SafeSend will not publish a price. It does publish the contract, and that is the better document.

SafeSend One is the software a tax firm bolts onto the end of its return process. It gathers source documents from the client, assembles the finished return into a package the client can actually read, collects the electronic signature on the e-file authorization, distributes K-1s, takes the payment, and chases everyone who has not done their part yet. It sells alongside two other things from the same company: TicTie Calculate, a PDF workpaper markup tool that predates the rest of it, and a secure file transfer product.

Read the feature list and the temptation is to file this under convenience software. That is the wrong frame, and it is the reason firms buy badly in this category. Return delivery is not a document problem. It is a choreography problem with a compliance artifact at the end of it, and the artifact has rules that come from the IRS rather than from your firm. Everything upstream of the signature is scheduling and nagging. The signature itself is regulated, and we get to that below, because it is the single fact that determines which alternatives are real and which ones only look like alternatives.

The second thing worth naming early is the shape of the load. A tax firm's delivery volume is not spread across a year. It arrives in a few weeks, twice, and every hour of manual handling inside those weeks costs more than the same hour in September. That is what makes this category worth paying for, and it is also what makes the terms of the deal matter more than they would for software you use evenly. An outage in the second week of April is not the same event as an outage in the second week of July, and no service level agreement we have read in this category prices that difference.

This is the change most comparison articles have not caught up with, and it is the first thing a buyer should know. SafeSend's own About page says the company joined the Thomson Reuters family in 2025. The site footer carries a badge reading "Since 2025 part of" alongside the Thomson Reuters mark, the support address is supportsafesend@thomsonreuters.com, and the legal policies index links a Thomson Reuters privacy statement next to SafeSend's own. The contracting entity on the master agreement is still cPaperless, LLC doing business as SafeSend, which is the company that started in 2008 as CPA Paperless.

Why this matters is structural rather than sinister. SafeSend's whole value proposition was that it sat neutrally on top of whichever tax package your firm ran. Its integration list covers CCH Axcess Tax and ProSystem fx Tax from Wolters Kluwer, UltraTax CS and GoSystem Tax RS from Thomson Reuters, and Lacerte from Intuit. Two of those five belong to its new owner. A firm on CCH Axcess is now routing its most client-facing workflow through software owned by a competitor of the vendor that makes its tax engine.

We want to be precise about what we can and cannot see, because the fair version of this is narrower than the dramatic version. SafeSend still publishes support for all five packages, and its AI intake page states in plain prose that collected documents flow into CCH Axcess Tax and ProSystem fx Tax, Thomson Reuters UltraTax CS and GoSystem Tax RS, and Intuit Lacerte. Its sales forms still list all five. What has visibly changed is emphasis: there is a dedicated integrations page for Thomson Reuters, headed "SafeSend One and Thomson Reuters, Unified" and pitching SafeSend as the number one choice for Thomson Reuters users, while the equivalent Intuit and Wolters Kluwer integration URLs return 404. That is a marketing signal, not a technical one, and we are not going to inflate it into a claim that support is being withdrawn. There is no evidence of that and we did not find any.

There is a second half to this that almost no comparison mentions, and it changes what the word alternative means on this page. The product most often recommended as the alternative to SafeSend is TaxCaddy, usually paired with SurePrep for scan and populate. Both are also Thomson Reuters now. The TaxCaddy site footer reads that it is part of Thomson Reuters, and sureprep.com no longer resolves to its own site at all: it redirects to a Thomson Reuters product page. So the two best known products in tax return gathering and delivery, the incumbent and its most cited rival, report to the same owner. A firm negotiating one against the other is negotiating with itself.

The correct action for a buyer is small and concrete. If you are on CCH Axcess, ProSystem fx or Lacerte, ask where your tax package sits on the integration roadmap for the next three seasons, and ask for the answer as a term rather than as reassurance on a call. If you are on UltraTax CS or GoSystem Tax RS, the acquisition is straightforwardly good news for you and you can stop worrying about this section. And if your shortlist is SafeSend against TaxCaddy, understand that you are choosing between two products from one vendor, which is a fine thing to do deliberately and an expensive thing to do by accident.

One more gap worth knowing before you shortlist: neither Drake nor Intuit ProConnect appears anywhere in SafeSend's published integration prose, and neither is an option in the tax software dropdown on its own forms, which lists UltraTax CS, GoSystem Tax RS, CCH Axcess, ProSystem fx Tax, Lacerte, and Other. If your firm runs Drake, that Other option is the beginning of a conversation rather than the end of one.

SafeSend has a pricing page. It loads normally, it is titled around simple pricing that fits your firm's needs, and it says that the pricing model is transparent and carries no hidden software fees. It contains no dollar figure of any kind.

What it does contain is genuinely useful, so credit where it is due. It names three packages, Essential Gather, Essential Deliver and Premium, and it tells you which modules sit in each. It states the shape of the model, which is pay per return or per delivery rather than per seat. And it publishes an explicit list of things the company says it does not charge for: per seat fees, monthly or annual maintenance and renewal fees, support fees, fees that scale with firm size, network and terminal server fees, and knowledge-based authentication fees. That last one is not filler. KBA is a per-transaction cost with a third party behind it, and a vendor stating in public that it does not meter you for it is a real commitment that a buyer can hold them to.

So the page tells you the pricing architecture and refuses to tell you the price. The only route to a figure is the Request Pricing form, which asks for your email, phone number, firm, job role, which tax software you run, and how many returns you process, banded as under 100, 100 to 999, 1,000 to 4,999, and 5,000 or more. Those bands are the most informative thing on the page, because they tell you what the quote is keyed to. You are being priced on volume, and your negotiating position is your volume.

We say this on every page in this series and it applies here: a vendor that will not publish a number is not doing anything wrong, but the buyer should notice that the phrase "transparent pricing" is being used to describe a fee structure rather than a price. Those are different things. The structure is transparent. The price is not published.

Here is the part almost nobody does, and it is the most valuable half hour in this entire evaluation. SafeSend publishes its Master Software License and Services Agreement, its general terms and conditions, and its support and maintenance terms, in full, on its website, readable without a login and without filling in a form. That is unusual in this category and it deserves to be said before we quote anything out of it. Most of the vendors a firm shortlists here will not show you a contract until you are deep in a sales process.

Read it and you learn six things the marketing pages do not tell you.

One, AI training on your data is permitted by default. The Customer Information License Grant allows cPaperless to use Customer Information to improve and enhance the software and its other product offerings, and it says so specifically, including to create, train and improve artificial intelligence and machine learning models. There is a real condition attached: it must first remove, obfuscate or otherwise de-identify any personal information so that it no longer constitutes personal information. That is a meaningful guardrail and it is more than several competitors put in writing. It is also a default permission rather than an opt in, and the material that would be de-identified is tax data, which is unusually dense with identifying detail. If your firm's position is that client information is never used for model training, that has to be an amendment on your order form. It is not the published position.

It is worth seeing that clause next to what its competitors publish, because the three postures are genuinely different and all three are public. TaxDome's privacy policy states flatly that customer data is not used for AI model training, and adds that data sent to OpenAI is not retained beyond processing the request. Liscio's privacy policy takes the same shape as SafeSend's but with a switch: using its AI features is consent to Liscio using inputs and outputs to improve them, including for training or fine tuning, with direct identifiers removed, and a firm that objects can turn the AI features off. SafeSend's is the middle position, permission by default with de-identification and no described off switch. None of the three is hiding anything. They have simply made different promises, and a firm with a written policy on client data can compare them in an afternoon without talking to a single salesperson.

Two, ownership is split into two categories and only one of them is yours. The agreement states that the customer retains all right, title and interest in Customer Information and user IDs, with an exception carved out for Data collected under Section 8. Section 8 then defines Data as non-personally identifiable information resulting from your use of the software, and says that it is solely owned by cPaperless, may be used for any lawful business purpose without a duty of accounting to you, and is used only in aggregated form without identifying you as the source. Section 7.3 adds that Data is not treated as Confidential Information. The fair reading is that this is usage telemetry rather than your clients' returns, and the aggregation proviso is a genuine limit. It is still a category of information about your firm's operations that you do not control and cannot ask to have deleted under the confidentiality provisions.

Three, the agreement terminates itself if you stop using it. If you do not use or deliver a return for eighteen consecutive months, the agreement automatically terminates and any unused units are forfeited. Since the licence is granted up to a purchased number of returns, a firm that buys a block ahead of a season and then handles that season differently can lose the remainder. This is the clause most likely to cost a real firm real money without anyone noticing, and it is one sentence long.

Four, getting your data out is a paid professional service, not a button. The published terms describe a Transition Period of up to thirty days from termination, during which you may request services reasonably necessary to move to another vendor. cPaperless commits to commercially reasonable efforts, charges its then current service rates for the work, and reserves the right to require payment in advance. Separately, and this is a credit, retention inside the product is customer configurable from one month to seven years with automatic destruction at the end of the period you set. We have not tested the product's own export tooling and we are not going to claim it lacks one. What we can say is that the contract does not promise you one, and exit terms are cheapest to negotiate before you sign.

Five, the security commitments are better than most and they are enforceable. The agreement commits cPaperless to an independent third party penetration test that includes testing for SQL and other injection vulnerabilities, to an independent third party security and controls assessment no less than once every twelve months following an established framework with SOC 1 and SOC 2 named as examples, and to immediate notification of any known unauthorized use, disclosure, modification or destruction of Confidential Information, followed by a written statement of circumstances and a proposed remediation plan acceptable to the customer. Written into the contract, that is worth considerably more than a trust badge on a website, and it is the strongest single argument for taking SafeSend seriously.

Six, the service level is 99.4 percent quarterly, and the remedy is small. Below 99.4 percent availability in a quarter you are entitled to a credit: nothing at 99.5 percent or above, 2.5 percent from 98 to 99.4 percent, 5 percent from 97 to 97.99 percent, and 7.5 percent below 97 percent. It is the sole and exclusive remedy, you have to measure the unavailability yourself and claim within ten days of month end or lose it, and no credits are issued if your account is not current on payment. There is also a clause allowing the agreement to be amended at the vendor's discretion on thirty days notice, with your remedy being to terminate and recover prepaid unused fees.

None of that is disqualifying. Several of those terms are better than the category norm and the security section is genuinely strong. The point is that all six answers are sitting in public, for free, and the buyer who reads them walks into the pricing conversation knowing what to negotiate: the training carve out, the unit expiry, and the exit.

Every list of SafeSend alternatives eventually suggests that a firm could handle delivery with a secure portal and a general purpose e-signature product for a fraction of the money. It is the most common bad advice in this category, and the reason it is wrong is a document any firm can read for itself.

IRS Publication 1345, the handbook for authorized e-file providers of individual income tax returns, was revised in December 2025 and replaced the November 2024 edition. It specifies what the software has to do when an electronic signature is obtained on Form 8878 or 8879. The software must enable identity verification using knowledge-based authentication questions. It must record a defined set of items: a digital image of the signed form, the date and time of the signature, the taxpayer's computer IP address and login identification for remote transactions, the taxpayer's passed identity verification results, and the method used to sign the record or a system log or other audit trail evidencing completion. The provider has to produce that information to the IRS on request. The signature must be linked to its record so that it cannot be excised, copied or transferred, and the record has to be tamper-proof after signing.

Then comes the sentence that decides the whole question. Publication 1345 requires the software to disable identity verification after three attempts, and provides that if the taxpayer fails knowledge-based authentication after three attempts, the provider must obtain a handwritten signature on Form 8878 or 8879 instead. There is a fallback path baked into the rule, and somebody in your firm walks it every season.

That reframes what you are actually buying. You are not buying a signature. You are buying the identity verification, the evidence record behind it, and a defined process for the taxpayers who fail it. A general purpose e-signature product signs documents beautifully and is not built around any of those three things. It can be part of an answer, and for entity returns and engagement letters where 8879 is not in play it can be most of one, but it does not replace a category built for this rule.

It also explains why SafeSend markets a knowledge-based authentication success rate at all, and how to interrogate that number. The company publishes 97 percent success on 1040 returns. We cannot audit that from outside and we report it as a vendor claim. What we can say is that the interesting number is the other one. Under the rule above, the taxpayers who fail after three attempts sign on paper, and your staff handle them by hand in the two weeks of the year when handling anything by hand hurts most. Ask every vendor on your shortlist for that rate, ask how it is measured, and ask what the product does at attempt four. The answer to the third question is worth more than the answer to the first.

Seven products a firm might reasonably shortlist against SafeSend, on the questions that can be answered from outside without entering a sales process. Every line was read on the vendor's own site on August 15, 2026. Where a figure exists only on a review aggregator we have deliberately left it out, because those numbers are frequently stale and are not the vendor's commitment to anything.

Two things fall out of that table that are worth more than the rest of the page.

The first is the KBA line, and it is the only genuinely apples to apples number in this category. Canopy publishes knowledge-based authentication at $1.25 per credit, billed on top of its per user subscription and on top of its $34 per client delivery module. TaxDome publishes $1 per KBA, on top of a seat price that already includes unlimited e-signatures. SafeSend publishes, on its own pricing page, that it does not charge KBA fees at all. Three vendors, three public positions on the same line item, all read off their own pages. At a few thousand returns, with some taxpayers needing more than one attempt, that stops being trivia. It is also why SafeSend's refusal to publish a headline price is less damning than it first looks: the careful buyer can still price a real part of the deal, and can hold the vendor to the part that is published.

The second is Liscio's per item rates. It is the only vendor here that will tell you, without a sales call, what one organizer and one return delivery cost. Whatever you end up buying, those two numbers are the anchor to walk into a SafeSend pricing conversation with, because they are a competitor's public commitment rather than your guess.

One caution about numbers you will find elsewhere. At least one competitor's marketing page runs a savings calculator that asserts a per return price for SafeSend. SafeSend does not publish one. We are not repeating a rival's estimate of a company's pricing as though it were that company's pricing, and neither should your evaluation memo.

How to tell which one you are.

The decision is usually settled by one question, and it is not a question about software. Where does the time actually go?

If the painful part is the last mile, the assembly, the signature, the chasing of people who have not signed, then you are buying a delivery product and you should buy the best one you can negotiate. That is a solved category and building anything here would be an expensive way to arrive back where you started.

If the painful part is the first mile, the documents that arrive in nine formats through four channels and have to be identified, sorted, matched against last year and chased when missing, then the delivery product is not your constraint and buying a better one will not move your season. This is the half where firms differ from each other, which is precisely why it is the half no vendor can package completely, and it is where a commissioned layer earns its keep.

And if the honest answer is that the product is fine and the contract is the problem, then the work is a renewal conversation with three specific asks, not a migration. Migrating a delivery platform between January and April is not a project, it is an incident.

What we would build, and what we have not built.

To be plain about our own position: ColabContent has not shipped a commission that replaces a tax return delivery platform, and we would talk a firm out of that project if it asked for it. What we build for accounting firms sits upstream, in the intake and classification work described above, and beside it in the reconciliation and review workflows that are specific to a firm rather than to the profession. On this page we are not the fifth vendor in the comparison table. We are the option you reach for after the table has told you that your problem was never delivery.

If you want the wider view of what does and does not automate cleanly in a tax practice, what actually works for CPA firms covers it without vendor framing, the AI tools evaluation guide covers how to shortlist, and what this costs covers the money before you spend any time on a call.

Who owns SafeSend?

SafeSend is part of Thomson Reuters. The company's own About page states that it joined the Thomson Reuters family in 2025, its footer carries the badge "Since 2025 part of" Thomson Reuters, its support address is supportsafesend@thomsonreuters.com, and its legal policies index links a Thomson Reuters privacy statement alongside its own. The contracting entity is still cPaperless, LLC doing business as SafeSend, which is the name on the master agreement and on the site copyright line. The business started in 2008 as CPA Paperless.

How much does SafeSend cost?

SafeSend does not publish a figure. Its pricing page loads normally and describes three packages, Essential Gather, Essential Deliver and Premium, and it describes the shape of the model as pay per return or per delivery rather than per seat. It contains no dollar amount anywhere on the page. The only route to a number is the Request Pricing form, which asks for your email, phone, firm, job role, which tax software you run and how many returns you process. Firms are banded on that form as under 100 returns, 100 to 999, 1,000 to 4,999, and 5,000 or more, so expect the quote to key off volume.

Does SafeSend train AI models on customer data?

Its published master agreement grants that right explicitly, with a condition attached. The Customer Information License Grant lets cPaperless use Customer Information to improve and enhance the software and other product offerings, including to create, train and improve artificial intelligence and machine learning models, and it requires that cPaperless first remove, obfuscate or otherwise de-identify any personal information so that it no longer constitutes personal information. Read it as written: the permission is the default term rather than an opt in, and the protection is de-identification rather than exclusion. A firm that wants its client data excluded from model training entirely needs that written into its order form, not assumed from the marketing.

Who owns the data inside SafeSend?

The agreement splits it in two, and the split is the part worth reading twice. Customer Information stays yours: the agreement says the customer retains all right, title and interest in it. Separately, Section 8 defines Data as non-personally identifiable information resulting from your use of the software, and states that this Data is solely owned by cPaperless and may be used for any lawful business purpose without a duty of accounting to you, provided it is used only in aggregated form without identifying you as the source. Section 7.3 also states that Data is not treated as Confidential Information. In plain terms that is usage telemetry rather than your clients' tax returns, and the aggregation proviso is a real limit, but it is a category of information you do not control.

What happens to a SafeSend account that goes unused?

It terminates on its own. The agreement provides that it automatically terminates if the customer does not use or deliver a return for eighteen consecutive months, and that on termination any unused units are forfeited. Since the licence is sold against a purchased number of returns, that clause has teeth for a firm that buys a block ahead of a season it then handles differently. Anyone buying units in bulk should know the expiry rule before the block, not after it.

Can we export our data if we leave SafeSend?

The published agreement does not describe a self-serve export. It describes a Transition Period of up to thirty days from termination during which the customer may request services reasonably necessary to move to another vendor, with cPaperless using commercially reasonable efforts, charging its then current service rates, and reserving the right to require payment in advance. Separately, retention inside the product is customer configurable from one month to seven years, with automatic destruction at the end of the period you set. We have not tested the product's own export tooling, so treat the agreement as the floor rather than the whole story, and ask for the export path in writing during diligence.

What service level does SafeSend commit to?

The published support terms set the bar at 99.4 percent quarterly availability and attach a credit schedule below it: nothing at 99.5 percent or above, 2.5 percent between 98 and 99.4 percent, 5 percent between 97 and 97.99 percent, and 7.5 percent below 97 percent. The credits are the sole and exclusive remedy, you have to measure the downtime yourself and claim within ten days of month end or forfeit it, and no credit is issued if your account is not current on payment. For a product whose entire load falls inside a few weeks, the honest reading is that the SLA is a posture rather than a hedge, because a 2.5 percent credit does not price an outage in the first week of April.

Why can a firm not just use DocuSign for Form 8879?

Because the IRS specifies the record, not just the signature. Publication 1345, revised December 2025, requires that software used to obtain an electronic signature on Form 8878 or 8879 enable identity verification through knowledge-based authentication, and that it capture a defined record set: a digital image of the signed form, the date and time of signature, the taxpayer's IP address and login identification for remote transactions, the taxpayer's passed identity verification results, and the method used to sign or an equivalent audit trail. The ERO has to produce that to the IRS on request. It also requires the software to disable identity verification after three attempts, and if the taxpayer fails knowledge-based authentication after three attempts the ERO must obtain a handwritten signature instead. A general purpose e-signature tool signs the document. It does not necessarily run the KBA, capture that record set, or handle the fallback.

What does the 97 percent KBA success rate actually mean for a firm?

SafeSend publishes 97 percent knowledge-based authentication success on 1040 returns as a marketing figure, and we have no way to audit it from outside, so treat it as a vendor claim. The useful way to read it is from the other end. Under Publication 1345 a taxpayer who fails knowledge-based authentication after three attempts has to sign Form 8879 by hand, so the residual percentage is not a rounding error, it is the population your staff chases manually during the worst two weeks of the year. Any vendor in this category should be asked for that number, how it is measured, and what the product does at attempt four.

Is SafeSend still neutral now that Thomson Reuters owns it?

It still publishes support for the competing stacks, and that is the fact to hold onto. Its Next Gen Gather AI page states that collected documents flow into CCH Axcess Tax and ProSystem fx Tax, Thomson Reuters UltraTax CS and GoSystem Tax RS, and Intuit Lacerte, and its sales forms list all five as supported. What has changed is emphasis. There is a dedicated integrations page for Thomson Reuters, headed SafeSend One and Thomson Reuters, Unified, and it markets SafeSend as the number one choice for Thomson Reuters users. The equivalent Intuit and Wolters Kluwer integration URLs return 404. Absence of a marketing page is not absence of an integration, so the honest conclusion is narrow: the roadmap now has an owner with its own tax software, and a firm on CCH Axcess or Lacerte should ask where its stack sits in that roadmap and get the answer in the contract.

Which alternatives publish a real price?

Three of the six products we checked do, and all three are practice platforms rather than pure delivery tools. TaxDome publishes $800, $1,000 and $1,200 per seat per year for Essentials, Pro and Business on a one year commitment, dropping to $700, $900 and $1,100 on a three year commitment, with e-signatures included and knowledge-based authentication charged at $1 per KBA. Canopy publishes $74, $109 and $149 per user per month billed annually, with its Premium tier flagged as coming soon, the tax return delivery module starting at $34 per client credit, and knowledge-based authentication billed separately at $1.25 per KBA credit. Liscio publishes $19, $49 and $99 per user per month billed annually, and it is the only one that publishes per item rates for the work itself: an organizer at $5 each and a tax return delivery at $5 each. SafeSend, TaxCaddy and Suralink publish no figure. Every one of those numbers was read on the vendor's own pricing page on August 15, 2026, and we deliberately left out figures that exist only on review aggregators.

Is TaxCaddy a real alternative to SafeSend?

It is the alternative most often named, and buyers should know that both now belong to Thomson Reuters. SafeSend's About page states it joined Thomson Reuters in 2025. TaxCaddy's own footer reads that it is part of Thomson Reuters, and sureprep.com, the site for the scan and populate product TaxCaddy is paired with, now redirects to a Thomson Reuters product page. TaxCaddy publishes no price either, and its pricing URL returns a 404. That does not make it a bad product, and the two have different strengths in gathering versus assembly. It does mean that a firm switching from one to the other is negotiating with the same parent company, which is worth knowing before you treat the switch as leverage.

Should a mid market firm buy a delivery platform or commission one?

Buy, in almost every case, and we will say that plainly on our own comparison page. Return delivery is a solved workflow with a compliance artifact at the end of it, the requirements come from the IRS rather than from your firm, and the load is seasonal, which is the exact profile that favours renting. Commissioning becomes worth discussing when the constraint sits upstream of delivery, in the gathering, chasing, classifying and review work that is specific to how your firm is organised, or when you are paying per return at a volume where the cost has become a line item somebody defends every year.

Related reading.

More on AI in accounting firms: what we would build for a CPA firm, what AI consulting for CPA firms covers, and the consultants accounting firms actually shortlist. On the tax engines themselves, the CCH Axcess playbook, the Lacerte playbook and the UltraTax playbook each cover building a layer on the software you already own. For choosing a partner, how to choose an AI consultant for a CPA firm and what a CPA automation consultant does. The neighbouring comparisons cover different layers of the same stack: Karbon and the practice management layer, Jetpack Workflow against Karbon, and renting AI against owning it. On ownership generally, why code handoff matters, and on diligence, the security questions to ask before a build. To size your own numbers first, the CPA teardown takes a few minutes, and how we work covers what happens next. The rest of this series sits on the comparisons hub.