Custom nonprofit software in Baltimore: when to build instead of renting by the record

A sector that is bigger than it looks, and smaller than it looks

Both things are true at once, and holding them together is the whole argument.

At the state level, this is a serious industry. Maryland nonprofits employed 288,231 people in 2022 — 13.4 percent of all private employment in the state, against 9.9 percent nationally, which puts Maryland sixteenth in the country. One in every seven or eight people working for a private employer in Maryland works for a nonprofit. Narrow the lens to Baltimore City and the figure stops being a statistic and becomes the defining fact about the local economy: 95,161 nonprofit jobs, or 35.5 percent of all private employment in the city. More than a third. That is driven heavily by Johns Hopkins — the university, the hospital and the health system together — so it is not evidence that small nonprofits employ a third of Baltimore. It is evidence that in this city, the nonprofit form is not a sideline. It is the main event.

A note on that number, since we have watched it drift. The figures still circulating in Maryland — around 280,000 workers and just under 13 percent — are 2017 data, usually printed without a year attached, including on the sector's own advocacy pages. The current ones come from the Bureau of Labor Statistics' research series that flags 501(c)(3) employers inside its quarterly census, published in 2025 for reference year 2022. Cite the current number with its year attached; being five years stale on your own sector's headline statistic is the kind of thing a funder notices. Worth knowing too that the long-running research program behind the older figures was for decades a Johns Hopkins institution here in Baltimore, and moved to George Mason after Lester Salamon's death — so the local reflex to cite Hopkins for it no longer quite works.

Now the other half of the picture, and it inverts completely. We pulled the IRS Business Master File — the July 2026 extract — and counted. Maryland has 37,637 registered 501(c)(3) organizations. Of those, just 7,848, or about one in five, file a full Form 990 or 990-EZ. Over half — 19,627 — file only the 990-N postcard, which is what you file when your gross receipts are under $50,000 and which contains no financial information whatsoever. The rest are churches and others with no filing requirement at all.

So four out of five "Maryland nonprofits" are, financially speaking, invisible, and that is not a data problem — it is a description of how small they are. Among the roughly 7,400 Maryland organizations that do report financials, 56 percent run on under $500,000 a year and 69 percent on under $1M. Baltimore City skews somewhat larger, because the anchor institutions are here: of its 4,710 registered charities only 1,247 file a full return, and of those, 44 percent are under $500,000 and 57 percent under $1M. Meanwhile the top ten organizations in the state account for something like 40 percent of all 501(c)(3) revenue, and the top five hundred for around 90 percent.

So the sector is a handful of giants and an extremely long tail, and the tail is who this article is for. In Baltimore City that tail leans in recognizable directions. Measured against the state as a whole, the city has roughly twice the expected share of community-improvement and capacity-building organizations, and twice the share of housing and shelter organizations; human services is the largest single category outright, and arts and culture is over-represented too. These are the neighborhood organizations, the workforce programs, the CDCs, the arts nonprofits — five staff, eleven grants, a board that meets quarterly, and a development director who is also the grant writer.

We are going to be blunt about what that implies, because it cuts against our own commercial interest and it is better said early: for most organizations in that tail, the right amount of custom software is none. A $75-a-month subscription that mostly works is a good deal for a five-person team. Keep reading anyway, because the part where it stops being a good deal is specific and recognizable, and it is not where most people expect.

Five vendors, five different meters

Let us be fair to the tools first, because they are decent software and most organizations should be on one of them. Little Green Light is genuinely excellent and honestly priced. Bloomerang is a good product with a real community around it. DonorPerfect has been serving small shops competently for decades. Neon is capable and its unlimited-records position is a real differentiator. Nobody should commission custom software to do what these already do well.

The difficulty is not the products. It is that you cannot compare them, and that appears to be structural rather than accidental. When we went through the category in July 2026 looking for a comparable rate, this is what we found.

Read down the middle column and the problem announces itself. These are not different prices for the same thing; they are different things being counted. An organization with 40,000 lapsed records and $200,000 of revenue is cheap under Neon's model and expensive under Bloomerang's. An organization with 900 records, $4M of government revenue and fourteen staff is the reverse. There is no unit that lets you lay two quotes side by side, and since six of the eight will not quote you without a call, you cannot easily get two quotes anyway.

Bloomerang is worth dwelling on, because it is the clearest case. Its pricing page leads with "One intelligent platform. Unlimited users," which is a genuinely generous position — most categories charge by the seat, and this one has largely stopped. It then tells you, in prose, that it offers "a predictable, constituent-based pricing model" so your budget will not spike as you grow. What it does not tell you is the rate. We pulled the full page source — all 851 kilobytes of it, scripts included — and there are exactly three dollar figures in the entire document: $125 for the CRM, $40 for Fundraising, $119 for Volunteer, each described as a starting point, with the actual bundle routed to Contact Sales. No slider, no band table, no ladder. The meter is named; the rate is not.

DonorPerfect is the same shape: "starting at $99 a month," three tiers called Core, Plus and Pro, and a Get pricing button under each. Its Canadian quote form is more revealing than its American pricing page — it asks you to select your number of records from bands running up to 200,001 and over, which confirms what drives the quote without attaching a number to any band. Bonterra goes furthest: its Apricot case management page publishes a complete feature comparison matrix across Essentials, Pro and Enterprise, with dozens of rows telling you exactly which capability sits in which tier, and not one dollar sign anywhere on the page. That is a lot of work to be that specific about everything except the price.

Against all that, Little Green Light simply prints the whole thing.

Above 50,000 records it is $15 a month for each additional block of ten thousand, the account ceiling is 200,000 constituents, users are unlimited and free — "Nothing. Nada. Zilch. Zero," as their own page puts it — every feature is included at every level, there is no contract, and you can export your data on cancellation with no fee. Annual prepayment takes ten percent off. It is possible to publish a price in this industry. One company does it.

There is a second problem with metering on records, which is that a nonprofit cannot prune. In most software categories, if the meter counts rows you can delete rows. Here you cannot, and for good reasons: you need gift history for stewardship, for the 990, for the auditor, and for tracking which restricted gift paid for what. Lapsed donors are not dead weight, they are the reactivation list — reactivating a lapsed donor is materially cheaper than acquiring a new one, which is exactly why you keep them. So the count only ever goes up. The meter is attached to the size of the community you exist to build, and it ratchets.

The line that actually costs you money

Now the part that matters more than everything above, and that we almost never see modeled: for most fundraising operations the subscription is not the largest software cost. The processing is.

Credit to Bloomerang here, because it publishes its processing rates in full and with a date on them — 12 June 2026 — which is more transparency than most of the category manages on any dimension. Through Bloomerang Fundraising, Visa, Mastercard and Discover run 3.95% plus $0.30 per transaction; American Express is 4.95%; ACH is 1.95% plus $0.95; peer-to-peer adds another one percent on top. Through Bloomerang CRM the card rate is 3.2% plus $0.30 and ACH is 1.8%. Little Green Light sits at the other end of the category: it takes no platform cut whatsoever, and you pay your processor directly, starting around 2.2% plus $0.30.

Put those side by side across realistic giving levels. This is just the percentage, setting aside the flat per-transaction fee, which both routes charge.

At half a million dollars of online giving, the spread is $8,750 a year — more than five times what Little Green Light charges at the very top of its published ladder. An organization that spent three months choosing between two subscriptions on a $30-a-month difference, and then never looked at the processing line, optimized the wrong number by an order of magnitude.

Two caveats, because the comparison is not quite like for like and we would rather say so than let it stand unqualified. Bloomerang Fundraising is a substantially larger product than a bare CRM plus a payment processor — that 3.95% is buying you forms, events, auctions, text fundraising and peer-to-peer campaigns that you would otherwise assemble yourself, and for plenty of organizations that bundle is worth the money. And a fully assembled alternative has its own costs, in tooling and in somebody's time. The point is not that one vendor is fleecing you. The point is that this is the largest line, it scales directly with your success at the thing you exist to do, and it belongs in the model.

The escape hatch is narrower than it looks, too. Bringing your own processor does not remove the platform's cut: Bloomerang applies a platform fee of 1.95% on Fundraising or 1% on CRM on top of whatever your outside processor charges you, and Neon charges from $50 a month for the privilege of using an external processor at all. That is a reasonable commercial position — the platform is doing real work either way — but it does mean the percentage is a property of the platform rather than of the payment rails, and switching rails will not shake it off.

And this is a category norm rather than one vendor's quirk. Blackbaud, which publishes nothing at all about what its software costs, publishes a complete payment rate card: 2.99% plus $0.30 on standard cards, 3.5% on American Express, 1% on ACH capped at $5 — and then, separately, a platform fee on funds raised through its online forms, at 1.5% for both eTapestry and Raiser's Edge NXT and 2% for its older web products. So the one number Blackbaud will tell you in advance is the one calculated on your donations. Read the category as a whole and the pattern is hard to miss: the subscription is negotiable, private and quoted per deal, while the percentage of your fundraising is fixed, public and applies to everybody.

The six hours nobody counts

Here is the other number that never makes it into a software comparison, and it is the one we would put at the center of any honest build-or-buy conversation.

The Center for Effective Philanthropy has been measuring what grant reporting costs grantees for years, across a large multi-funder dataset. Its finding: the median time a grantee spends on a single funder's reporting and evaluation requirements fell from about eight hours before the pandemic to about six hours in 2021 and 2022. That decline is real progress and the funders who drove it deserve the credit. But read the unit carefully. Six hours is the median per funder. Nobody has one funder.

Two of those inputs are ours rather than CEP's, and we want to be explicit about that: the assumption of two reports per grant per year, which matches the common interim-plus-final pattern but will be low for anything on a quarterly cycle, and the $30 hourly loaded cost. We took the $30 deliberately from a vendor arguing against custom builds — it is the figure they use to price the burden of maintaining your own system — and it seemed only fair to apply the same rate to the burden their alternative leaves in place.

At twenty active grants that is 240 hours a year: six full working weeks, spent transcribing information your organization already holds into somebody else's template. It is more than four times what the software subscription costs at that scale. And CEP's wider findings make the picture worse rather than better — among the grantmaking institutions it surveyed, fewer than half share back what they learn from the reports they collect, and only about half use any digital technology to manage the process at all. Two-thirds will now accept some alternative form of reporting, including, at sixty-nine percent, a report you already wrote for a different funder. Many grantees do not know that, and never ask.

This is the single clearest software opportunity in the sector, and it is nearly untouched. Not because it is technically hard — it is not — but because it sits in the gap between the products. The donor CRM knows the money came in. The accounting system knows what was spent. Neither knows that Funder A wants participant counts by ZIP code quarterly in a spreadsheet, Funder B wants a narrative against three named outcomes twice a year in a portal, and the city wants demographic breakdowns on a schedule of its own. That translation layer is where the six hours go, every time, forever.

The part no national platform models

Everything above is true anywhere in the country. This section is the reason a Baltimore organization's software problem is not the same as a Denver organization's, and it is where we would spend the money.

Start with restricted funds, because it is the foundation and because the software category is remarkably silent about it. Under FASB's ASU 2016-14, a nonprofit carries net assets in two classes — with donor restrictions and without — and recognizes a release from restriction in the period the restriction is actually satisfied. That sounds like an accounting detail and it is really a data-model requirement: satisfying a purpose restriction means matching specific expenditures against specific gifts, gift by gift, not netting a revenue total against an expense total at year end. The same standard requires you to publish an analysis of expenses by both natural and functional classification, and a liquidity disclosure explaining what you actually have available to spend in the next twelve months — a question you cannot answer without knowing which of your assets are spoken for.

ASU 2018-08 then adds the harder test, and it is the one that catches people. A grant is conditional if it carries both a barrier to overcome and a right of return — and a conditional grant is not revenue until the barrier is met. Money already sitting in your bank account is a refundable advance, which is to say a liability. Barriers include measurable performance requirements, matching requirements, and stipulations limiting your discretion over how the work is done, which describes a great many government and foundation awards. One thing to flag because it is still repeated in guidance everywhere: the old rule that a condition could be ignored when the chance of failing it was remote was superseded in 2018. It no longer applies, and ambiguous stipulations are now presumed conditional.

Sit that beside the products and the gap is obvious. A donor CRM records that a gift arrived and what it was for. It does not carry a per-gift restriction ledger, it does not model barriers, and it does not compute releases. Most small organizations bridge the gap with class tracking in QuickBooks and a spreadsheet that reconciles the two, maintained by one person, and it works right up until that person leaves.

Two practical notes on that bridge, since this is the most common setup in Baltimore and the details are worth having. Class and location tracking — the feature the entire workaround depends on — starts at the QuickBooks Online Plus tier, currently $115 a month at list, and it is capped at forty classes and locations combined; unlimited requires Advanced at $275. Neither Simple Start nor Essentials has it at all, which catches people who budgeted for the cheap tier. But before paying any of that, check TechSoup: Intuit donates a one-year, five-user QuickBooks Online Plus subscription for an $80 administrative fee, once in an organization's lifetime. If you are a small Baltimore nonprofit currently paying list price for Plus, that is most of a year's accounting software recovered for the cost of an afternoon.

And if the fund-accounting layer is genuinely the problem rather than the reporting on top of it, there is a real published option before anyone builds anything: Aplos publishes $79 and $129 a month for its lower tiers and includes fund-level balance sheets and income statements at every level. The serious fund-accounting products above it — Blackbaud's Financial Edge NXT, Sage Intacct, MIP — publish nothing at all; MIP does not have a pricing page in the first place. As ever in this category, the moment the product gets more capable, the price disappears.

Then the Maryland layer, which moved three weeks ago and which almost nothing published online has caught up with yet. Chapter 394 of the 2026 Laws of Maryland — Senate Bill 354 — was approved by the Governor on 12 May 2026 and took effect on 1 July 2026. It raised the threshold for a CPA review from $300,000 to $400,000, and the threshold for an independent CPA audit from $750,000 to $1,000,000. If you are checking this against another source, note a trap in the drafting: the bill as introduced set the review figure at $500,000 and was amended down to $400,000 before passage, so the enrolled text carries both figures and anyone reading the introduced version will confidently tell you $500,000. It is $400,000.

What makes this more than a trivia item is the denominator. Maryland computes those thresholds on gross income from charitable contributions, assembled from named lines of your 990 — and government grants are expressly excluded. Which produces a genuinely strange situation for Baltimore, where so much of the human-services sector runs on city, state and federal pass-through money.

The Maryland audit trigger and the federal Single Audit trigger are both $1,000,000, and they are computed on completely different money.

The federal one — raised from $750,000 by the April 2024 overhaul of the Uniform Guidance, effective for fiscal years starting on or after 1 October 2024 — counts federal awards expended. Maryland's counts charitable contributions received, minus the government money. So a Baltimore organization with $3M of revenue, nearly all of it public funding, can owe Maryland a $50 registration fee and no CPA report at all, while simultaneously tripping a federal Single Audit. And a donor-funded arts organization with a fraction of that revenue can be the other way around. Any system that claims to tell you where you stand has to compute both numbers off the same ledger, on two incompatible bases. We have not seen an off-the-shelf product that does, and we would be glad to be shown one.

The same 2024 overhaul moved several other figures worth having right, since they change what your system needs to track: the de minimis indirect cost rate went from 10% to up to 15% of modified total direct costs, the equipment capitalization threshold went from $5,000 to $10,000, and the portion of each subaward included in the modified-total-direct-cost base went from the first $25,000 to the first $50,000. That de minimis figure is the one to watch out for — the regulation sets 15% as a ceiling you elect within, and a great deal of published guidance describes it as though it were a flat rate. There is also a further rewrite of the Uniform Guidance out in proposed form, with comments closed on 13 July, so this is a live area and worth rechecking before you rely on it.

The registration calendar itself is straightforward and no platform tracks it: annual renewal is due within six months of your fiscal year end, with an automatic extension that lands on the fifteenth day of the eleventh month — deliberately aligned with the extended federal 990 deadline. The fee runs from nothing under $25,000 of contributions up to $300 above $500,001, and lateness costs $25 for every month or part of a month.

Finally, the local funding layer, and this is where the argument stops being abstract. The Baltimore Children and Youth Fund is written into the City Charter at Article I, § 13 — approved by voters as Question E in November 2016 with 82 percent in favor — which requires an annual appropriation of at least three cents per $100 of assessed property value. Worth stating precisely, because it is routinely described as a dedicated three-cent property tax: it is not a separate levy, it is a floor on an appropriation pegged to the assessable base. In practice that has meant $14.2M in fiscal 2024, $15.2M in 2025, $16.2M in 2026, and a proposed $17.0M for 2027.

Now look at what it asks of a grantee. Monthly Financial Expense Reports are due on the 25th of the following month — a fixed date, weekends and holidays included — on a cash basis, itemized, with invoices, receipts and proof of payment attached. A month in which you spent nothing still requires a filing. Reports submitted after the month closes are refused outright and need a formal late appeal. On top of that sit quarterly programming reports on a mid-December, March, June and September cycle that does not line up with the city's own fiscal year, monthly staffing reports to another city office by the fifth business day, an annual site visit, and five-year record retention. And the reporting is not merely administrative: reporting zero expenses blocks your first disbursement, and spending under half your award blocks the second. Fail at the paperwork and the money stops.

You can guess what that costs a five-person organization, but you do not have to. In a November 2025 report from the Baltimore City Inspector General, BCYF's own president told investigators that for some grantees those monthly expense reports consume up to fifty hours a month.

Fifty hours a month is a full-time person, three-quarters gone, to report on a grant of $50,000 a year.

We want to be careful about what that number means. It is the top of a range, not a median, and it will be describing the organizations with the least administrative infrastructure — which is rather the point, because those are the same organizations the fund exists to reach. Nothing about it suggests bad faith on anybody's part; a public fund distributing tax revenue has a real obligation to account for it, and BCYF has been under genuine oversight pressure to tighten exactly these controls. But it is a precise, locally sourced measurement of the thing this whole article is about. The reporting burden is not a rounding error against the software budget. For a grassroots Baltimore grantee it can exceed the value of the grant's own program time, and no donor CRM on the market touches it.

The exit question, and when to ask it

Ask before you sign, not when you leave, and ask about specifics rather than about "export" in the abstract. The right question is whether gift history, soft credits, attachments and custom fields come out — those are the four that reliably do not.

The terms vary more than people expect, and the good end is genuinely good. Little Green Light states plainly that you can export your data on cancellation with no penalty or fee. Bloomerang is better still, and deserves real credit here: its export runs self-serve from the settings screen, produces a ZIP of CSVs, and its documentation publishes both an inclusion list and — far rarer — an explicit exclusion list, so you can see in advance that engagement scores, split-gift detail and membership-module data will not be coming with you. Publishing what you don't get is the single most useful thing a vendor in this category does, and almost nobody else does it.

Blackbaud is the one to read slowly, and it is the incumbent at the larger end of Baltimore's sector. Its master agreement requires you to ask for a copy of your data at least thirty days before your termination date, and leaves it to Blackbaud's option whether you receive guidance on extracting it or an actual backup copy. After termination the data is destroyed, and its cancellation policy warns that access ends on the effective date. For Education Edge it states outright that once access is off, it cannot pull the information. Its knowledgebase notes that Raiser's Edge data cannot be exported into a single file because it lives across linked tables, and media and attachments are separate processes again. Most pointedly, eTapestry's own help documentation describes a migration-ready "Final Data Export" as something your organization can purchase — the free path being to build two custom exports yourself. None of this is improper; it describes old, genuinely complicated software. But it means the moment you decide to leave is the worst possible moment to start finding out what leaves with you.

Neon is the one we would want a written answer from. Its terms of service run to the usual length and confirm that you own your data — and then contain no data-export, data-return or post-termination retrieval provision at all. The words "extract," "migrate," "retrieve" and "download" do not appear; the only uses of "export" in the document refer to export-control law. What the terms do establish is Neon's right to delete customer data after termination, and a perpetual, irrevocable license to aggregated data that survives the agreement. Its public API has no bulk-export endpoint, so retrieval means paging through records object by object. There may well be a perfectly good export tool behind the login — we could not check, because the help center requires an account. That is exactly the sort of thing to establish before signing rather than after.

The general lesson is the one we drew in our guide for property managers, where data export and API access were themselves paid upgrades, and again in our piece on legal software, where several platforms will not carry out the very records the state requires firms to keep. Nonprofit software is better behaved than either of those. It is still worth an hour and a written answer before signing.

What custom nonprofit software costs in Baltimore in 2026

The numbers you will find online for this are large and mostly unhelpful. Development shops will tell you a custom nonprofit platform starts somewhere north of $60,000; one agency page we read while researching this article offers the example of a $200,000 build being cheaper than $50,000 a year of SaaS, which is arithmetic that only works because the build side carries no maintenance, no hosting and no staff time. On the other side, a SaaS vendor asserts that custom systems cost 30 to 40 percent of the build annually to maintain — roughly double the usual rule of thumb, and inflated in precisely the direction that sells subscriptions. Both numbers describe a way of working rather than a property of the software.

A traditional agency price is high for reasons that have little to do with your organization. You are renting a team — a project manager, a couple of engineers, a designer, an account lead — each holding a slice of the context, each billing by the hour, plus the overhead of every handoff between them. We think that model is broken at this scale, so we do not use it. One senior builder with AI in the loop across the whole job, scope frozen before we start, and a fixed price attached to it: a real number you agree to before any code is written. For most organizations it lands like this.

Every one of those is a fixed price against a fixed timeline — half to start, the balance when it ships — and you own every line of the code, the keys and the accounts at the end. The full breakdown of what is and is not included lives on the pricing page. For most organizations the right first purchase is the Custom App, because what a nonprofit needs is almost never a replacement for its CRM — it is the one or two things the CRM was never designed to do. The Donation Portal is the same engine as our online store package pointed at a different job: giving, event registration and membership under your own brand, on your own domain, running through your own payment processor at your own negotiated rate rather than through somebody's platform percentage. On the arithmetic in the table further up, that last detail alone can cover the build.

What we'd actually build — and the case for not replacing anything

Here is the conclusion that separates this article from every agency page on the subject, and it goes against our own commercial interest, so take it in that spirit.

For an organization under roughly $2M a year, replacing your donor CRM with a custom build is the wrong call, and we will tell you so on a call. Little Green Light at the top of its ladder is under $5,000 across three years. No custom donor database, plus the maintenance it will need, comes close to competing with that, and anyone who tells you otherwise is quoting you a number they have loaded. The four break-even calculations we could find in this category are each authored by a party with something to sell, and each one is tilted toward its author's product — including the most thorough of them, which prices staff workaround time honestly on the option it is arguing against and assigns zero workaround hours to the option it sells.

What does pay is the layer around the CRM, because it replaces staff hours rather than records. That is a much smaller piece of software and a much larger saving.

The first build, nearly always, is the grant and restricted-fund tracker — the thing currently living in a spreadsheet. Each award, its restriction, its budget lines, its spend-down against those lines, its barrier conditions, its report calendar, and the release entries that follow when the restriction is satisfied. That object does not exist in a donor CRM and it does not exist in QuickBooks; it exists in the head of whoever built the spreadsheet, which is the actual risk. Build it once and the audit, the 990, the liquidity disclosure and the two different thousand-dollar thresholds all fall out of the same ledger.

Second is per-funder outcome reporting, which is the six hours. One data model underneath, many report shapes on top — the ZIP-code spreadsheet, the narrative against named outcomes, the city's demographic breakdown — generated rather than transcribed. This is unglamorous and it is the highest-return software a grant-funded organization can own.

Third, for human-services organizations especially, is program and participant data, which has nowhere good to live. A donor CRM was designed to hold donors; putting participants in it is a compromise everyone makes and nobody likes, and the case management products that do it properly are the ones that would not tell us their prices. Fourth is the donation and event portal, for the processing reasons above. Around all of it sit the integrations that keep it honest: your CRM's API, your accounting system, your payment processor, and clean exports for whoever audits you.

The principle throughout is the one we described in the piece on internal tools teams actually adopt: software that mirrors the real workflow gets used, and software that asks people to change how they work gets quietly worked around. If you want to see what these budgets buy, we keep five real apps running live in the browser on the demos page.

Build or buy: how to tell which side you're on

You do not have to take anyone's positioning on faith, ours included. The useful test is not "custom or SaaS" in the abstract — it is a few plain questions about your own organization, and the answers usually point clearly one way.

• Are you paying people to bridge the software? If a real slice of your development director's month goes to assembling funder reports by hand or maintaining the restricted-fund spreadsheet, you are already paying for custom software in salary and getting nothing you own for it.
• What is the true annual number? Add the subscription, the processing percentage on everything you raise online, the modules, and the implementation. The processing line is usually the biggest and is almost always left out.
• Is the gap a whole platform, or two screens? If the CRM is broadly right and two things are missing, build the two things. Replacement is rarely the answer under about $2M, and we will say so.
• What happens on the way out? Ask, in writing, before you sign: does gift history come with us, do soft credits, do attachments, do custom fields?

If those questions mostly land on "the CRM is fine and we barely think about it," keep it — that is the right and cheaper answer. If they mostly land on "we have built a whole shadow operation in spreadsheets to make this work," that shadow operation is the specification for the thing worth building. For how the timeline collapsed from the industry-standard months down to weeks, we wrote separately on how long it really takes to build a custom app, and the broader decision framework sits in custom versus SaaS versus no-code. This discussion also sits inside a local one — our guide to custom software development in Baltimore covers the trades, the Port logistics economy, healthcare and the rest of the city's small organizations alongside the sector.

Built in Baltimore, yours to keep

We are a small studio of ex-founders based here in Baltimore, and we build custom software for small and growing organizations — custom apps, internal tools, portals and operations systems — roughly the way we wish someone had built it for us when we were running our own companies. Fixed price, fixed timeline, direct with the builders, and fully yours at the end. We are not accountants or auditors, and nothing here is accounting or legal advice; everything above about FASB, the Uniform Guidance and the Maryland Solicitations Act is offered as engineering context and should be checked against the sources themselves, which is a habit we would recommend generally given how much of what is published about them is out of date. The Maryland thresholds in particular changed on 1 July, and most of the internet has not noticed yet.

If you run an organization that has outgrown the software around it — funder reports assembled by hand, restricted funds tracked in a spreadsheet only one person understands, program data with nowhere to live — the way to find out what it would take is a free thirty-minute call. Bring the spreadsheet. We will tell you honestly what we would build, what you should keep renting, how fast it could ship, and the fixed price that goes with it.

Common questions from Baltimore nonprofits

How much does custom nonprofit software cost in Baltimore?

Our fixed prices are the anchor. A one-week Prototype Sprint — one core flow, such as a restricted-fund dashboard or a grant report generator, clickable and deployed — is $3,500. A branded donation, event registration and membership portal on your own domain starts at $6,000. A custom internal tool, or a full operations system covering programs, grants, restricted funds, outcomes and the compliance calendar, starts at $12,000. Each is a fixed price agreed before any code is written, and you own the code, the keys and the accounts at the end. Most nonprofit builds land in the $6k–$30k range. Treat published numbers in this category with suspicion, including the ones arguing against building: the four break-even calculations we could find are all authored by parties with something to sell, and each one loads the assumptions toward its own product.

How much does nonprofit CRM software actually cost per year?

Almost nobody in the category will tell you. Of the platforms we checked in July 2026, Little Green Light publishes a complete ladder — $45 a month up to 2,500 constituent records, rising in bands to $135 at 50,000, then $15 for each additional 10,000, with unlimited users and a 200,000-record ceiling. Instrumentl publishes $299, $499 and $999 a month for grant management. Salesforce publishes Nonprofit Cloud at $60 and $100 per user per month on an annual contract. Everyone else publishes a floor and a form. Bloomerang shows $125 a month for its CRM, $40 for Fundraising and $119 for Volunteer, and routes the actual bundle to Contact Sales — there is not a single record band anywhere in its pricing page. DonorPerfect publishes "starting at $99 a month" with Core, Plus and Pro all saying Get pricing. Neon CRM publishes "starts at $99". Bonterra's Apricot case management publishes a full feature comparison across three tiers with no dollar figure at all. So the honest answer is that the list price is unknowable in advance, and you should budget for the processing line instead, which is published and which is usually larger.

Is the subscription or the payment processing the bigger cost for a nonprofit?

The processing, at almost any real giving level, and it is not close. Bloomerang publishes its rates openly and with a date, which is more than most of the category manages: through Bloomerang Fundraising, Visa, Mastercard and Discover run 3.95% plus $0.30 a transaction, American Express 4.95%, ACH 1.95% plus $0.95, with another 1% on peer-to-peer. Through Bloomerang CRM the card rate is 3.2%. Little Green Light, by contrast, takes no platform cut at all — you pay your processor directly, starting around 2.2% plus $0.30. On $500,000 of online giving, a 1.75-point spread is $8,750 a year, which is more than five times Little Green Light's entire top-of-ladder subscription. Note also that bringing your own processor does not escape it: Bloomerang applies a platform fee of 1.95% on Fundraising or 1% on CRM on top of whatever your outside processor charges, and Neon charges from $50 a month for using an external processor. If you only model one number, model this one.

What changed about Maryland's charitable registration audit thresholds in 2026?

They went up, three weeks ago, and almost nothing published online reflects it yet. Chapter 394 of the 2026 Laws of Maryland — Senate Bill 354 — was approved by the Governor on 12 May 2026 and took effect on 1 July 2026. It raises the threshold for a CPA review from $300,000 to $400,000, and the threshold for an independent CPA audit from $750,000 to $1,000,000. A useful detail if you are checking sources: the bill as introduced set the review figure at $500,000 and was amended down to $400,000 before passage, so the enrolled text shows both numbers and anyone reading the introduced version will quote $500,000. The enacted figures are $400,000 and $1,000,000. The threshold is computed on gross income from charitable contributions, which is a narrower base than total revenue and — importantly for Baltimore — expressly excludes government grants.

Why do the Maryland audit threshold and the federal Single Audit threshold both sit at $1,000,000?

Coincidence, and it is the most useful piece of trivia in this whole article, because the two numbers are computed on completely different denominators. Maryland's $1,000,000 is gross income from charitable contributions and expressly excludes government grants. The federal Single Audit threshold, raised from $750,000 by the April 2024 revisions to the Uniform Guidance and effective for fiscal years beginning on or after 1 October 2024, is $1,000,000 of federal awards expended. So a Baltimore human-services organization running largely on city and state pass-through money can post $3M in revenue, owe Maryland a modest registration fee and no CPA report at all, and still trip the federal Single Audit. The reverse happens too. Any system that claims to tell you where you stand has to compute both figures from the same ledger, on two different bases, and essentially no off-the-shelf product does.

Can off-the-shelf donor software handle restricted fund accounting?

Not really, and this is the gap that sends most organizations to a spreadsheet. A donor CRM records that a gift arrived and what it was for. Restricted fund accounting requires something structurally different: under FASB ASU 2016-14 you carry net assets in two classes, with and without donor restrictions, and you recognize a release from restriction in the period the restriction is satisfied — which means matching specific expenditures against specific gifts, gift by gift, not against a revenue total. ASU 2018-08 adds a second and harder test: a grant is conditional if it carries both a barrier to overcome and a right of return, and a conditional grant is not revenue at all until the barrier is met. Money already in your bank account sits as a refundable advance, which is a liability. Worth noting because the old rule is still repeated everywhere: the test of whether the condition was unlikely to be met was superseded in 2018 and no longer applies. Almost no donor CRM models any of this, most small nonprofits fake it with QuickBooks classes, and the reconciliation between the two lives in somebody's spreadsheet.

What happens to my donor data if I leave Blackbaud or another nonprofit platform?

Ask before you sign, in writing, and ask specifically about gift history, soft credits, attachments and custom fields rather than about export in general. The terms vary far more than people expect. At the good end, Little Green Light states plainly that you can export your data on cancellation with no penalty or fee, and Bloomerang says it provides easy exports at any time. Blackbaud is the one to read carefully: its own contract review guidance tells departing customers to extract their data while they still have access, offers a zipped file of all your data on request to Customer Support, and — for Education Edge specifically — states that it cannot pull the information once access is turned off. A Blackbaud knowledgebase article on Raiser's Edge also notes it is not possible to export all data into one file, because the data lives across linked tables. None of that is unusual or improper, but it does mean the moment you decide to leave is the worst possible moment to start finding out what leaves with you.

When should a Baltimore nonprofit keep its subscription instead of building custom software?

Most of the time, and we would rather say so than sell you a build you do not need. Only about a quarter of Baltimore City's registered charities file a full Form 990 at all, and of the ones that do, some 57 percent run on under $1M a year — statewide the figure is closer to 70 percent, while the ten largest organizations in Maryland account for roughly 40 percent of all 501(c)(3) revenue. These are small teams, and for a small team a $75-a-month subscription that mostly works is a genuinely good deal. If your gift processing, acknowledgments and appeals fit inside Little Green Light, DonorPerfect, Bloomerang or Neon, keep them. Building starts to make sense at the point where the work has outgrown the product in a specific, nameable way: when a real slice of someone's month goes to assembling funder reports by hand, when restricted-fund tracking lives in a spreadsheet nobody else can maintain, when program and participant data has nowhere to live because a donor CRM was never designed to hold it, or when the number your board actually manages by does not exist in the product at any tier.