What Salesforce Really Costs a Nonprofit in 2026

Salesforce donates software to nonprofits, which is genuinely generous and routinely misunderstood. Organizations hear "free CRM" and budget accordingly, then discover that the software was never the expensive part. This guide sets out what you actually get, what you actually pay for, and how to phase a project so it finishes instead of stalling at 70 percent with the money gone.

What Power of Us actually gives you

If your organization is approved for the Power of Us Program, you receive:

• 10 donated Agentforce Nonprofit CRM licenses. Agentforce Nonprofit is the current name for what Salesforce called Nonprofit Cloud.
• 10 donated Agentforce Sales and Agentforce Service CRM licenses. These are the standard Sales and Service products, without the nonprofit industry function.
• Discounted pricing on licenses beyond those, and on various other products and services.

This changed in December 2025. Before then the program offered 10 free Nonprofit Success Pack licenses, and that option has been removed. New qualifying organizations are now provisioned on Agentforce Nonprofit. If you are weighing that platform decision, see Agentforce Nonprofit vs NPSP.

You do not need an existing org or a trial to apply. Approved organizations get provisioned with their licenses, and approval can take up to two weeks. Salesforce itself recommends identifying an administrator and an implementation partner before you request product, which is good advice: an unused org with donated licenses is a common and demoralising outcome.

Who qualifies, and who does not

Broadly, you must be recognised as a charitable, nonprofit, educational, or social change organization in your country, and be compliant with the Salesforce acceptable use and external-facing services policies.

The specifics that catch people out:

• In the United States, only 501(c)(3) and 501(c)(4) organizations are eligible. Other IRS subsections do not qualify. If your organization holds multiple designations, the donated licenses and discounts may only be used by the qualifying entity.
• Fiscally sponsored organizations can qualify while that sponsorship remains in effect.
• Social enterprises, B Corps, community interest companies, and cooperatives generally receive a discount but not a donation.
• Some countries are excluded from donations and discounts entirely, and in some countries Salesforce can donate but not resell.
• You will need documentation. In the United States that means your IRS tax ID and evidence of status. For sales tax exemption, the name on the certificate must match your application, the seller must be entered as Salesforce, Inc., and the certificate must be signed.

Check current eligibility rules against the Salesforce Power of Us pages before you plan a budget around them, because program terms change and this article will age.

What the donated licenses do not cover

This is the gap that surprises boards. The donation covers software access. It does not cover any of the work that turns software into a working system.

• Design. Your data model, your sharing model, your gift and program structure. Decisions, not clicks.
• Configuration and build. Objects, fields, page layouts, automation, and validation.
• Data migration. Getting constituents and gift history out of Raiser's Edge, Bloomerang, DonorPerfect, Access, or a set of spreadsheets, deduplicated and reconciled.
• Integrations. Payment processor, online giving forms, email tool, accounting package, form tools.
• Reports and dashboards, including the funder reports that are the reason you are doing this.
• Training. The staff who have to use it on Monday.
• Ongoing administration. Somebody has to own the system after go-live.

A useful mental model: the software is the building materials. Everything above is the builder.

Where license costs start again

Even with 10 donated licenses, you can end up paying, and it is better to know which triggers apply to you before you design.

• User 11 and beyond. Discounted, not donated. Budget for growth.
• Add-on products. Marketing tools, CPQ, Shield Platform Encryption, extra sandboxes, and additional storage are generally separate.
• Experience Cloud licenses, if you build an applicant portal, a grantee portal, or a volunteer portal. This is the single most common surprise in grantmaking and volunteer projects.
• Data 360, formerly Data Cloud, consumed as credits. If you plan to unify external data or ground an AI agent in documents, this becomes a running cost.
• Agentforce usage. Agent conversations are consumption based, so a customer facing agent has a cost that scales with volume.
• Third party applications from AppExchange. Many vendors offer nonprofit pricing. Ask, always.
• Storage, which nobody thinks about until a document heavy program hits the limit.

We are not publishing per-user or per-conversation prices here, because published pricing changes and a stale number in a budget is worse than no number. Get current figures from your Salesforce account executive and ask specifically what nonprofit pricing applies to each line.

Where the implementation money goes

Rough shape of a typical nonprofit implementation budget. Proportions vary, but if a quote is wildly different from this shape, ask why.

Data migration is the line most often underestimated, because the effort scales with how bad the data is, and nobody knows how bad it is until they look. If a quote gives you a firm migration price without having seen your data, treat that number as a guess.

The best money you can spend early is an assessment of your existing data. It converts the largest unknown in the project into a known.

The line items nonprofits forget

• Someone to own the system. An administrator, even part time. Without one, the org degrades within a year and you pay a consultant to fix what an internal admin would have prevented. This is the single most common cause of a failed nonprofit Salesforce investment.
• Staff time during the project. Your development director and program manager will spend real hours in workshops and testing. That is not free, and if you do not plan it, the project slips while everyone does their day job.
• Data cleanup labour. Some of it has to be done by people who know your constituents. A consultant cannot tell you whether two similar records are the same family.
• Training for turnover. Nonprofit staff turnover is high. Budget for onboarding new people every year, and record the training rather than repeating it live.
• Year two changes. A new grant with new metrics arrives. Hold a modest annual budget for change rather than treating every request as a crisis.
• The system you keep running in parallel. If you cannot switch off the old database on day one, you are paying for both. Plan the overlap period explicitly.

What AI adds to the bill

Agentforce Nonprofit ships purpose-built agents for prospect research, participant management, volunteer coverage, and donor support. They are genuinely useful and they are not free, in two different ways.

The direct cost is consumption. Agent conversations are metered, and Data 360 usage is metered in credits. A donor support agent handling volume has an ongoing cost that scales.

The indirect cost is readiness, and it is usually larger. An agent reasons over the data you already have. If you have duplicate donors, missing gift designations, and a half-populated program model, the agent will produce confident wrong answers, in front of a donor. Fixing that is data work, not AI work.

The honest sequencing for almost every nonprofit is: get the data model right, get the data clean, get the reports trustworthy, then add one agent against one measured use case. Buying agents to fix a data problem is the most expensive way to discover you had a data problem. If you want the technical detail on doing it properly, see the donor support agent blueprint.

How to phase a build around a grant cycle

Most nonprofit budgets are not annual pools of flexible cash. They are grants with restrictions and dates. Phase the work to match that reality.

Phase one should be one side of the house, live and used. Either fundraising or one program area. Not both. A working half is worth far more than two unfinished halves, and it gives you something to show a funder.

What makes a good phase one:

• It replaces a system you can then switch off, so you stop paying twice.
• It produces at least one report somebody currently builds by hand.
• It has a frozen scope, agreed in writing before build starts.
• It can go live in weeks, not quarters.

Then phase two, funded by a different cycle, adds the other side. By then your staff understand the platform, your data is cleaner, and your estimates are better because they are based on your actual org rather than an assumption.

If a funder will pay for technology as part of a program grant, scope the phase to that program. It is much easier to fund "the system that tracks and reports this program" than "a CRM".

Quote red flags

• A firm data migration price with no look at your data. That is a guess, and you will pay for the difference through a change order.
• No discovery phase. Anyone who can price your build without understanding your funders and your metrics is pricing something other than what you need.
• Migration urgency based on an NPSP deadline. There is no announced end of life for the Nonprofit Success Pack. Ask them to cite it.
• AI in phase one, before data cleanup. This is selling the exciting thing before the necessary thing.
• No named administrator in the plan, internal or contracted. Somebody must own it afterwards.
• Training as a single one hour session. Adoption is what determines whether the money was worth spending.
• No mention of reconciliation. If nobody has told you how they will prove the new numbers match the old ones, they have not planned the hardest part.

Legitimate ways to spend less

Not corner cutting. These genuinely reduce cost without reducing what you get.

1. Migrate less history. Bring recent years in full detail and older years as summary. This can remove a large share of the migration cost and most of the reconciliation pain.
2. Clean your own data first. Your staff know which records are duplicates. Consultant hours spent guessing are the most wasteful hours in the project.
3. Use standard function. Every custom object is something you maintain forever. Designations, soft credits, tributes, source codes, and outcome tracking already exist. Use them as designed.
4. Cut named licenses, not capability. Frontline staff who only submit information may need a form or a portal, not a full license.
5. Ask every vendor for nonprofit pricing. Salesforce, AppExchange vendors, and consultancies. Many have a rate they do not advertise.
6. Buy an assessment before a build. A couple of weeks of design work regularly removes more cost from the build than it adds, because it stops you building the wrong thing.
7. Invest in an internal admin. The cheapest long term Salesforce is one where somebody inside your organization can make a change without a purchase order.

Frequently asked questions