Investigator-initiated study management, explained.

What happens after a drug company says yes to a doctor's study idea

One funded study, from a signed contract to its final report

1. The proposal is approved and the contract signedScientific review is finished. The agreement fixes the budget, the payment schedule, the enrollment plan and the product quantity. KESTREL-IgAN's was signed March 3, 2025.
2. The baseline is confirmedSeven dates become the plan: contract, first patient, interim analysis (a planned early look at the data), last patient in, last visit, final study report and manuscript.
3. The study is run, quarter by quarterReports are read, enrollment is compared with plan, product ships, invoices are matched and payments classified. This is where the 9 against 16 is found, and it is the subject of this guide.
4. The plan changes when the study changesAmendments move dates and money, and a study that drifts gets a recovery plan or, sometimes, an early close.
5. The study closesFinal report, manuscript, registry results, product reconciled, last payment made and every payment and shipment reported as a transfer of value.

Step 3 is the center of this guide. The people, companies and figures in this example are fictional.

Four surprises about funded studies: the sponsor, the vial's value, the recruitment gap and a missing count

Baseline, drift, milestone, in-kind: the words this work runs on

Fourteen terms are defined in the one-page summary.

Who holds which duty when a doctor sponsors the study and a company funds it

In shortThe sponsor answers for the study. In an investigator-initiated study that is normally the doctor or the institution. The company answers for what it promised: money, product and its own reporting. Every part of an ESR manager's job sits on one side of that line, and the research agreement draws it.

The US definition is short. A sponsor is the person who "takes responsibility for and initiates a clinical investigation". A sponsor-investigator is an individual who "both initiates and conducts" it, under whose immediate direction the drug is given, and the term excludes any person other than an individual.1 ICH E6(R3), the international good clinical practice guideline, uses nearly the same words and adds that the obligations "include both those of a sponsor and those of an investigator".6

When a hospital or university takes the role, it is the sponsor and the doctor is its investigator.

The European definition puts the same idea differently. A sponsor is "an individual, company, institution or organisation which takes responsibility for the initiation, for the management and for setting up the financing of the clinical trial".7 That word, financing, is one reason a European agreement needs care. It has to be clear who set up the money and who merely gave some.

Companies say this in public. One device maker's research program page states it is a "supporter" of such studies "but is not the study 'sponsor'", and that the sponsor-investigator "is responsible for conducting the study and assuring the validity of the study data".8

The words get used loosely. A short 2024 paper separates them: "initiated" is about who conceived the study, "sponsored" about who takes on the regulatory responsibilities, and the second word is "often used inaccurately to describe the funding source".9 Check which arrangement your agreement describes before you apply anything in this guide.

Two ways a marketed product gets studied without the company's own IND

A US study of a drug usually needs an investigational new drug application, an IND, unless it meets a list of conditions. A study of a lawfully marketed drug is exempt only if it will not support a new indication or a significant labeling change, will not support a significant advertising change, does not significantly raise the risks of the drug through its dose, route or patient group, follows the ethics committee and consent rules, and stays within the ban on promoting an investigational drug.13

KESTREL-IgAN takes the other route: it runs under an IND held by the investigator's side. The doctor rarely writes the whole file from scratch. FDA's guidance for sponsor-investigators, a May 2015 draft that is still a draft, describes a letter of cross-reference authorization: the company lets FDA read the manufacturing and earlier study information in its own file, so the doctor need not supply it again.14 The letter gives FDA access, and the same draft notes that it does not give the sponsor-investigator a right to read the company's confidential content.

An ESR manager often has to track this letter as one more item in the contract's list of commitments.

The rules around a funded study, and where each one stood in November 2026

In shortNo single rule covers investigator-initiated studies. Good clinical practice, FDA's IND rules, the EU Clinical Trials Regulation, CMS Open Payments, fraud and abuse guidance, and the results rules each govern one piece. Several moved in 2025 and 2026.

Two different rules share the number 11. 42 CFR Part 11 covers ClinicalTrials.gov. 21 CFR Part 11 covers electronic records and signatures, and section 13 returns to it.

Eight documents that carry a funded study, and the numbers buried in each

In shortAlmost every number an ESR manager needs already exists on paper, written by somebody else: enrollment in a progress report, lots in a packing list, money in an invoice. The work is getting each number out of its document, tying it to the agreement and keeping a pointer back to the page.

Even a modest portfolio creates a lot of these. Take 40 studies, four reports a year each, four invoices each, a shipment every other month and a handful of amendments. That is well over 500 documents a year for one person, and our arithmetic is illustrative: use your own volumes. The sample company in this guide estimates about 2,400 reports, 1,100 invoices and 380 shipments a year across 640 studies, and those figures are fictional.

Each document also has a home outside the ESR team. The agreement sits in contract management, the invoice in accounts payable, the lot in clinical supply, the report in a grants system or a mailbox. So the study's true state lives in five places, and the spreadsheet is somebody's attempt to pull it together.

The research agreement and the company's quality system list further duties for each party during the study. They are the authority on those duties, and a team should follow them as written. This guide stays with dates, product, money and deliverables.

Working out how late a study will be from one quarterly report

In shortTake the patients still needed, divide by the recent enrollment rate, and add that to today. For KESTREL-IgAN that is 15 patients at 1.17 a month, which lands on October 25, 2027, 30 weeks after plan. The method is simple. The judgment lies in which rate you believe, and in reading why patients failed screening.

Here is the calculation on the example's numbers. The plan was 24 patients between October 1, 2025 and March 31, 2027, a straight line of 16 by September 30, 2026. The report says 9.

Why one enrollment rate is never the whole answer

The rate you pick moves the date by months. Over the whole study so far, 9 patients in under 11 months is about 0.84 a month. Over the last quarter alone it is 1.33. And a recovery plan changes the rate itself, so a forecast is a statement of what happens if nothing changes. A range says that out loud: the sample forecast gives August 2027 to January 2028.

Reading the screen failures, where the fix usually hides

A screen failure is a patient checked for the study who does not qualify. Page 2 of the report says 14 patients were screened and 5 failed, and three of the five failed because their kidney function, measured as eGFR, fell below the study's entry limit of 45. Five of 14 is 36%. That single line suggests a question for the investigator: would widening the limit to 30 fit the study's science? Page 3 adds that four patients are in screening and that the clinic's referrals come only from its own patients.

The two findings together point to a recovery plan: more referrals, a second site, and a conversation about the entry criterion. The investigator decides, because the protocol is hers.

The evidence that investigator-led trials miss enrollment more often, in three data sets

In shortStudies from 2014 and 2015 found failed recruitment far more often outside industry. Our own count of ClinicalTrials.gov for 2015 to 2021 starts, using a stand-in group, shows the same pattern: about the same share terminated, twice the share of those terminated citing low enrollment, and many more records left with unknown status.

Carlisle and colleagues examined 2,579 phase 2 and 3 trials closed in 2011. Nineteen percent, 481 trials, were terminated for failed accrual, meaning too few people enrolled, or finished under 85% of planned enrollment. Non-industry funding was among the factors tied to failure, at 25% against 16%.32

Kasenda and colleagues followed 1,017 randomized trials from ethics committee files in three countries. A quarter were discontinued. Poor recruitment was the most common reason, at 26.5% for investigator-sponsored trials and 8.4% for industry-sponsored ones.3 Both studies are old. We wanted a current check, so we asked ClinicalTrials.gov directly.

What the registry shows for 2015 to 2021 starts, read in November 2026

The registry has no field called investigator-initiated. Our closest proxy is an interventional study whose lead sponsor is classed as "other", meaning a university, hospital or individual, with an industry collaborator. It misses a funded study that names no company as a collaborator, and it takes in other partnerships. We compared those 9,896 studies with 35,255 led by industry. Everything in this part of the section is our count.33

Terminated studies run at about the same share in both groups, 12.4% and 12.0%. The reasons differ. Among terminated studies with the investigator in the lead, 567 of 1,223 used the words enrollment, accrual or recruitment when explaining why, or 46.4%, against 910 of 4,245 among industry-led studies, 21.4%. That is our count. It reads only what sponsors typed, 198 of those records give no reason, and a study with low enrollment described in other words is missed, so treat those shares as an indication.

The share in unknown status is the finding that most concerns an ESR lead. It reads 16.3% against 9.9%, and each of those records is a study whose public entry stopped being kept up.

Study size explains part of the fragility. A review of 5,834 kidney, bladder and prostate cancer trials found most investigator-initiated trials were open-label phase 2 studies of 20 to 49 people, open-label meaning everyone knows which treatment each patient gets.34 In a study of 24, a single site's staffing change can move the whole forecast. Money is the other part: 48 people interviewed in Switzerland described a systemic problem of these trials being underfunded, with budgets often inaccurate because of poor planning and limited oversight.35

Three fulfillment routes, one expiry date and the records behind every vial

In shortProduct reaches a funded study by one of three routes, depending on how it must be labeled. For each shipment someone must check the request against the approved total, capture the lot and expiry from the reply, make sure the lot outlasts the study and collect the site's receipt. All of it has to be on record.

Choosing among clinical kits, commercial packs and relabeled packs

Label wording is a rule. The immediate package of a US investigational drug "shall bear a label with the statement 'Caution: New Drug—Limited by Federal (or United States) law to investigational use.'"21 In the EU, authorized products used in a trial may be labeled under the trial rules or under the normal marketing rules, with additional particulars on the pack where the protocol's circumstances require them.25 If a study runs under an IND exemption, ask regulatory affairs what your packs need.

Checking a lot's expiry against today's forecast, and again when it moves

The example's packing list shows 40 vials, lot RQ26-1187, expiry September 30, 2028. That date looks fine on its own, two years out. It stops looking fine beside the forecast: enrollment has slipped, so the last patient's last visit has moved to November 20, 2028, seven weeks after the lot expires. Against the original plan, April 26, 2028, the lot would have been fine. So this check has to run against today's forecast, and it has to be rerun whenever the forecast moves.

Whose records prove where each vial went

In an investigator-held IND, FDA's rules put these records on the sponsor-investigator. A sponsor keeps records "showing the receipt, shipment, or other disposition" of the drug, with the investigator's name and the date, quantity and batch of each shipment.22 An investigator keeps records of the "dates, quantity, and use by subjects", and returns or disposes of unused supply when a study ends.10,36

E6(R3) says responsibility for product accountability rests with the investigator or institution, and that a sponsor may facilitate it.6 The company keeps its own shipping records, because it shipped the product and because it may have to report it as a transfer of value. In FDA's fiscal 2024 inspection summary, failing to maintain accountability records appears among the common sponsor-investigator observations.37

The study's agreement decides what the site reports back about stock on hand, so the company can see a shortage coming. Record retention runs two years after approval, or two years after the investigation ends if no application follows, under US rules, and 25 years for the master file under the EU regulation.10,24

Paying for research only when the work that earns it is on file

In shortA research payment should be fair market value for real work, kept away from sales, and tied to evidence. Each invoice line needs a line in the schedule, a document that proves the event happened, and a check of what was already paid and the cap. The government has treated research grants without that discipline as kickbacks.

E6(R3) asks only that "the financial aspects of the trial" be documented in an agreement between sponsor and investigator or institution.6 The harder rules come from fraud and abuse law.

The 2003 guidance for drug makers from the US health department's Office of Inspector General, OIG, said payments for research services "should be fair market value for legitimate, reasonable, and necessary services". It said grants linked to product purchases can be misused, and that "manufacturers should insulate research grant making from sales and marketing influences".

It also listed research "initiated or directed by marketers" and post-marketing research used as a pretense to promote product among the signs of questionable research.28 That text now sits in OIG's archive, and no replacement for drug makers has a date.29 Compliance teams still build on it.

Tying each invoice to a milestone, with the arithmetic

KESTREL-IgAN's sample contract is worth $684,800: seven milestones totaling $524,000, plus $6,700 per enrolled patient, up to 24, which is $160,800. Here is where the money stood on October 28, 2026.

Why these budgets drift: what the real costs of 93 trials showed

Griessbach and colleagues collected real costs from 93 investigator-sponsored randomized trials in Switzerland, Germany and the UK. The median total cost was $645,824, and over a quarter of it went on planning, which is typically not fully funded. Of the 66 with an original budget available, 46 exceeded it by more than 50%.40 So a budget amendment request may trace back to a budget that was never realistic.

Separation of duties is the plain control behind all this. The person who reviews an invoice should never approve it. Holding a payment is also a legitimate tool, provided the agreement allows it: a recovery plan can hold the next milestone until a new site opens, and the investigator should hear about the hold from a person, in advance.

What Open Payments and the EFPIA Code ask of a research payment and a shipped vial

In shortUS makers report research payments to CMS once a year, by the 90th day, with the institution, the study, the product and the principal investigators. CMS answers that study drug needs no specific value. Europe's EFPIA Code discloses R&D transfers of value in aggregate. Classifying each record when it is approved saves the year-end hunt.

The fields a research record needs, and the special rules that apply

Any payment "made in connection with an activity that meets the definition of research" and subject to a written agreement or protocol follows special rules. It is reported separately, with the name of the receiving institution or person, the total research payment "including all research-related costs", the study name, the related product, the principal investigators and, optionally, the ClinicalTrials.gov number.11

CMS allows up to five covered principal investigators per payment, and a payment to a non-teaching institution is still reportable when a physician who is a covered recipient, the rule's term for the clinicians and teaching hospitals it tracks, is a principal investigator.2

Publication may be delayed for new products and new applications of existing ones, for up to four years or until FDA approval.26 Reports are due by the 90th day of the year, so Program Year 2026 is due March 31, 2027.17

By our count, in Program Year 2025 the in-kind share came to $1.91 billion on 128,187 records, against $7.58 billion in cash on 803,772 records.5 These totals include every kind of research a maker funds. They show the scale of in-kind support and say nothing about funded studies on their own. Small-payment thresholds of $13.82 and $138.13 for 2026, and $14.31 and $143.01 for 2027, apply to very small transfers, so they rarely touch a research record.18

The penalties, and the one case that set the record

Failing to report brings civil penalties of $1,443 to $14,432 per item, capped at $216,490 a year, and for knowing failures $14,432 to $144,329, capped at $1,443,275. Those are the 2025 levels, and HHS made no adjustment for 2026.41,42 In August 2026 the Justice Department announced a Veloxis settlement of more than $46 million, including a $1.55 million penalty it called "the largest Sunshine Act recovery since the law was passed in 2010". That case concerned meals and attendee lists, and it shows that reporting records get read closely.43

Final reports, manuscripts and registry results: the dates and who owes each one

In shortFunded studies owe a chain of deliverables, some to the company under contract and some to the public under law. A sponsor-investigator normally owes the registry results personally. FDA has now sent 11 public notices, 2 of them to individual physicians, and by our count one in six records in our registry group has gone quiet.

On the US side, the responsible party is the sponsor, unless and until a principal investigator has been designated.12 When a doctor holds the IND, the doctor normally is that party. Status changes must reach the registry within 30 calendar days, and so must the actual primary completion date, the day the last patient has the final exam or treatment used to collect the main outcome's data.31 The company doesn't own these filings, but a missing result reflects on the product and the program, so ESR teams track them as if they did.

A Notice gives 30 days to fix the problem. After that FDA may seek penalties of up to $15,107, plus up to $15,107 for each day the violation goes uncorrected, at 2025 levels.41,19 In Germany, a late or missing Article 37 document can bring a fine of up to 25,000 euros, though we found no published case of one.46

The wider record is poor: in a 2020 Lancet study, 1,722 of 4,209 trials due to report, 40.9%, met the one-year deadline, and 63.8% had reported at any time. Industry sponsors had about three times the odds of complying that non-industry, non-government sponsors had.47

For the ESR team the practical move is to compute each date from the current forecast, and to chase it through the investigator. Our count of registry records, method in section 6, found results posted for 36.1% of completed studies in the proxy group and 44.3% of industry-led ones, which is no legal test, since many of these are not applicable clinical trials.33 It does point to where to look first.

A final report needs reading too. Check it against the protocol's objectives, section by section. A report missing the table for a secondary endpoint, a lesser outcome the study measures, is an unfinished deliverable, and the contract's milestone usually says "accepted", which gives the company grounds to wait for the missing section.

Re-baselining after an amendment, and closing a study that cannot recover

In shortA protocol amendment is a scientific document with financial consequences: dates, budget, product and contract may all move. Someone has to work those out and get them approved. And some studies should stop. A clean close needs the enrollment history, recovery attempts, final payment, product reconciliation and registry updates in one place, with two signatures.

One amendment worked through: twelve more months of follow-up in a rare kidney disease

GLOM-DURA is a sample study of Glomira in 24 adults with C3 glomerulopathy, a rare kidney disease. Amendment 3 extends follow-up from 12 to 24 months after the core period, with two more visits, to study how long the response lasts. Working out the new dates, budget and product, and getting them approved, is called a re-baseline.

Because the budget moves more than the sample company's 10% rule, the contract needs an amendment as well. Amendments are common everywhere: a Tufts study of phase I to IV protocols found 76% had at least one, up from 57%, with an average of 260 days from identifying the need to the last oversight approval, and 215 days during which sites ran under different versions. The authors deemed 77% of amendments unavoidable.48 Those figures come from company-run protocols; no one publishes the same for funded studies.

An early close on a pediatric study that reached 3 of 20 patients

GLOM-KIDS, a sample study in children with C3 glomerulopathy, had enrolled 3 of 20 in 22 months. Two recovery attempts failed: a second site withdrew, and an amendment widening the age range never went in. On October 14, 2026 the investigator agreed to stop enrolling and let the three children finish their 52 weeks. The case has to balance on every line.

An early close carries its own obligations. Status updates reach the registry within 30 days.31 Discontinued trials are more likely to go unpublished, 55.1% against 33.6% in Kasenda's cohort, and only 37.9% of the discontinuations in that cohort were reported to the ethics committees.3 A good close plan names who will write up what was learned. Unused supply is returned or disposed of, with written records of the disposition.36,22

Eight ways a funded study goes wrong after signature, and what each one costs

In shortThe failures are mostly ordinary: a late warning, a payment before the work, a lot that runs out, a classification done from memory in January. No public source counts them, so your baseline has to come from your own files. A sample of 30 closed studies will show you which ones you have.

Why these happen: five causes behind almost every entry on that grid

FDA's inspection summary for sponsor-investigators lists common observations that include failing to maintain the Form FDA 1572, on which an investigator commits to follow FDA's rules, inadequate case histories, not following the plan and missing accountability records.37 Most of those fall on the investigator. Yet a company that funds a study and ships its product has a real interest in whether the file holds up, and a clean paper trail on its own side is what it controls.

What software can read, what people must decide, and what regulators say about AI in this work

In shortSoftware does well at reading documents, matching lines and doing date arithmetic. It should draft recovery plans and leave every decision on money, product, baselines and closes to a named person. No regulator has written AI rules for this work, and the general texts are drafts, delayed, or scoped elsewhere. Part 11 and data integrity rules still apply to the records.

The limits to hold on to when you evaluate document-reading software or a forecast

Reading software can misread a scanned table, so each value needs a link to its page and a person who confirms it. A forecast built on few patients has wide error, so it should show a range, and nobody should present it as a prediction of what the site will do after a change.

Classification follows rules that CMS and your counsel set, and an exception should go to the transparency analyst. Separately, we found no independent test of any ESR software's reading accuracy. Ask a vendor to measure it on your documents.

And keep the line from section 2 in view. Automation helps a supporter see what it paid for. It should not start managing a study the investigator sponsors, so what the software sends a site should stay within what the agreement allows, and should go out only when a person sends it.

How ESR teams run funded studies today: the roles, the hand-offs and the minutes each quarter

In shortAn ESR manager carries the portfolio and hands off to finance, supply, medical and transparency colleagues, mostly by email, around a grants database and a spreadsheet. Most time goes to reading and retyping. We found no public time study, so the numbers below are an estimate for you to replace with your own.

ESR manager to QA auditor: who touches a funded study

Where an ESR manager's minutes go on one study in one quarter

Spreadsheets, grants tools and outsourcing: three ways teams run it

We found no public market size for this category and no reliable count of programs. Commercial software for it exists under the names investigator-initiated study management, grants management and research management.

KESTREL-IgAN's quarter: a three-page report read, a study moved to At risk, a later lot, a signed plan and a payment held

In shortThis is the example from the top of the page, inside the product. Dr. Lindqvist's report arrives at 9:12 am. Agents read it, cite every figure, forecast the last patient 30 weeks late and draft a recovery plan. In the same hour the supply tracer flags a lot. Nadia Brooks asks for a later lot and approves the plan, and milestone 3 is held. Finance approves the invoice as the second person.

1

Nadia opens 38 studies drawn as one river

Each study is one row. Hollow marks are its baselined milestones, solid marks are actual or forecast ones, and the colored stretch between them is the drift, printed in weeks at the end of the row.

On the morning of October 28 she has 2 studies At risk and 10 Drifting. KESTREL-IgAN sits at +11 weeks with a chip that says its Q3 report is in. A line across the top lists the work waiting on her: 4 reports, 3 shipments and 6 invoices, plus a new study, an amendment and a close-out.

2

The quarterly report reader works through the three pages in 6.4 seconds

Seven steps run in view: received, read the enrollment table, compared with the plan, forecast the last patient, moved the band, drafted the recovery, and handed to Nadia.

Each step says what it found. Three steps turn amber: 7 behind plan (44%), the rate of 1.17 a month giving October 25, 2027, and the rule that moves the study from Drifting to At risk.

3

Every figure links back to its page

Screened 14, enrolled 9 of 24, treated 9, completed 2. Each tile carries a page tag. Click one and the passage lights up in the report on the left, so Nadia checks it and types nothing.

The left pane is Dr. Lindqvist's own report, unaltered. Three screen failures for eGFR below 45 and the note about a limited referral base are highlighted, because the recovery plan rests on them.

4

A forecast with a range, a band with its rule, a plan with owners

The chart draws plan, actual and forecast together: last patient in on October 25, 2027, with an 80% range of August 2027 to January 2028, and the interim analysis, final report and manuscript moving with it.

Below it sit the band change with its rule, a message to Dr. Lindqvist and a plan of five actions: a second site by January 15, 2027, referral letters to 6 clinics by November 30, a protocol amendment to the ethics committee by December 15, a monthly call and a hold on milestone 3. The message waits for Nadia, and investigators never see bands or forecasts.

5

In the same hour, a packing list shows a lot that will not last

The vendor's dispatch email and packing list reach the supply mailbox. The supply tracer fills ship date, quantity, lot, expiry, tracking, logger and labels, each with a link to its line, and nothing is typed.

It then checks the lot against today's forecast. Lot RQ26-1187 expires September 30, 2028, seven weeks before the last visit, now forecast for November 20, 2028, so it raises a flag. It also records the shipment as a research payment, in-kind items.

6

Nadia asks for a later lot, and the record refills itself

One click sends a drafted message to the vendor, asking for a lot that expires after November 20, 2028. The reply offers lot RQ26-1342, expiry March 31, 2029, same ship date. The tracer reads it, updates both fields, clears the flag and logs the change.

Nadia confirms the shipment record, and the site sees lot and expiry in its portal.

7

One signature confirms the figures and approves the plan

Nadia signs once, with the statement in front of her: the Q3 figures are confirmed, the five actions go to their owners and milestone 3, $96,000, is held until the second site opens. The signature records her name, the date and time, and its meaning, which here is Approved.

The study page shows the hold straight away, and the investigator's portal shows the milestone on hold until the second site opens.

8

Finance approves the invoice as the second person

Confirming the report makes the per-patient invoice payable: 4 × $6,700 = $26,800, tied to the schedule and to page 2. Nadia marks it reviewed.

She cannot approve it herself. Marcus Webb, the finance approver, sees the line, the evidence and the cap, and signs. The payment joins that afternoon's file to accounts payable, and the transfer of value is already classified.

The solution reads and proposes; named people confirm, approve and sign. It does not change a baseline, release a payment, ship product or message an investigator on its own.

The rest of an ESR manager's week: chased reports, blocked shipments, unearned invoices, re-baselines and closes

In shortBeyond the one case above, the same pattern repeats: agents read and match, then a person decides. Here are reports confirmed in bulk, a request above its approved total, exceptions in payments, a re-baseline, an early close, final report checks, the investigator's view and the administrator's settings.

Reports confirmed together, and the ones that never came

Product requests routed by supply type, and stopped when they exceed the total

Invoices that were not earned, and files that reach accounts payable

A re-baseline with three approvers, and an early close with two

The final report check, the investigator's view and the settings behind it all

At go-live, the portfolio import mapper maps an existing portfolio file's columns to fields, checks each row against its business unit's rules and prepares a dry run. In the sample, 598 studies came in on February 3, 2025.

Seven ESR agents, the five people who sign and what the solution reads and writes

In shortSeven agents do the clerical half of active study management, and each has something it must never do. Five roles confirm, approve and sign, and the person who prepares never approves. Documents come in from the portal, email and your systems, and cited records and files go out.

Report reader to import mapper: each agent's job and its limit

Who signs what: the second-person rule across five roles

The transparency analyst alone can change a classification, with a reason kept in the audit trail, and a second administrator approves each configuration version. A confidence below 0.95 sends a figure to a person. Signatures carry the printed name, the date and time, and the meaning, and the trail exports as an inspection pack. These are controls that support your compliance with 21 CFR Part 11; your validation and procedures complete it.49

From reports, replies and invoices to cited records: what goes in and what comes out

Systems an ESR program already runs, connected by API or flat file

The outcomes this was designed for

These are design targets, to be measured in your own pilot. The sample screens show 640 active studies at the portfolio level. This solution classifies each transfer of value and hands it to your transparency reporting system; the transparency team files the report. Proposal intake and scientific review belong to a separate review solution.

What a late warning, an unearned payment or a lost study costs your ESR program

In shortPublished figures frame the question: how often recruitment fails, how much these trials cost, how often results run late. What matters to your decision is your own count of late warnings, wrong payments and hours spent retyping, and no one publishes those. A week with your closed studies will give you your first numbers.

Reading 30 closed studies to find your own ESR baseline

The investigator lost, the question left unanswered, and other costs a return calculation misses

What automating funded study management returns when you count ESR hours and wrong payments against the tool's cost

In shortPut in your own numbers. Every default marked illustrative is an assumption to replace. The calculator values manager time and payments not made in error, and shows recovered studies on a separate line, because that value is real and much harder to prove.

Hours, payments, warning and classification: where ESR automation can save

The ESR return calculator: studies, invoices and your hourly cost

Eight questions that show whether your ESR program is ready to automate its post-award work

In shortScore each question from 0 to 2. The total places you in one of three bands, each with a next step. Nothing you choose leaves this page.

From one clean baseline file to a rollout by business unit: what each score tells you to do

What to ask an ESR software vendor about reading, forecasts, supply, money and who decides

In shortAsk every vendor the same questions, us included. Many tools in this category record what staff type. The answers show what the tool reads, how you check it, how it handles the sponsor line, and how you would measure it on your own files.

A twelve-week ESR pilot: one business unit, one closed quarter replayed, one live quarter end in parallel

In shortStart with documents you have already checked by hand, so you know the right answers. Then run one quarter end live, with people deciding everything. Decide with five numbers: field accuracy, citation validity, warning lead, exceptions caught and minutes per study.

1. Build the baseline file for one business unit: dates, budget and product for each study, with a named owner.
2. Gather one closed quarter: every report, reply and invoice, plus what you decided. Plant a few errors in copies: a duplicate invoice, an early milestone, a short-dated lot.
3. Replay it and score every field against your own record, by document type.
4. Run one live quarter end in parallel, with nothing automatic: your team still checks by hand and decides everything.
5. Decide with your numbers, and set the thresholds you are willing to live with for bulk confirmation.

Reading accuracy, warning lead and exceptions: the ESR numbers to score

What people ask about investigator-initiated studies and how they are run

In shortEach answer stands on its own. For the detail behind it, follow the section it came from.