Salesforce PDF generation cost: 12,000 documents a month, priced four ways

For a Salesforce org generating 12,000 documents a month, the four available routes differ in price by about seventy times. Visualforce PDF rendering costs nothing in licences and stops working the moment a document needs a real font or a chart. A native managed package quoted per user costs roughly $42,000 a year at 140 seats. The published per-document tools land near $8,000 a year, and both of their ladders run out before they reach this volume. A rendering API called from Apex costs $594 a year. Every published figure below was read from the vendors' own pages on 16 September 2026.

The job we are pricing

Fixing one shape is the only way to make this comparison honest, because the meters in this category respond to completely different things. Our example org runs Salesforce CPQ, has 140 licensed users, and produces about 12,000 documents a month: roughly 2,000 quotes generated by sales reps on demand, and 10,000 monthly account statements produced overnight by a scheduled job. Six people build templates. Nobody else needs to touch one.

That split matters more than the volume. Five sixths of the output is automated, which means the people who could be charged for seats are not the people producing the documents. Hold that thought through the next four sections, because it is what inverts the ranking.

Route 1: Visualforce renderAs, and exactly where it stops

Salesforce has generated PDFs natively for years. Add renderAs="pdf" to a Visualforce page and the platform returns a PDF. It costs nothing, it involves no vendor, and for an internal picking list or a simple summary it is genuinely the right answer. Teams leave it for reasons that are documented, not mysterious.

The Visualforce Developer Guide sets out the ceilings plainly. The maximum response size "must be less than 15 MB before being rendered as a PDF file", the maximum size of a generated PDF is 60 MB, and the maximum total size of all images inside one is 30 MB. More restrictive than any of those, and the reason most teams actually give up: "PDF rendering doesn't support JavaScript-rendered content", and "the font used on the page must be available on the Visualforce PDF rendering service. Web fonts aren't supported." Images encoded as data: URIs are not supported either.

Read together, those rules describe a print engine from a different decade. No brand typeface unless it happens to be on Salesforce's list. No client-side chart library, so every graph has to become a server-rendered image that then counts against the 30 MB image budget. No modern layout behaviour you would take for granted in a browser. For an internal document that is fine. For a quote that goes to a customer with your brand on it, the first design review usually ends this route.

So the real cost of Visualforce is not a licence fee. It is the engineering time spent discovering these limits one at a time, and then the rebuild. Budget it as weeks, not dollars.

Route 2: a Salesforce-native app, priced per user

The obvious next step is a managed package from the AppExchange: Conga Composer, Nintex DocGen, S-Docs or PDF Butler. All four are quoted per user, and none of the four publishes a rate card, which we audited in detail on our Salesforce document generation pricing comparison. Third-party estimates for Conga alone span $6 to $50 per user per month, so any single number is an illustration rather than a quote.

Take $25 per user per month as the middle of that range. At 140 users that is $42,000 a year, before any document allowance on top. Whether you actually need 140 seats is the question worth fighting over: if a licence is only required for the six people who build templates, the same product costs $1,800 a year. That is a factor of 23 riding on one contractual definition, and it is why the first question to any vendor here should be whether triggering a document requires a licence or only authoring one does.

There is a second trap specific to our example. Ten thousand of these documents are produced by a scheduled job with no human involved. That job runs as some user, and vendors differ on whether that user needs a paid document generation licence. Get the answer in writing before signing, because it decides whether automation is free or whether it quietly re-prices your whole org.

Route 3: a published per-document tool

Platform-agnostic tools connect to Salesforce and meter documents instead of people, and unlike the native camp they print their prices. Two are worth pricing here, and the result is more interesting than the arithmetic suggests.

Documint publishes three tiers: $39 a month for 2,400 documents a year, $124 for 12,000 and $249 for 60,000, which is 5,000 a month. It also publishes its overage, at 1,000 documents for $59.80 on the top tier. Our org needs 12,000 a month, so it sits on the $249 plan plus 7,000 documents of overage at $418.60, which is $667.60 a month, or $8,011.20 a year.

Docupilot runs from $29 a month for 100 documents up to $699 for 10,000, with extra seats at $4. There is no published overage: the plan notifies at 80, 90 and 100 percent of the allowance and then you upgrade. At 12,000 a month our org is past the top published tier, so the honest answer is $8,388 a year and a sales conversation.

That is the finding worth carrying away from this section. Both published ladders in this category stop just below the volume at which generation becomes automated. Documint's top rung is 5,000 a month and Docupilot's is 10,000, and the shape of an org that outgrows them is not exotic. It is any business sending its customers a monthly statement. The moment you are generating documents on a schedule rather than on a click, per-document pricing built for human-initiated work starts behaving like a metered utility with no cap.

Route 4: a rendering API called from Apex

The last route removes the merge engine entirely. Your Apex builds HTML, posts it to a rendering service, and saves the returned PDF back onto the record as a ContentVersion. There is no managed package, no template count and no seat count. We price this on our pricing page: $29 a month for 10,000 renders, $99 for 50,000 and $299 for 250,000, with yearly billing costing the same as six months. At 12,000 documents a month our example org needs Growth, which is $594 a year and leaves it using under a quarter of the allowance.

The Apex side has real constraints and they shape the design, so they are worth stating before you cost the build. A transaction may make at most 100 callouts with a cumulative timeout of 120 seconds, and the Apex heap limit is 6 MB in synchronous code and 12 MB asynchronous. A returned PDF blob has to fit in that heap, which is the practical reason this work belongs in an asynchronous context: mark the method @future(callout=true) or use a Queueable, and you get the 12 MB ceiling instead of 6 MB. For the overnight statement run, batch the work so each execution handles a slice well inside the callout budget rather than looping 10,000 times in one transaction.

The trade is the one every hosted option carries. Record data leaves the platform, so this route is off the table for orgs whose compliance posture forbids that, and it should be mapped as a subprocessor in your next security review. The document also becomes code: changing the wording of a quote is a deployment rather than a click by the sales operations manager.

All four routes, side by side

Route	Annual cost at 12,000 a month	Who edits the template	Main risk
Visualforce renderAs	$0 in licences	Developers	No web fonts, no JavaScript content, 15 MB response ceiling
Native app, per user	About $42,000 at 140 seats, or $1,800 at 6, depending on the licence definition	Business users, in Salesforce	No published price, and an unpublished document allowance on top
Documint, per document	$8,011.20	Business users, outside Salesforce	Top tier is 5,000 a month, so most of the bill is overage
Docupilot, per document	$8,388 and still short of the volume	Business users, outside Salesforce	No overage mechanism, so you leave the published ladder entirely
Sitepdf, per render	$594	Developers, in HTML	Data leaves the platform, and template edits become deployments

The spread is not subtle, and it is almost entirely caused by the meter rather than by the quality of the products. Every tool in that table generates a perfectly good PDF. They just disagree about what to count.

When the native app is still the right purchase

We sell the last row, so here is the case against it. If your sales operations team rewrites the quote template every few weeks and expects to do it without filing a ticket, a merge app pays for itself on that alone, and no per-render price makes up for putting a deployment between a business user and a wording change. If your security review forbids record data leaving the org, S-Docs generating entirely on-platform is the correct answer and the price is irrelevant. And if your contracts genuinely live in Word because your legal team redlines them there, template merge is the only route that respects that file at all.

The rendering route earns its place in one specific situation, and our example org is squarely in it: most documents are produced automatically, the design matters because customers see it, and the number of people who edit templates is far smaller than the number who hold a Salesforce licence. In that shape, paying per seat is paying for an editing workflow nobody uses.

One more thing worth planning for while you are costing this. Generating 10,000 statements a month is the cheap half of the problem. The expensive half starts afterwards, when finance has to match every one of those invoices against the payment that eventually arrives, and the volume that made document generation a budget line makes reconciliation one too. Teams tend to solve the first and discover the second a quarter later.

If you are still deciding between the camps rather than between vendors inside one, our document generation API comparison prices the non-Salesforce field on the same basis, and the Salesforce document generation pricing page covers what each native vendor will and will not tell you before a call.