Client Document Collection Statistics (2026)

I found no published study showing that AI shortened the document chase. Every AI adoption survey cited here rose sharply: attorneys whose office uses AI tools went from 10.9% in 2023 to 30.2% in 2024 (ABA). Yet 64–65% of surveyed accounting firms still named client documents among their biggest workflow challenges in 2024–2025. And US purchase loans still take about 37 days to close (ICE, 2026).

Last verified: 2026-10-25 · 98 statistics in the dataset, each checked against its source by script · 3 original analyses of public data · Download the CSV · Cite or embed

I build DokuTrak, a tool that asks clients for documents and follows up for you (pricing). This page is about the data, not the product. I opened every source below, matched the number to the wording, and labeled who paid for each survey. A script then re-checks that every figure is still printed on its source page.

Key takeaways

• 65.2% of 367 surveyed accounting firms named getting documents from clients among their biggest workflow challenges in 2024; in 2025, 64% named chasing clients for information, the most common answer (Financial Cents, vendor surveys).
• 49% of 240 surveyed accounting professionals spend 1–2 days gathering client information for a typical engagement, and 36% spend 3–7 days (Canopy, vendor survey, 2024).
• 1,449,500 accountants and auditors and 76,480 tax preparers held wage-and-salary jobs in the US in May 2025 (Bureau of Labor Statistics, 2026).
• 650,667 CPAs held an active US license as of August 27, 2026 (NASBA, 2026), and 886,338 people held a 2026 PTIN (IRS, 2026).
• 57% of individual taxpayers, over 85 million people, used a paid preparer in fiscal year 2024 (GAO, 2026, from IRS data).
• Tax professionals e-filed 75,316,000 individual returns in the 2026 season through May 8, versus 65,730,000 self-prepared e-filed returns (IRS, 2026).
• 27.6% of individual returns e-filed by tax professionals through October 17, 2025 arrived after April 11, against 23.5% of self-prepared returns (DokuTrak analysis of IRS filing season statistics).
• 92% of lawyers use email to communicate with clients, and 35% say their firm offers clients a secure client portal (ABA, 2023).
• 695,006 US mortgage applications, 5.9% of those with a final decision, were closed for incompleteness in 2025 (DokuTrak analysis of public CFPB HMDA data).
• A data breach in financial services cost $6.3 million on average, against $4.99 million across industries (IBM, 2026, vendor).
• The IRS took about 20 months to resolve identity theft victim cases (National Taxpayer Advocate, 2026).
• Workers report generative AI time savings equal to 1.6% of all US work hours (Federal Reserve Bank of St. Louis, 2025).
• 35% of tax and audit professionals use AI tools their firm hasn't authorized (Thomson Reuters Institute, vendor research arm, 2026).
• The Charlotin database lists 898 US court decisions on AI-hallucinated material in 2026 through September 30, more than the 526 for all of 2025 (DokuTrak analysis, 2026).

Has AI changed how professionals collect client documents?

AI changed the tools, not the chase. Adoption surged between 2023 and 2026, the indicators of document chasing did not improve (several rose), and AI added new risks around documents: forged files in lending, unapproved tools and unchecked output. No source measures AI's effect on document-chasing time directly, so this section puts the series side by side and lets each keep its own definition.

AI adoption rose. The chase did not shrink.

Each panel is a separate survey or dataset with its own axis.

Do not compare panels with each other.

AI ADOPTION

THE CHASE

Attorneys whose office uses AI

0%

50%

100%

10.9%

2023

30.2%

2024

ABA TechReports, 2023 and 2024

Documents named as a top problem

0%

50%

100%

53.8%

2023

65.2%

2024

64%

2025

Financial Cents (vendor); wording changed 2025

Legal pros using AI for work

0%

50%

100%

27%

2024

31%

2025

69%

2026

8am (vendor); 2024-25 via LawNext

Days to close a purchase loan

0

25

50

37.3 d

2025

36.8 d

2026

ICE loan-level data, March

Mortgage lenders using AI/ML

0%

50%

100%

15%

2023

38%

2024

STRATMOR Group, 2024 study

Loan files closed incomplete

0%

5%

10%

5.3%

2023

5.4%

2024

5.9%

2025

DokuTrak analysis of HMDA

Sources and definitions: dokutrak.com/blog/client-document-collection-statistics

Adoption surged. Attorneys whose office uses AI tools went from 10.9% in 2023 to 30.2% in 2024 (ABA TechReports, 440 and 512 respondents).¹² A different survey measures personal use of general-purpose GenAI. It found 69% of legal professionals using it for work in 2026, up from 27% and 31% in its two previous editions (8am, a vendor, 1,300+ US legal professionals; the 2024 and 2025 points as reported by LawNext).³⁴ Professionals whose organization uses GenAI went from 22% to 40%, and 34% at tax firms (Thomson Reuters Institute, the research arm of a software vendor; global sample of 1,514 respondents in 27 countries).⁵ Mortgage lenders using AI or machine learning went from 15% in 2023 to 38% in 2024 (STRATMOR Group, 2024 Technology Insight Study).⁶

The chase did not shrink. Getting info and documents from clients was named by 53.8%, then 65.2% of surveyed firms; in 2025, 64% named chasing clients for information. It was the most-cited workflow problem in 2024 and 2025 (Financial Cents, vendor surveys; the question wording changed in 2025).⁷⁸ Purchase mortgages close in 36.8 days versus 37.3 a year earlier, a record low in ICE's series but still over a month (ICE loan-level data, vendor).⁹ Mortgage applications closed for incompleteness went from 5.3% to 5.4% to 5.9% of decided applications between 2023 and 2025 (DokuTrak analysis of public CFPB HMDA data). Tax pros still name missing or incomplete client information as the season's biggest problem (NATP, 2026).¹⁰

Measured savings are small. Workers report GenAI time savings equal to 1.6% of all work hours (St. Louis Fed, 2025).¹¹

AI added new risks. AI-generated document fraud grew nearly 5x from May to December 2025, from a small base (Inscribe, vendor).¹² 35% of tax and audit professionals and 34% of law-firm professionals use AI tools their firm has not authorized (Thomson Reuters Institute, vendor research arm, global sample, 2026).¹³¹⁴ The Charlotin database lists 38 US court decisions on AI-hallucinated material in 2024, 526 in 2025 and 898 in 2026 through September 30 (DokuTrak analysis).

Three honesty notes before the details. First, the adoption surveys ask different questions: office use versus personal use, "any AI" versus GenAI. Read each number with its definition. Second, nothing here shows AI caused or failed to cause any of these numbers. Third, the fair summary is narrow: no published data shows the chase got shorter. The survey and HMDA indicators did not improve, and closing time improved by half a day.

How many professionals collect documents from clients in the US?

The four core occupations hold 3.65 million wage-and-salary jobs, and 4.3 million with paralegals and loan officers. In May 2025, 1,449,500 accountants and auditors, 1,373,680 bookkeeping and accounting clerks, 76,480 tax preparers and 754,500 lawyers held wage-and-salary jobs (Bureau of Labor Statistics, 2026). Paralegals held 392,880 jobs and loan officers 274,330.

Jobs and wages come from the May 2025 Occupational Employment and Wage Statistics, released May 15, 2026.¹⁵ They count wage-and-salary jobs only. Self-employed practitioners are not in them, and not every one of these jobs involves outside clients. Projections come from the Occupational Outlook Handbook, which now runs from 2025 to 2035.^1617181920 The Handbook has no page for tax preparers.

How many CPAs are licensed in the US?

650,667 CPAs held an active license as of August 27, 2026, according to NASBA's Accountancy Licensee Database.²¹ The count covers 53 of the 55 licensing jurisdictions. Hawaii and New Mexico are not in it. NASBA is the National Association of State Boards of Accountancy, the body that coordinates the state boards that license CPAs.

How many tax preparers are there in the US?

886,338 people hold a current preparer tax identification number (PTIN) for 2026, according to the IRS Return Preparer Office, with data as of September 1, 2026.²² Among them are 209,076 certified public accountants and 71,966 enrolled agents. A PTIN is the number the IRS requires of anyone paid to prepare a federal return. It is an administrative count, and some preparers hold several credentials.

How many new accountants enter the profession each year?

40,817 students earned an accounting bachelor's degree in the 2023–24 academic year, based on provisional federal data reported in the AICPA's 2025 Trends report.²³ 28,082 new candidates entered the CPA Exam pipeline in 2024. That was down from 42,626 in 2023, the highest tally since 2016, and AICPA notes that the year before a major exam change has historically drawn a surge of test takers. BLS expects about 115,300 openings a year for accountants and auditors over the decade, and about 144,100 a year for bookkeeping clerks even as that occupation shrinks.¹⁶¹⁷

How many accounting, tax and law offices are there?

62,904 private establishments operated as offices of CPAs in 2025, with 526,519 employees on average (BLS Quarterly Census of Employment and Wages, 2025 annual averages).²⁴ Tax preparation services counted 37,505 establishments and 99,127 employees.²⁵ Legal services counted 211,361 establishments and 1,221,471 employees.²⁶ An establishment is one location, not one firm, and the census covers employers only. For lawyers as people, the ABA counts 1,374,720 in 2025, up from 1,355,963 a year earlier.²⁷ That count includes lawyers the BLS occupational table (OEWS) leaves out, such as the self-employed.

Accounting and tax: how hard is it to get documents from clients?

It is the most-cited workflow problem in Financial Cents' annual survey: 64–65% of surveyed firms name it among their biggest challenges. A typical engagement takes days of back-and-forth, and 49% of surveyed professionals say their firm lets clients email sensitive forms.

What share of accounting firms say chasing client documents is a top problem?

65.2% of 367 surveyed firms named "getting info and documents from clients" among their biggest workflow challenges in Financial Cents' 2024 report, the top answer, up from 53.8% of the previous year's respondents.⁷ The 2025 edition surveyed 816 firm owners. It found "chasing clients for information" was the most common bottleneck at 64%, in a question where respondents could pick several challenges. It also said getting documents from clients "is still the biggest workflow issue."⁸

The 2024 sample was 367 owners and admins of accounting, bookkeeping and tax firms, mostly in North America, drawn from the vendor's email list between November 2023 and January 2024. Financial Cents sells workflow software to these firms. In 2022, a Wolters Kluwer survey of 1,445 US tax and accounting firms found that "half of this year's survey respondents reported late or unprepared clients to be a top challenge."²⁸ Different vendor, different wording, same problem near the top.

How long does it take to gather a client's information for one engagement?

49% of accounting professionals spend 1–2 days communicating with clients to gather everything for a typical engagement. 36% spend 3–7 days, and 16% spend two weeks or more (Canopy, a vendor of accounting software, with CPA Practice Advisor, 2024).²⁹ The survey covered 240 accounting professionals in June 2024. The same respondents spend 9.3 hours a week on client communication and want to bring it down to 7.2.

Financial Cents asked firms to recall how fast clients sent documents before they automated. 49.9% said clients took several days, 18.5% said weeks or months, and 1.9% said same or next day.⁸ Read this with care: it is a vendor survey, and a self-reported before-and-after from users or prospects of workflow software, so it leans toward firms that already adopted it.

How stretched are accounting and tax firms in 2026?

Very. 90% of US accounting and bookkeeping professionals say fatigue and burnout have been a significant issue. 70% say IRS staffing cuts have increased their administrative workload (Intuit QuickBooks, 2026 Accountant Technology Survey, 725 respondents, vendor).³⁰ Half of the 600 firm respondents in Thomson Reuters' 2026 State of Tax Professionals Report say they spend too much time on low-value, low-margin work. They named it the top barrier to profitability.³¹ Thomson Reuters sells tax software to these firms, and 64% of its sample is in the US. Chasing a missing 1099 is the kind of work I would put in that bucket.

Are clients ready to send documents online?

About half are. Accounting professionals in the same Intuit survey say 53% of their clients are tech-advanced or proficient and 47% are tech-limited or challenged.³⁰ That is the firm's estimate of its own client base, not a survey of clients. A request a tech-limited client can't open is a request that waits.

Do accounting firms still accept tax documents by email?

Yes, about half do. 49% of respondents said their firm allows clients to email sensitive information such as tax forms, and 51% said it does not (Canopy with CPA Practice Advisor, 2024, 240 respondents, vendor).²⁹ What the rules say about that is in the security section below. If you are building a list for clients, my tax document checklist covers what to ask for.

Do missing documents push tax returns past the deadline?

Tax pros say so, but no one counts it. The National Association of Tax Professionals reported in April 2026 that "the biggest issue wasn't complex tax situations – it was missing or incomplete information." The result: returns "being filed closer to the deadline or being placed on extension."¹⁰ That comes from early responses from NATP members, with no sample size and no percentage. The IRS estimates that "more than 20 million taxpayers" were expected to file by the October 15, 2025 extended deadline.³²

To see whether professionals carry a longer tail than self-filers, I compared the IRS weekly filing tables. More than a quarter (27.6%) of the returns tax professionals e-filed through October 17, 2025 arrived after April 11, compared with 23.5% of self-prepared returns (DokuTrak analysis of IRS filing season statistics).

Returns e-filed after April 11, 2025

Share of individual returns e-filed through October 17, 2025

Tax professionals

27.6%

Self-prepared

23.5%

Calculation: (cumulative e-file to Oct 17 minus to Apr 11) / cumulative to Oct 17.

The Apr 11 cutoff is before the Apr 15 deadline: the tail includes some on-time returns.

Professionals prepare more complex returns; the gap may reflect complexity, not documents.

The IRS does not publish why returns come in late.

Source: DokuTrak analysis of IRS Filing Season Statistics, 2025

How much time does a client spend preparing their own tax paperwork?

The IRS estimates a Form 1040 takes 12 hours and costs $290 on average, and recordkeeping is the largest part of that time.³⁵ Taxpayers with business income average about 21 hours and $610. These are national averages from the IRS's Paperwork Reduction Act model in the 2025 instructions. That recordkeeping is what the professional waits for.

How many tax returns do professionals file?

Slightly more than last year, and more than taxpayers e-file on their own. Tax professionals e-filed 75,316,000 individual returns in the 2026 season through May 8, against 65,730,000 that taxpayers e-filed themselves (IRS). In fiscal year 2024, 57% of individual taxpayers used a paid preparer (GAO, from IRS data). Most of those returns started with documents the client had to send.

How many returns did the IRS process in fiscal year 2025?

The IRS processed 271.4 million federal tax returns and supplemental documents in fiscal year 2025, including almost 162.8 million individual income tax returns (IRS Data Book, June 2026).³⁶ 93.7% of individual returns were filed electronically.³⁷ Fiscal year 2025 ran from October 2024 to September 2025, so it mostly covers the 2025 filing season.

How many 2026 returns did tax professionals e-file?

The IRS received 144,992,000 individual returns in the 2026 season through May 8, 0.6% fewer than at the same point in 2025.³⁸ Tax professionals e-filed 75,316,000 of them, up 0.6%, and self-filers e-filed 65,730,000, up 1.7%. The weekly tables count returns, not taxpayers.

How many taxpayers use a paid preparer?

Over 85 million individual taxpayers, 57% of the total, used a paid preparer in fiscal year 2024. The figure is IRS data, cited by the Government Accountability Office in February 2026.³⁹ For more than half of filers, the return depends on someone else getting their documents.

Law firms: how do clients send documents, and how fast do firms answer?

Email is the default channel: 92% of lawyers use it with clients, and fewer than half work at firms that offer secure online document sharing. No source measures how the documents themselves arrive. Many firms are also slow to answer new inquiries.

How many law firms let clients share documents securely online?

35% of lawyers said their firm offers clients "a secure client portal," and only 12% regularly use one to communicate with clients. 53% said their firm does not offer one (ABA 2023 Websites & Marketing TechReport).⁴⁰ The data comes from the ABA Legal Technology Survey of private-practice lawyers. The ABA has not republished these questions after 2023.

Firm size is a clear divide. 25% of solo respondents say their firm offers one, compared with 66% of respondents at firms with 500 or more lawyers. Small firms are where the gap sits.

Firms that offer clients a secure "portal", by size

Share of respondents, ABA Legal Technology Survey, 2023

0%

25%

50%

75%

100%

Solo

25%

2-49 lawyers

~31%

50-99 lawyers

41%

100-499 lawyers

50%

500+ lawyers

66%

Source: ABA 2023 Websites & Marketing TechReport ("approximately 31%" for 2-49)

Online document sharing more broadly follows the same curve. In 2023, 48% of respondents said their firm offered document sharing and 45% said it did not, with solos lowest at 36%.⁴⁰ In 2022, the figure was 40%.⁴¹

How do lawyers communicate with clients?

Email first. 92% of lawyers use email with clients, 86% the telephone and 18% text messaging (ABA, 2023).⁴⁰ Face to face follows at 45%, videoconferencing at 37%, and postal mail at 19%.

What do legal clients expect?

In 2020, 69% of consumers said they prefer working with a lawyer who can share documents electronically "through a web page, app, or online portal" (Clio Legal Trends Report, vendor, 2020).⁴² In 2019, 79% of legal consumers said they expected a response within 24 hours (Clio, 2,000 US consumers).⁴³ Both are vendor surveys, and both are several years old: read them as what clients said they expected in those years. They measure different people in different years, so I don't compute a gap between them and the ABA figures above.

How fast do law firms answer new clients?

Often not at all. In a 2024 secret shop of 500 law firms, only 33% responded to an email inquiry, down from 40% in 2019, and 48% were essentially unreachable by phone (Clio, vendor, 2024).⁴⁴ The shops ran from June 20 to July 5, 2024. Slow replies plausibly slow the documents too, though no study measures it.

Mortgage and lending: how often do files stall on documents?

Often enough to measure. HMDA-reporting lenders close hundreds of thousands of US applications a year because the file was never completed, 16% of borrowers redid paperwork, and a purchase loan still takes over a month to close.

How many mortgage applications are closed because the file is incomplete?

In 2025, HMDA-reporting lenders closed 695,006 US mortgage applications because the file was incomplete, 5.9% of all applications that reached a decision (DokuTrak analysis of public CFPB HMDA data). For refinances, it was about 1 in 12. By purpose, the rate was 8.2% for refinances, 8.3% for cash-out refinances and 3.7% for home purchases. The overall rate went from 5.3% in 2023 to 5.4% in 2024 to 5.9% in 2025.

HMDA is the Home Mortgage Disclosure Act, the federal law that requires most mortgage lenders to report every application and what happened to it. The CFPB publishes the data each year, and "closed for incompleteness" is one of the outcomes lenders must choose from.

Mortgage applications closed for incompleteness, 2025

Share of US applications with a final decision (HMDA action taken 1 to 5)

All applications

5.9% (695,006)

Home purchase

3.7%

Refinance

8.2%

Cash-out refinance

8.3%

All applications

by year

5.3%

2023

5.4%

2024

5.9%

2025

"Closed for incompleteness" = the lender sent a written notice and the file

was not completed in time. HMDA-reporting lenders only; HMDA does not record which

documents were missing. Withdrawn applications are not counted.

Source: DokuTrak analysis of public CFPB/FFIEC HMDA data, pulled 2026-09-25

How many borrowers had to redo paperwork or delay closing?

16% of borrowers redid or refiled paperwork because of processing delays, and 21% had a closing delay, according to a February 2025 Federal Reserve Bank of Philadelphia report.⁴⁶ It links the National Survey of Mortgage Originations to administrative data for 38,648 borrowers with loans from 2013 to 2021. The same report found that delays at origination cut the likelihood of refinancing in response to rates "by more than half (by 5.7 percentage points)." A slow file costs the borrower later, not only at closing. My mortgage document checklist lists what lenders usually ask for.

How do borrowers send documents today?

69% of borrowers "used an online portal for uploading documents electronically" (ICE Mortgage Technology, 2025 Borrower Insights Survey).⁴⁷ The survey covered 2,002 US adults, including 1,000 homeowners with a mortgage taken in the last 10 years. ICE sells mortgage technology to lenders, so this is a vendor survey.

How long does a mortgage take to close in 2026?

The average purchase loan closed in 36.8 days in March 2026, down from 37.3 days a year earlier. That is the fastest since ICE began tracking the metric in 2019 (ICE Mortgage Monitor, May 2026, vendor data from ICE's loan-level origination records).⁹ The typical purchase loan moved from application to rate lock in 11 days, then from lock to closing in another 26. ICE does not attribute the change to AI, and neither do I.

Do slow onboardings cost financial firms clients?

Yes, by their own account, in financial services broadly. Fenergo surveyed 600 senior decision-makers at financial firms in the UK, US and Singapore in 2025. 70% said they lost clients in the past year due to inefficient onboarding, and abandonment rates average around 10% (Fenergo, vendor, 2025).⁴⁸

Is it safe to collect client documents by email?

For many firms, plain email does not meet the rules. US regulation requires covered firms to encrypt customer information in transit. The IRS tells tax pros not to send client files by unencrypted email. And the IRS warns that fake "document request" emails are used to target tax professionals. This is a summary of public sources, not legal advice.

What do the FTC and IRS require?

The Safeguards Rule is the FTC regulation that sets data security duties for non-bank financial institutions. It treats tax preparers as one of them: "An accountant or other tax preparation service that is in the business of completing income tax returns is a financial institution." The rule's scope list also includes mortgage brokers and tax preparation firms.⁴⁹ It requires firms to "protect by encryption all customer information held or transmitted by you both in transit over external networks and at rest" (16 CFR 314.4(c)(3)).⁴⁹ The rule's exception: if encryption is infeasible, a firm may use effective alternative compensating controls reviewed and approved by its Qualified Individual.

IRS Publication 4557 is blunter: "Send only password-protected and encrypted documents if you must share files with clients via email or use Secure File Transfer Protocol (SFTP) to transmit files instead of email."⁵⁰ My guide on how to send and receive documents securely turns that into steps.

How often are accounting and law firms breached?

Professional-services firms had 478 publicly reported data compromises in 2025, up 39% from 344 in 2024 and up 162% from 182 in 2021 (Identity Theft Resource Center, 2025 Annual Data Breach Report).⁵¹ The ITRC notes these firms are "increasingly used as 'stepping stones' to access multiple corporate clients."

Through spring 2024, the IRS had received reports of nearly 200 tax professional data incidents potentially affecting up to 180,000 clients (IR-2024-180, the latest published figure).⁵² Among lawyers, 29% said their firm had experienced a security breach, 19% did not know, and 52% said it had not (ABA 2023 Cybersecurity TechReport).⁵³ The ABA's definition of breach includes lost devices and break-ins.

Across all industries, the ITRC tracked 3,322 US data compromises in 2025, up 5% from 3,152 in 2024. Financial services was the most breached industry, with 739 compromises, ahead of healthcare and professional services.⁵⁴ Verizon's 2026 Data Breach Investigations Report, a global dataset, counts 3,578 security incidents in professional, scientific and technical services, 2,558 of them with confirmed data disclosure.⁵⁵ Verizon sells security services, and its data comes from its own caseload and contributing organizations, so treat it as vendor data.

How much does a data breach cost a financial firm?

$6.3 million on average in financial services, against a global average of $4.99 million across industries (IBM, Cost of a Data Breach Report 2026, global sample, vendor).⁵⁶ Only 37% of breached organizations said they encrypt sensitive data both at rest and in transit, the standard the FTC Safeguards Rule sets for tax preparers. Third parties were involved in 48% of the breaches Verizon studied, up 60% from the year before, and phishing was the first way in for 16%.⁵⁵

How much fraud and phishing do Americans report?

The FBI's Internet Crime Complaint Center received 1,008,597 complaints in 2025, with $20.877 billion in reported losses.⁵⁷ 191,561 of them were phishing or spoofing complaints. Consumers also filed 3 million fraud reports with the FTC in 2025 and reported $15.9 billion in losses.⁵⁸ Both counts are complaints people chose to file, so they understate the total.

How long do tax identity theft victims wait?

About 20 months. The National Taxpayer Advocate reported in June 2026 that the IRS was taking about 20 months to resolve identity theft victim assistance cases. More than half a million cases were pending at the end of the 2026 filing season.⁵⁹ A stolen client file is not a one-day problem for the taxpayer.

How do attackers use document requests?

They impersonate them. The IRS's 2026 Dirty Dozen warns that tax professionals "remain targets of 'new client' or 'document request' emails that deliver malicious links or attachments." Scammers also use hacked client inboxes to find "a genuine email from a previous victim's email account sent to their tax professional."⁶⁰ Business email compromise produced $3,046,598,558 in reported losses from 24,768 complaints in 2025 (FBI IC3).⁵⁷ IC3 counts only complaints, so the real total is higher.

How common are misdirected-email incidents?

49% of cybersecurity leaders said they experienced incidents caused by misdirected email (KnowBe4, State of Human Risk 2025).⁶¹ That survey covered 700 security leaders and 3,500 employees in 14 countries, and KnowBe4 sells misdirected-email protection. Among lawyers, 42% said email encryption was available to them, falling to 33.1% for solos (ABA, 2023).⁵³ More on this in why email fails for document collection.

Do reminders actually get people to act?

Yes, modestly. The evidence comes from courts, vaccines and taxes, because no study measures reminders for document requests. Treat everything in this section as an analogy, not a measurement of the chase.

How much do reminders change behavior?

Two New York City field studies tested a clearer summons form and text-message reminders. These interventions cut failures to appear in court by 13 to 21% and led to 30,000 fewer arrest warrants over three years (Fishbane, Ouss and Shah, Science, 2020).⁶² The authors' lab experiments found that criminal justice professionals see missed court dates as "relatively unintentional." That sounds a lot like a client who forgot the request was in their inbox.

In a 689,693-patient randomized trial, reminder texts raised pharmacy flu vaccination by 2.0 percentage points, or 6.8%, and the most effective messages "delivered reminders on multiple days" (Milkman et al., PNAS, 2022).⁶³ In UK field experiments with more than 200,000 people, those who received a reminder letter were "nearly four times more likely to pay their tax bill" than those who did not (Hallsworth et al., NBER, 2014).⁶⁴

How big is the effect at scale?

Smaller than the papers suggest. Two US government nudge units ran 126 randomized trials on more than 23 million people. Their nudges raised take-up by 1.4 percentage points, an 8.0% increase, versus 33.4% in published academic studies (DellaVigna and Linos, Econometrica, 2022).⁶⁵ Expect a modest lift from reminders, not a miracle. The practical side is in how to stop chasing clients for documents and my document request email templates.

What new risks does AI bring to client document intake?

AI is spreading fast, mostly in the work around documents rather than in getting them. Meanwhile clients are using it too, fake documents are getting easier to make, and about a third of tax and legal professionals use AI tools their firm hasn't approved.

How fast did professionals adopt AI?

Fast, in every survey cited here; the four series and their definitions are in the first section. One detail they add: solos lag. 17.7% of solo attorneys said their office used AI-based tools in 2024, against 47.8% at firms of 500 or more lawyers (ABA 2024 AI TechReport).²

What do firms actually use AI for?

Mostly writing and summarizing. Among professionals at firms using AI, 75% personally use it for drafting client emails and communication, and 71% for summarizing documents or meetings. 47% use it for data entry or document extraction, and 9% for tax return preparation (Financial Cents, State of AI in Accounting & Bookkeeping 2026, vendor).⁶⁶ The survey's top uses sit around the documents, not in getting them. Tax firms report the same pattern. In a survey of more than 1,000 US tax professionals, 36% of those describing how they use AI named document analysis, behind advisory projects (44%), tax planning (40%) and compliance research (39%) (CPA.com and Blue J, June 2026, vendors).⁶⁷ If you are comparing options, I keep a guide to ChatGPT for accountants and one on AI tools for lawyers.

Did AI save time?

A little, on average. Factoring in all workers, including nonusers, "workers report generative AI time savings equivalent to 1.6% of all work hours," while adoption rose to 54.6% (Federal Reserve Bank of St. Louis, November 2025, nationally representative survey of US adults 18–64).¹¹ Vendor surveys of AI users report much larger savings. Karbon's respondents say they save about 60 minutes a day, or approximately 21 hours a month, with AI (State of AI in Accounting 2026, 593 professionals on six continents, mostly US, at least 48% of them Karbon users).⁶⁸ Self-reported savings among a vendor's own users and a national average across all workers measure different things. Accountants still report "losing approximately 5 hours per week moving, re-entering, or reconciling information across tools" (Intuit QuickBooks, a vendor, 2026 survey of 725 US accounting professionals).³⁰

What are clients doing with AI?

Using it, and worrying about it. Nearly one in five tax filers used AI in the 2026 season, 11% for the first time (Bipartisan Policy Center poll of 1,200 filers).⁶⁹ More than half of consumers have used or would consider AI for legal questions. 28% were directed by their AI to contact a lawyer, and in 12% of cases an AI convinced them their problem wasn't worth pursuing (Clio, vendor, 2025).⁷⁰

Trust lags. 71% of Americans think AI will make their personal information less secure (Pew Research Center, 5,119 adults, 2026).⁷¹ 71% of mortgage borrowers say it is "very important" for their lender to tell them when AI is used, while 54% are "completely comfortable" with it (J.D. Power, 10,067 borrowers, 2025).⁷² And 60% of accountants say clients asked for proof of AI data protection always or frequently in the last 12 months (Intuit QuickBooks, 2026).³⁰

Are clients sending AI-generated fake documents?

Some are, and the numbers are still small. Inscribe, a fraud-detection vendor, reports that monthly AI-generated document fraud grew nearly 5x from May to December 2025, from a small base. AI-generated documents were still under 5% of the fraudulent documents it detected, and 1 in 16 documents showed signs of manipulation (Inscribe, 2026).¹² That is a rate among documents lenders and fintechs sent to a fraud vendor, not a population rate. Sumsub, an identity-verification vendor, found AI-assisted forgery rose from 0% to 2% of fake documents in 2025 (global data).⁷³

In mortgages, an estimated 1 in 119 applications show indications of fraud, and "documents altered with AI" are "not always easy to spot on a loan-by-loan basis" (Cotality fraud data, vendor, September 2026).⁷⁴ The same release reports the fraud risk index down 4.6% year on year. The US Treasury's FinCEN warned in November 2024 that "criminals employed GenAI to alter or generate images used for identification documents."⁷⁵ Catching a wrong or altered file at intake is the subject of my guide to checking client files for wrong or unreadable documents.

How many professionals use AI tools their firm hasn't approved?

About a third of tax and legal professionals. 35% of tax and audit firm professionals say they use AI tools their firm hasn't authorized for work. So do 34% of law-firm professionals (Thomson Reuters Institute, vendor research arm, Future of Professionals 2026, respondents in 40 countries).¹³¹⁴ Concern is rising with use: 83% of accounting professionals in Karbon's 2026 survey are concerned about data security, "+7% YoY" in Karbon's words.⁶⁸ 87% of AI-using accounting firms have no formal written AI policy, and 23% have no plans to create one (Financial Cents, 2026, vendor; 87% of 350+ AI-using firms in a survey of 486).⁶⁶ For lawyers, ABA Formal Opinion 512 is explicit: under Model Rule 1.6, "a lawyer using GAI must be cognizant of the duty to keep confidential all information relating to the representation of a client … unless the client gives informed consent."⁷⁶

How many court decisions involve AI hallucinations?

The Charlotin database lists 898 US court decisions on AI-hallucinated material in 2026 through September 30, already more than the 526 it lists for all of 2025 (DokuTrak analysis). It lists 11 for 2023 and 38 for 2024. Decisions involving lawyers, not only self-represented litigants, went from 4 to 14 to 207 to 358. It is a public, CC BY count, and most of its 2026 decisions involve self-represented litigants; its counts also depend on how much it covers. Even so, the lawyer cases show unchecked AI output reaching professional work, which is why the confidentiality and unauthorized-tool numbers above matter.

US court decisions on AI-hallucinated material

By decision year and by the party who relied on the hallucinated content

0

300

600

900

11

2023

38

2024

526

2025

898

2026*

Lawyer involved

Pro se litigant

Other parties

Lawyer involved, by year:

4, 14, 207, 358

• 2026 = January 1 to September 30, a partial year (hatched). Includes 7 decisions

where AI use was alleged, not confirmed (1 in 2025, 6 in 2026).

The database is volunteer-maintained: earlier years can be revised up.

Source: DokuTrak analysis of the Charlotin AI Hallucination Cases Database (CC BY 4.0)

What nobody measures yet

Six questions every firm that collects documents would want answered have no published measurement. Here is the closest proxy, where one exists.

1. Hours spent chasing documents. No source isolates it. Closest: 9.3 hours a week on all client communication, and 1–2 days or more of back-and-forth per engagement (Canopy, 2024).
2. Reminders sent before a client sends everything. No source in any vertical.
3. Share of documents received by email, as a share of documents rather than of firms. Closest: 49% of surveyed accounting professionals say their firm allows email (Canopy, 2024); 92% of lawyers use email with clients (ABA, 2023).
4. Share of delays caused by missing documents. Closest: NATP's qualitative reports, the 695,006 HMDA "incomplete" closures, and the 16% of borrowers who redid paperwork (Philadelphia Fed).
5. Share of client uploads that are wrong, unreadable or incomplete. No neutral source. The "not in good order" figures that circulate are vendor claims (see below).
6. Whether AI shortened any of the above. No before-and-after study exists.

If you have a dataset that answers one of these, I would like to cite it.

Statistics that circulate but I couldn't verify

These numbers appear often in posts about document collection, onboarding and follow-up. I tried to trace each one to its source and couldn't. The table names the claim, not the pages that repeat it.

Real numbers that are often misread

• "62% of breaches involve the human element." The Verizon 2026 DBIR says the human element "was present in 62% of breaches."⁵⁵ That does not mean 62% of breaches are caused by email.
• "74% of billable law-firm tasks can be automated." Clio's 2024 Legal Trends Report says nearly three-quarters of hourly billable tasks are "exposed to AI automation." The full report describes it as a model estimate built by scoring 7M+ time entries with GPT-4, not an observed result.⁴⁴
• "65% of Americans have used an AI chatbot for legal help." The number is real: a Rev survey of 1,002 US adults, run on SurveyMonkey on December 2–3, 2025 and published January 28, 2026. Rev sells transcription and AI tools to legal professionals, so it is a vendor survey. Cite it with its date and sample; it measures consumers asking a chatbot, not lawyers using AI.
• "19 million taxpayers file extensions" is outdated. The IRS's latest estimate is "more than 20 million."³²
• "AI adoption" among lawyers depends on the question. 30.2% of attorneys said their office used AI tools in 2024 (ABA); 69% of legal professionals said they personally used general-purpose AI tools for work in 2026 (8am). Different questions, so quote the definition with the number.

Methodology, sources and how to cite this page

I built this page to be the source other pages can cite. Here is how the numbers got here.

Selection. Statistics about collecting information and documents from clients in accounting, tax, law and mortgage lending, the size of the professions and firms that do it, and cross-cutting evidence on security, reminders and AI. Primary sources only: the agency, association, academic paper or company that produced the number, never a roundup that repeats it. US sources come first; non-US sources are labeled. I used the latest edition of each series; older editions appear only as trend points and carry their year.

Verification. I opened the original 59 sources on 2026-09-25 and again on 2026-10-08, and the 39 added sources on 2026-10-25. Each time I matched the number and the surrounding wording to the source.

Since 2026-10-25 a script does the check for every row of the CSV. It fetches the source URL and requires an HTTP 200 answer. Then it requires the exact strings in the row's source_match column, the figure as the source prints it, to appear on that page or in that PDF. For the three DokuTrak analyses, the strings are the raw inputs as the source prints them (the HMDA count, the IRS cumulative counts); the Charlotin row only checks that the CSV download answers, because that database changes daily. The script fetches pages like a browser does: some government sites, such as BLS and GAO, refuse generic link checkers but answer a browser.

All 98 rows passed on 2026-10-25. Four rows point to a copy of their source. Three publishers (Science, PNAS and Business Wire) block automated access, and the NBER page shows only an abstract. The copies: the PubMed abstracts of the Science and PNAS papers, the full-text PDF of the NBER paper, and a verbatim reprint of the 2022 Wolters Kluwer press release. The page quotes what those copies say.

Only the 2024 and 2025 trend points of the 8am legal survey (27% and 31%) could not be confirmed at the primary source; they say "as reported by LawNext" on the page. I left out numbers read off chart images, numbers without a reachable source, and modeled estimates presented as measurements.

The 98 rows are the dataset. Figures quoted in "Statistics that circulate but I couldn't verify" and "Real numbers that are often misread" sit outside it: they are examples of claims, linked where a source exists. No figure on this page comes from DokuTrak's own product data: the product is too young for a sample that would mean anything.

Source types. Every row in the CSV carries one label:

The three DokuTrak analyses. HMDA applications closed for incompleteness, IRS late e-filing by professionals, and AI-hallucination decisions by year. Each has a "How I calculated this" box with the source, filter, numerator, denominator, pull date and limitations. The raw pulls and the script are available on request.

Updates. I re-verify the full page at least quarterly. I also refresh it when a major series publishes: Clio Legal Trends in October, Financial Cents yearly, IRS weekly tables, HMDA yearly, Charlotin continuously. The "Last verified" date changes only on a full re-verification, which now means a clean run of the checking script on every row. Next scheduled refresh: January 2027, after the IRS opens the 2027 filing season.

Corrections. If a number here no longer matches its source, tell me and I will fix it and log the change below.

Changelog. 2026-10-25: 39 statistics added, for 98 in total. New: workforce and wages (BLS), CPA licenses (NASBA), firm counts (BLS employer census), IRS filing volume, paid-preparer use (GAO), breach costs, fraud complaints, identity theft case times and 2026 accounting surveys. Every row re-verified by script. Four sources repointed to accessible copies. The court-reminder figure now quotes the Science abstract (13 to 21%). Dataset markup and an embed snippet added. 2026-10-22: the "65% of Americans" claim moved from "couldn't verify" to "often misread" after its source (Rev, January 2026) was found. 2026-10-08: all 59 statistics re-verified; Charlotin count refreshed (898 US decisions in 2026 through September 30); key findings expanded to 13. 2026-09-25: first publication, 59 statistics, 3 analyses.

Embed the citation. Paste this HTML where you use a number from this page:

<p>Source: <a href="https://dokutrak.com/blog/client-document-collection-statistics">Client Document Collection Statistics (2026)</a>, DokuTrak, last verified 2026-10-25.</p>

Embed a chart. Each chart is a standalone SVG. Swap the file name for any of the five charts above:

<figure>
  <a href="https://dokutrak.com/blog/client-document-collection-statistics">
    <img src="https://dokutrak.com/blog/images/client-document-collection-statistics/chart-2-hmda-incomplete.svg" alt="US mortgage applications closed for incompleteness, 2023 to 2025" width="720" loading="lazy">
  </a>
  <figcaption>Source: <a href="https://dokutrak.com/blog/client-document-collection-statistics">DokuTrak, Client Document Collection Statistics (2026)</a></figcaption>
</figure>

Sources

Sources first retrieved 2026-09-25, or 2026-10-25 for those added that day. Every source re-checked by script on 2026-10-25.

Footnotes

1. ABA Legal Technology Resource Center, "2023 Artificial Intelligence (AI) TechReport" (10.9%, 440 respondents). ↩

2. ABA Legal Technology Resource Center, "2024 Artificial Intelligence TechReport" (April 2025). ↩ ↩²

3. 8am (MyCase/LawPay), "2026 Legal Industry Report" (March 5, 2026). ↩

4. LawNext, reporting the 2024 and 2025 trend points of the 8am (MyCase/LawPay) "2026 Legal Industry Report" (March 2026). ↩

5. Thomson Reuters Institute, "2026 AI in Professional Services Report" (2026). ↩

6. STRATMOR Group, "Laying the Foundation for a Seamless Digital Mortgage Experience" (April 2025), 2024 Technology Insight Study. ↩

7. Financial Cents, "2024 State of Accounting Workflow Automation Report" (2024). ↩ ↩²

8. Financial Cents, "2025 State of Accounting Workflow & Automation Report" (2025), p.14 and p.62. (full report PDF) ↩ ↩² ↩³

9. ICE, "May 2026 Mortgage Monitor" (May 11, 2026). ↩ ↩²

10. National Association of Tax Professionals, "Early post-tax day feedback highlights what created filing challenges," press release via National Law Review (April 30, 2026). ↩ ↩²

11. Alexander Bick, Adam Blandin and David Deming, "The State of Generative AI Adoption in 2025," Federal Reserve Bank of St. Louis (November 2025). ↩ ↩²

12. Inscribe, "2026 State of Document Fraud Report" (January 2026). ↩ ↩²

13. Thomson Reuters Institute, "Future of Professionals 2026: Tax and accounting report" (2026). ↩ ↩²

14. Thomson Reuters Institute, "Future of Professionals 2026: Legal report" (2026). ↩ ↩²

15. U.S. Bureau of Labor Statistics, "Occupational Employment and Wages — May 2025," Table 1, national employment and wage data by occupation (May 15, 2026). ↩

16. U.S. Bureau of Labor Statistics, Occupational Outlook Handbook, "Accountants and Auditors," 2025–35 projections. ↩ ↩²

17. U.S. Bureau of Labor Statistics, Occupational Outlook Handbook, "Bookkeeping, Accounting, and Auditing Clerks," 2025–35 projections. ↩ ↩²

18. U.S. Bureau of Labor Statistics, Occupational Outlook Handbook, "Lawyers," 2025–35 projections. ↩

19. U.S. Bureau of Labor Statistics, Occupational Outlook Handbook, "Paralegals and Legal Assistants," 2025–35 projections. ↩

20. U.S. Bureau of Labor Statistics, Occupational Outlook Handbook, "Loan Officers," 2025–35 projections. ↩

21. National Association of State Boards of Accountancy, "How many CPAs are there?" (as of August 27, 2026). ↩

22. IRS Return Preparer Office, "Federal tax return preparer statistics" (data as of September 1, 2026). ↩

23. AICPA & CIMA, "Accounting firms report strong hiring outlook, AICPA report finds," on the 2025 Trends report (October 27, 2025). ↩

24. U.S. Bureau of Labor Statistics, Quarterly Census of Employment and Wages, 2025 annual averages, NAICS 541211 Offices of certified public accountants, US private sector (CSV). ↩

25. U.S. Bureau of Labor Statistics, Quarterly Census of Employment and Wages, 2025 annual averages, NAICS 541213 Tax preparation services, US private sector (CSV). ↩

26. U.S. Bureau of Labor Statistics, Quarterly Census of Employment and Wages, 2025 annual averages, NAICS 5411 Legal services, US private sector (CSV). ↩

27. American Bar Association, "ABA 2025 Profile of the Legal Profession" (December 2025). ↩

28. Wolters Kluwer, "New Wolters Kluwer annual accounting survey reveals how technology is helping firms tackle top 5 challenges and achieve 2023 goals," Business Wire release (December 2, 2022), verbatim reprint. ↩

29. Canopy and CPA Practice Advisor, "Survey Results Are In: Charting the Future of Accounting" (September 15, 2024). ↩ ↩²

30. Intuit QuickBooks, "2026 Accountant Technology Survey" (2026). ↩ ↩² ↩³ ↩⁴

31. Thomson Reuters Institute, "2026 State of Tax Professionals Report" (June 2026). ↩

32. IRS, "IRS reminds taxpayers who filed for extensions of the Oct. 15 deadline," IR-2025-104 (October 14, 2025). ↩ ↩²

33. IRS, "Filing season statistics for week ending April 11, 2025.". ↩

34. IRS, "Filing season statistics for week ending Oct. 17, 2025.". ↩

35. IRS, "2025 Instructions for Form 1040.". ↩

36. IRS, "IRS describes agency's activities in fiscal year 2025 Data Book" (June 2026). ↩

37. IRS Data Book, FY 2025, "Returns filed, taxes collected and refunds issued.". ↩

38. IRS, "Filing season statistics for week ending May 8, 2026.". ↩

39. U.S. Government Accountability Office, "Paid Tax Return Preparers," GAO-26-108723 (February 2026). ↩

40. ABA Legal Technology Resource Center, "2023 Websites & Marketing TechReport" (January 2024). ↩ ↩² ↩³

41. ABA Legal Technology Resource Center, "2022 Websites & Marketing TechReport.". ↩

42. Clio, "Highlights from the 2020 Legal Trends Report" (2020). ↩

43. Clio, "Clio's Legal Trends Report reveals law firms struggle to respond to client inquiries," press release (October 21, 2019). ↩

44. Clio, "2024 Legal Trends Report," press release (October 7, 2024). (secret-shop dates) ↩ ↩²

45. CFPB / FFIEC, HMDA Data Browser API, nationwide aggregations, 2023–2025 data (pulled 2026-09-25). ↩

46. Neil Bhutta and Paulina Zeballos Doubinko, "Do Mortgage Borrowing Experiences Differ by Race and Ethnicity?", Federal Reserve Bank of Philadelphia (February 2025). ↩

47. ICE Mortgage Technology, "2025 Borrower Insights Survey" (2025). ↩

48. Fenergo, "Financial Crime Industry Trends 2025," press release (October 7, 2025). ↩

49. FTC Standards for Safeguarding Customer Information, 16 CFR Part 314 (eCFR, current as of 2026-09-01). ↩ ↩²

50. IRS, Publication 4557, "Safeguarding Taxpayer Data" (Rev. 6-2024). ↩

51. Identity Theft Resource Center, "2025 Annual Data Breach Report" (January 2026). ↩

52. IRS, "Tax pros continue to be targeted by identity thieves," IR-2024-180 (July 2, 2024). ↩

53. ABA Legal Technology Resource Center, "2023 Cybersecurity TechReport.". ↩ ↩²

54. Identity Theft Resource Center, "2025 Annual Data Breach Report" summary (January 29, 2026). ↩

55. Verizon, "2026 Data Breach Investigations Report" (May 2026). ↩ ↩² ↩³

56. IBM, "Cost of a Data Breach Report 2026," press release (July 29, 2026). ↩

57. FBI Internet Crime Complaint Center, "2025 IC3 Annual Report.". ↩ ↩²

58. Federal Trade Commission, prepared statement on "The Rising Scam Economy" before the Joint Economic Committee (March 25, 2026). ↩

59. National Taxpayer Advocate, "National Taxpayer Advocate issues mid-year report to Congress," IR-2026-79, on the Fiscal Year 2027 Objectives Report (June 24, 2026). ↩

60. IRS, "Dirty Dozen tax scams for 2026," IR-2026-30 (March 5, 2026). ↩

61. KnowBe4, "State of Human Risk 2025" (December 2025). ↩

62. Alissa Fishbane, Aurelie Ouss and Anuj K. Shah, "Behavioral nudges reduce failure to appear for court," Science 370 (2020), abstract on PubMed. ↩

63. Katherine L. Milkman et al., megastudy of text-message nudges to encourage flu vaccination at Walmart pharmacies, PNAS (2022), abstract on PubMed. ↩

64. Michael Hallsworth, John A. List, Robert D. Metcalfe and Ivo Vlaev, "The Behavioralist As Tax Collector," NBER Working Paper 20007 (2014), full text. ↩

65. Stefano DellaVigna and Elizabeth Linos, "RCTs to Scale: Comprehensive Evidence From Two Nudge Units," Econometrica 90(1) (2022). ↩

66. Financial Cents, "The State of AI in Accounting & Bookkeeping 2026" (2026). ↩ ↩²

67. CPA.com and Blue J, "Blue J and CPA.com survey finds AI adoption among tax firms has nearly doubled in one year" (June 8, 2026). ↩

68. Karbon, "The State of AI in Accounting Report 2026" (January 2026). ↩ ↩²

69. Bipartisan Policy Center, "Takeaways from BPC's 2026 Tax Filing Season Poll" (2026). ↩

70. Clio, "2025 Legal Trends Report" (October 2025). ↩

71. Pew Research Center, "Americans and AI 2026" (June 17, 2026). ↩

72. J.D. Power, "2025 U.S. Mortgage Origination Satisfaction Study" (November 12, 2025). ↩

73. Sumsub, "Identity Fraud Report 2025–2026," press release (November 25, 2025). ↩

74. Cotality, "Annual Mortgage Fraud Risk Report" press release (September 15, 2026). ↩

75. U.S. Department of the Treasury, FinCEN, Alert FIN-2024-Alert004 on deepfake media (November 2024). ↩

76. American Bar Association, "ABA issues first ethics guidance on a lawyer's use of AI tools," Formal Opinion 512 (July 29, 2024). ↩

77. Damien Charlotin, AI Hallucination Cases Database, CC BY 4.0. (CSV download, pulled 2026-10-08) ↩