fields, trained staff, historic data. A customer who has run the same system for eight years is exactly the customer who can be repriced with the least churn risk.

My own position on this is unsentimental. If you have been on the same plan since 2019 and have never tested the market, you have been carrying a discount that the vendor treats as an option it can withdraw. Price that option at zero in your planning, because that is what it is worth.

The consolidation pressure runs underneath all of it. As point tools get absorbed into suites, the renewal conversation stops being about one product, a dynamic set out in the piece on point SaaS absorption into agent ecosystems.

What the aggregate renewal data says

Individual vendor notices tell you the mechanism. Buyer-side indices tell you the scale, with the caveats set out further down.

Vertice reported its SaaS price inflation index at 16.4% in June 2026, its highest recorded reading, calculated from more than $75 billion of global processed spend. April was 12.1% and May 14.2%. The previous peak was 14.7% in November 2025. That is a 4.3 point move in two months.

Zylo's 2026 SaaS Management Index, built on a survey of 218 IT leaders alongside analysis of more than 40 million licences, found 78% met unexpected consumption or AI charges in the trailing year. 61% cut projects or initiatives because of unplanned SaaS cost increases.

Set that against the demand side. Gartner's May 2026 revision put worldwide IT spending at $6.31 trillion for the year, growing 13.5%, with the software segment growing 15.1%. Software spend is growing faster than IT spend overall, and buyer-side price indices are running in the same band. Volume is not the only thing moving.

The renewal window is the whole negotiation

Most buyers treat the renewal date as the deadline. It is not. The deadline is the notice date written into your contract, which sits 30, 60 or 90 days earlier depending on what was negotiated.

Past that date, an auto-renewal clause has already committed you to another term. The vendor knows this. Every concession after the notice date is a favour rather than a trade, and it is priced accordingly.

So the first task in any renewal is not price discovery. It is finding two dates in your own paperwork: the renewal date and the notice deadline. Everything else follows from the distance between today and the second one.

The second task is data. Pull 12 months of licence counts, active users and feature adoption before you open the quote. Zylo's index puts average unused licences at 36% of the estate. If that is even directionally true in your organisation, the strongest response to a 20% price rise is a 20% smaller seat count, and it requires no negotiation at all. Rationalising the estate first is the argument made in the piece on SaaS sprawl and what rationalisation actually recovers.

What you can actually trade

An objection is not a trade. Vendors hear objections all quarter and have scripts for them. What moves a quote is giving the account team something they can report internally.

The useful frame is to ask what the vendor's own quarter needs. Term length, timing of signature, reference rights and expansion commitments all show up in someone's compensation plan. A flat refusal does not.

The last row is the one buyers skip. A longer notice period costs the vendor nothing today and hands you a bigger negotiating window every year afterwards. Ask for it when the commercial points are already agreed.

One tactic I would avoid: threatening to leave when you have not priced the alternative. Account teams can usually tell, and a bluff that gets called removes the only credible thing you had. Price the alternative properly, even if you have no intention of moving. The exercise is what makes the conversation real.

Where this argument is weakest

This post argues that grandfathering is ending and that loyal customers are paying for it. Here is what would have to be true for that to be wrong, and where the evidence is thinner than it looks.

The case that the increases are justified

Many of these products genuinely do more than they did in 2019, and some of the added capability carries a real marginal cost. Inference is not free, and a vendor absorbing token costs at a 2019 price is subsidising usage out of gross margin. The economics behind that are set out in the piece on inference costs and what they do to software margins.

A legacy plan priced before those costs existed was never sustainable at scale. Repricing it is not automatically extraction. Some of it is a correction that should have happened in smaller increments and did not.

What buyer-side indices cannot see

The two headline figures in this post come from Vertice and Zylo. Both companies sell software spend management. Both benefit commercially from a narrative of rising software prices. Neither publishes the underlying contract set, so the sample cannot be independently checked.

Curiously, both cite roughly $75 billion of managed or processed spend, which does not mean the samples overlap but does mean the two figures should not be read as independent confirmation of each other. Treat both as directional, from interested parties, and weight the vendor notices far more heavily.

The part nobody can settle

Whether this is a one-off reset or a permanent regime change is not knowable yet. If repricing the installed base is a single catch-up move, renewals normalise in 2027 and the current indices are a peak. If it is the new default because new-logo growth does not recover, the compounding continues.

The honest answer is that both readings fit the data available in August 2026. Anyone selling you certainty in either direction is selling something. The wider market question sits in the analysis of what the SaaS selloff was actually pricing in.

Frequently asked questions

Before your notice date

Open a spreadsheet with one row per software contract. Fill three columns: renewal date, notice deadline, and the notice deadline minus today. Sort ascending. Any row showing fewer than 90 days is this week's work, and any row where you cannot find the notice deadline is worse than a short one.

Then take the single largest contract on that list and write down what you would do if the quote came back 30% higher. Not what you would say. What you would do. If the honest answer is pay it, you have found the number that your vendor already knows and you did not.

Related reading