CRO bid analysis, explained.

Four CRO totals that price four different studies: CRO bid analysis in plain words

Five steps that take NXL-301 from an RFP to a signed award

1. The RFP goes outOn August 26, Northwind Bio sends four CROs an assumptions sheet and a monitoring plan for a 620-patient Phase 3 study.
2. Bids come backBetween September 29 and 30, four proposals of 74 to 128 pages arrive, each with a budget workbook, one of them in euros.
3. Every bid is priced to one studyTwelve assumptions are checked and 17 differences found. The lowest total moves from first to third of four. Sections 05 to 07 cover this step in depth.
4. Questions, scores and a bid defenseEach CRO answers written questions. A panel scores technical approach, like-for-like price, quality and team. Two CROs present on October 14.
5. A person signs the awardThe outsourcing lead signs a recommendation, a senior approver above an agreed amount signs next, and the record goes to the trial master file, the set of a trial's essential records.

The people, CROs and figures in this example are fictional.

Bids that fall, a guideline that drops the word CRO and other things newcomers miss

Reading a CRO budget starts with pass-through cost, unit rate, bid grid and change order

The one-page summary defines 14 terms.

What a CRO proposal prices: service fees, pass-through costs and the assumptions behind both

In shortA bid has two kinds of money and a layer of words. Service fees pay the CRO for its work. Pass-through costs travel through the CRO to others. Assumptions, often in prose or footnotes far from the numbers, say what study those numbers describe. The assumptions decide what the totals mean.

What published studies say trial budgets are spent on

The best public breakdown comes from a study for the US health department, on data from three commercial cost databases from 2004 to 2012. Across phases, excluding site overhead and the sponsor's own monitoring, the largest drivers were clinical procedures, administrative staff and site monitoring.4 These are old figures, measured on a convenience sample (a set chosen because it was available), and they describe whole trials with no split by CRO. They still explain why outsourcing teams read the monitoring plan first. We read the study's abstract, not the full paper.

Full-service, FSP or in-house: how the buying model changes what you compare

ICH E6(R3) section 3.6 and 21 CFR 312.52: what a sponsor owes when it picks a CRO

In shortNo rule tells a sponsor how to compare bids or what weight price deserves. The rules say three other things: the sponsor may transfer work and stays responsible, it must assess suitability before it selects, and the transfer and the selection must be documented. ICH E6(R3) took effect in the EU on July 23, 2025, and FDA published it as final guidance in September 2025.

Assumptions sheet, proposal, workbook and bid grid: what each document in a CRO selection proves

In shortA clear RFP does half the comparison, because each assumption it fixes is one a bidder cannot choose. The proposal says what the CRO intends, and the workbook says what it priced. They do not always match, so read both. Sponsors have long used a common bid grid to force the numbers into one format, and CROs have long found those grids hard to fill in.

How common a shared bid grid was, and how CROs felt about filling one in

The only public survey we found on bid grids dates from 2009. It was run by the University of the Sciences in Philadelphia and TTC, llc, and its articles do not state the sample size, so treat it as a sign of direction.12 Nearly two-thirds of respondents said their company used a common grid across all projects.

The same year, a consultant wrote in Applied Clinical Trials that he had seen grids of over 100 budget lines, and that dividing a grid into major categories is enough, given the latitude CROs have in allocating costs.1 His advice still fits: fix the assumptions in detail, and compare the money in broad groups. A CRO that puts electronic patient diaries inside data management and a CRO that prices them separately can both fill in a grid honestly and still look different on that row.

How to normalize a CRO budget by hand: re-pricing Bluefen's $45.95M line by line

In shortFind each assumption that differs from the RFP. Take the volume the RFP asks for and multiply it by the CRO's own unit rate. Fill anything left out with the RFP's estimate. Leave a risk that may never happen out of the total, and flag it. Every change has two citations: where the CRO said it and where the RFP asks for something else.

Twelve assumptions, four bids, and where each one departs

Start with a table of your assumptions against each bid. In the example, no bid matches all 12. Aldermoor has the fewest differences, two, where Hollowmere has three once travel is counted, and Bluefen differs on eight rows.

Bluefen's nine differences, worked by hand

Each line below takes the volume in the RFP and the rate in Bluefen's own workbook or rate card. Monitor travel follows the visit count, which is why on-site monitoring has one row on the table above and two lines here.

Two rates change on the way: Bluefen's own rate card quotes a higher visit rate for full checking of critical data ($8,744 against $8,301), and a lower site rate at the larger footprint ($33.6K against $37.7K). Patient narratives here are the short case summaries in the clinical study report. The visit counts are as the workbooks state them; the plan itself rounds its count to about 1,970. Use your own CRO's stated visit counts, and ask how they were built. A medical monitor is the medical expert on hand for a study's medical questions.2

Seven more rules the other three bids teach

Monitoring visits and source data verification: why 8 weeks against 12 moves a CRO bid by millions

In shortMonitoring is the largest service line in the example and a large share of trial cost in published data. Frequency and the share of data checked are choices, and current guidance lets them vary with risk. So first decide what plan you want, then price every bid to it. If a CRO proposes a lighter plan, ask for that as a priced alternative beside the bid.

Monitoring means checking that a site follows the protocol, the study's written plan, and that the data are reliable. Source data verification, or SDV, is one piece of it: comparing what was entered with the original record. A monitoring visit costs staff time and travel, so the visit count moves a CRO budget fast. In the example, the monitoring line and its travel carry $6.97M of Bluefen's $11.95M rise.

What FDA and ICH say about visit frequency and full verification

In 2013, FDA wrote that for major efficacy trials, companies "typically conduct on-site monitoring visits at approximately 4- to 8-week intervals", at least partly because of a perception that frequent visits with 100% verification were FDA's preference. It added that there was "a growing consensus" that risk-based approaches are "more likely than routine visits to all clinical sites and 100% data verification to ensure subject protection and overall study quality".3

ICH E6(R3) says the sponsor should decide the extent and nature of monitoring from identified risks, and that frequency should also follow those risks.2 Neither text sets a number of weeks or a percentage. FDA's April 2023 questions-and-answers guidance expands on the 2013 text.13

Measured data point the same way. In an after-the-fact (post hoc) study of three Phase 3 trials, covering 2,566 subjects and more than 3 million data fields, complete SDV left an error rate of 0.27% and partial SDV 0.53%: about 370 data points checked to avoid one error. The authors concluded that the results did not support complete SDV as a way to make data meaningfully more accurate.14

Most of the authors work at one company that helped collect the data, and several hold shares in it. We read the abstract and the authors' disclosures, not the whole paper. One study does not settle your trial, and its error rate covers any field, whatever its importance.

What to do with a bid that prices a lighter plan

Do the same for a bid above your plan. In the example, Hollowmere offers a visit every 6 weeks to protect enrollment quality in a disease area where trials compete for patients, at $1.22M more. That may be worth buying, as a decision. Removing it to compare is a step in the arithmetic, and the panel still gets to ask what the extra visits buy.

Where CRO bid grids go wrong: footnotes, exclusions, mismatched rows and a stale RFP

In shortThe expensive errors come from an assumption nobody read, a cost left out, a row that means different things to two CROs, or an answer that never reached the spreadsheet. We found no published study of how often these happen, so count your own on one past selection.

Where to look first when you have one afternoon

Some differences are large and easy to find, such as a stated site count. Others are small and well hidden, such as a footnoted fee. A review with limited time should begin where the money is large and the words are hard to find.

Do not stop at the money. An RFP with gaps produces gaps in every bid. One consultant's list of RFP mistakes includes leaving out vital information: missing details such as enrollment dates and monitoring frequency force bidders to make their own assumptions.1 If three of four bids differ from your RFP on the same row, the row may be your problem. In the example, the outsourcing lead runs the next RFP draft through the same assumption list before it goes out.

CRO change orders and scope drift: how a low price comes back once the work starts

In shortA bid priced on a lighter study does not stay light. The missing work returns as change orders, some of them unavoidable and some bought at the award. Published data are thin: a 2009 survey on budget variance, a 2023 study by Tufts and a large CRO on on-budget delivery, and Tufts data on amendments, formal changes to a study's protocol. Nobody we could find publishes change orders as a share of CRO budgets.

At a 2016 conference, a sponsor's executive director of development outsourcing warned: "Most times you end up with a contract that has an incomplete scope." She advised setting aside a percentage of the trial budget for change orders, and named no percentage.

The article's author concludes that change orders are "for the most part, unavoidable".17 He also says a large part of CRO revenue comes from them, with a figure of up to 100% additional revenue and no dataset behind it. That is one writer's claim about revenue, so do not read it as a share of budgets. We found no public figure for that share.

Telling a protocol amendment from a mispriced bid

A change order can come from two places, and a better comparison reaches only one. An amendment changes the study after the award: that is honest new scope, and the rates in the contract should set its price. A bid priced on 105 sites when the RFP asked for 140 was always going to cost more. Keeping the two apart is the reason to file the like-for-like grid with the contract: when a change order arrives, you can ask whether the work was in the RFP you sent.

What a price cannot tell you about quality

Some failures have nothing to do with the budget. FDA inspects sponsors and CROs as one group. In fiscal 2024, it classified 99 such inspections: 83 needing no action, 15 voluntary action and one official action. The first of the most common observations it lists is failure to ensure proper monitoring.19

Separately, FDA keeps a page of notifications on data integrity, four between 2016 and 2025, covering five contract research organizations whose studies it found unacceptable, and it reminds sponsors and applicants that they are responsible for the integrity of their study data.20 In the 2025 case, a law firm's summary says sponsors relying on in vitro (lab-dish) data at issue had to repeat the study at a different site.21

Those cases mostly concern bioequivalence work, tests of whether two versions of a drug act alike, a world away from a Phase 3 proposal. For a Phase 3 selection the step is the same: ask for qualification evidence, SOPs and metrics, as E6(R3) 3.6.8 says.2

How much a Phase 3 CRO contract costs, what a day of delay is worth and two laws to add to the RFP

In shortThere is no public price list for CRO work. Published figures describe whole trials, from older data or a modeling tool. They can sanity-check a budget and cannot set a price. A day of delay can outweigh a large price gap, and two recent US laws may change what a bid costs you.

Published Phase 3 cost figures, and what each one measures

The example's bids near $55M for 620 patients at 140 sites in 16 countries are fictional. They sit well above that older dermatology average, which measured a different scope, at US sites, a decade and more ago. So compare a real bid with your own past awards, scaled for sites, months and patients, and treat outside figures as background. We found no public benchmark for CRO unit rates. Commercial databases exist, and a sponsor's own awards are often the better set because the scope is known.

Reading a price gap as days of delay

Tufts CSDD estimates the direct cost of a day at about $40,000 for Phase 2 and 3 trials, and $55,716 for Phase 3 alone. It puts lost sales at about $500,000 a day in its journal article and about $800,000 in its white paper, from one dataset of 645 drugs.23,24 We read the white paper and the article's abstract.

Dermatology is among the areas with the highest relative daily direct costs.24 If you turn a price gap into days, you can ask a sharper question: would this CRO need to be that much faster for the difference to stop mattering?

The CRO market: size estimates that disagree

The two commercial reports we could open put this market at $83.6 billion for 2025 and $103.4 billion for 2023, and we read only their public summaries.25,26 We could not see their methods or boundaries, so take them as an order of magnitude.

BIOSECURE and section 174: two laws a bid total does not show

A weighted CRO scorecard and a bid defense meeting: how price meets everything else

In shortComparing like for like narrows the price gaps, and that can make technical approach and quality decide. Agree weights before the bids arrive, score price with arithmetic, score the rest with a panel, and use a bid defense to meet the people who would run the study. A person awards the work, on a record that shows why.

A scorecard turns several judgments into one number that a panel can argue with. Teams usually score technical approach, price, quality and the proposed team. We found no public standard for the weights. The example uses technical 40%, price 30%, quality 20% and team 10%, and gives the lowest like-for-like total 5 points and the others a score in proportion: $54.92M ÷ $57.90M × 5 = 4.74 for Bluefen.

On submitted totals, the score hands the top mark to the bid that priced the smallest study. Priced alike, four bids that seemed $14M apart are about $5M apart, so the 70% of the score that is not price carries more of the decision.

What a bid defense is for

A bid defense is a meeting at which a shortlisted CRO presents and defends its proposal to the sponsor's panel. It previews how the partnership might work. A trade article advises the sponsor to review the proposal in depth and come with prepared questions, and the CRO to bring its team lead in person with the people who will spend most of their time on the project.28

Where AI fits in a CRO proposal comparison, and what FDA's AI draft says about it

In shortReading long proposals, matching line names and doing the arithmetic suit software. Deciding what study you want and whom to hire does not. FDA's draft AI guidance says it does not address internal operational uses, as long as they do not affect patient safety, drug quality or study reliability. Quality teams still need evidence, and a published test on financial documents shows why every number needs a citation a person can open.

Which regulatory texts reach a bid comparison tool, as of November 2026

Why every extracted number needs a page beside it

We found no public test of AI on CRO proposals. The nearest evidence we found is a 2023 benchmark on public companies' financial documents. On a sample of 150 cases, GPT-4-Turbo used with a retrieval system answered wrongly or refused on 81% of questions, and the authors found all the models they examined had weaknesses that limit enterprise use.36

Those models are old, and that test differs from reading a CRO workbook. Treat the result as a reason to measure: run any tool on a finished selection, compare it with a checked answer, and require a citation for each value.

Limits to plan for: scanned tables, merged cells and hidden tabs; a term used differently by two CROs; a footnote that changes a line's meaning; and the temptation to read a benchmark as a price. A good process makes each value checkable in one click, gives a person the unclear mappings, and leaves the decision and the signature with people.

The CRO selection process today: rebuilt spreadsheets, a scoring panel and a signed memo

In shortAn outsourcing lead rebuilds each bid in a spreadsheet while a study lead reads the technical sections. Finance checks totals. Quality runs a qualification audit. A panel scores, a bid defense follows and a senior person signs. No public study measures where the time goes; our rough estimate puts most of it in reading and rebuilding. The reasoning often stays in email.

Who shapes a CRO award: outsourcing, the study team, finance, quality and the bidders

Eight steps from RFP to contract, with their hand-offs

Building the grid yourself, paying a consultant or buying bid comparison software

Bluefen falls from first to third and Aldermoor leads at $54.92M: the NXL-301 comparison on screen

In shortPriya Ranvelle, the clinical outsourcing lead, opens four bids in columns beside what the protocol asks. Bluefen looks $7.38M cheapest. She runs the assumption checker, which reads the plan, then each bid in turn. Every bid is re-priced at its own rates and re-ranked in front of her, and each difference opens on its page.

1

Four bids as each CRO priced them

NXL-301 is a Phase 3 study of nexrakimab in atopic dermatitis: 620 patients, 140 sites, 16 countries, 34 months. The first column is what the protocol asks. Then come Hollowmere at $57.81M, Aldermoor at $53.33M, Bluefen at $45.95M and Quillon at $60.29M, priced in euros. Bluefen is 17% under the $55.33M benchmark from 23 past awards. The banner tells Priya to check assumptions before anyone is shortlisted.

2

The checker reads the plan, then each bid

Priya presses one button. The assumption checker reads the RFP assumptions sheet and the monitoring plan, which give 12 assumptions to test. It then goes bid by bid. As each finishes, its column lights up: a tick for a match, a colored cell for a difference, with its effect on the total.

3

Every bid priced to one study, and re-ranked

Aldermoor leads like for like at $54.92M. Hollowmere comes down to $57.19M, in second. Bluefen rises $11.95M to $57.90M, third of four. Quillon falls to $59.84M, fourth. The side panel lists why Bluefen moved, difference by difference, and the banner says it in one sentence.

4

A difference opens on its page

Priya clicks Bluefen's monitoring cell. The panel shows what the bid assumed, what the plan asks and the effect: +$6.97M with travel. Below sits page 37 of Bluefen's proposal with the sentence marked, and under that the monitoring plan it is measured against. Bluefen's wording says the frequency may increase "subject to a change order". She has a question to send, with its source.

5

A 6% fee in footnote 3

Aldermoor's page 44 carries three notes under a pass-through table. The third adds "a handling fee of 6%" on pass-through costs, which never reached the budget total. The RFP asks for pass-throughs at cost. The proposal reader takes notes as seriously as tables, so 6% of $17.53M, or $1.05M, joins Aldermoor's like-for-like total, and the question goes out.

The 865-line bid grid, 17 assumption gaps and the 12 rows every bid is priced to

In shortBehind the comparison sit the protocol basis, a line-by-line grid on 15 standard tasks with past awards beside each, and a list of gaps with effect and severity. Any number opens the proposal page or workbook row behind it, with the arithmetic.

Eleven drafted questions, one weighted scorecard and an award a person signs

In shortThe clarification writer drafts a question per gap, citing that CRO's page and the RFP line. Priya edits and sends them, CRO by CRO. An answer updates the total. The panel scores on like-for-like price, the bid defense follows, and the award memo writer drafts a recommendation that Priya signs and Omar Haddad approves.

1

Questions drafted for a person to send

Eleven questions wait across four CROs, each with its citations. Bluefen's five ask it to re-price monitoring to the plan, confirm fees for months 31 to 34, and price 140 sites or state the enrollment risk of fewer. Nothing goes out until Priya presses send, and no question mentions another CRO's price.

2

An answer changes the total

Aldermoor replied on October 7: it waives the fee for NXL-301 and withdraws footnote 3. Priya applies the answer. Aldermoor's like-for-like total falls from $54.92M to $53.87M, and the open gap count drops from 13 to 12. Its translation question is still out.

3

A scorecard on like-for-like price

Weights come from settings: technical approach 40%, like-for-like price 30%, quality and oversight 20%, project team 10%. Panel scores were entered from October 3 to 6. Before the fee answer, Aldermoor scores 90.4 of 100, Hollowmere 87.0, Quillon 85.5 and Bluefen 77.0. The screen notes that on submitted totals Bluefen would take the full price score, where like for like it scores 4.74.

4

A brief for the bid defense

The study lead, Dr. Maya Chen, chairs the panel. A one-page brief lists what to settle with each CRO: Aldermoor's translations at the $540K estimate and its enrollment plan, Hollowmere's case for visits every 6 weeks, and subcontracting with both. Price stays on like-for-like totals.

5

A six-section memo, every statement cited

The award memo writer drafts six sections: recommendation, how the bids were compared, like-for-like price, assumption gaps, quality and oversight, and conditions. Eight citations are checked. Dana Okafor's qualification audit goes into the quality section.

6

She signs; above $10M, he approves

Priya signs with the meaning "I recommend this award". The record holds her name, the date, the time and that meaning. Because the award is above $10M, it goes to Omar Haddad, head of clinical sourcing, under the delegation of authority. He approves subject to the October 14 bid defense. The grid and memo are filed to the trial master file.

A new bid landing mid-selection, an RFP checked before it goes out and your like-for-like rules

In shortThe outsourcing lead's day holds more than one study. Seven sourcing events, each a selection under way, sit on one list. A late bid is read and added while she watches. A draft RFP gets checked before bidders see it. The dashboard shows what the checker finds across selections, and your rules are settings.

The outsourcing lead sends every question and signs; six software agents read, map, check and draft

In shortEach agent has one job and limits it does not cross. None of them decides. A person sends every question and signs the recommendation, and above an amount you set a second person approves. Botminds reads your RFP, proposals, workbooks and past awards, and leaves a grid, questions, a scorecard and a signed memo in your own record.

What each bid analysis agent reads, checks or drafts

The lead sends and signs, the panel scores, the head of sourcing approves

The person who proposes differs from the person who approves. Every agent run, rule change, question sent, signature and approval goes on the audit trail with who and when, and proposals are visible only to the team on that selection.

These are controls that support your compliance with 21 CFR Part 11 and EU Annex 11; no product can be certified against them, and your quality team judges whether a record meets your obligations. Botminds' own credentials are SOC 2 Type II audited and information security managed to ISO 27001. Roles, separation of duties and integrations with your contract system and trial master file come as standard.

Who signs and scores in the NXL-301 screens

From RFP and past awards to a like-for-like grid and a signed memo

Works with what you already run: a contract system for the work order, a trial master file for the record, email for proposals and answers. Design targets, measured in the live solution: about 6 minutes to a like-for-like grid of four bids, and a median of 24 days from proposals in to award.

The bill for a mispriced CRO bid: change orders, late awards and fees found in invoices

In shortStart with your own past awards. Published figures can frame the question: how much scope moves budgets, what a Phase 3 day costs, how many trials run over. How much of your own change order spend was in the RFP you sent is a number only your files can give you.

Six places past CRO awards hide change orders, delays and missed fees

Goodwill, rescues and lost reasons: CRO selection costs that stay off the spreadsheet

Your return on like-for-like comparison: price on the award, outsourcing hours and days on the critical path

In shortThe calculator counts the price you could avoid paying on awarded work and the hours a comparison takes, against what the tool costs. It shows trial days separately, because they count only if the award sits on your study's critical path, the chain of tasks that sets its end date. Replace each default tagged illustrative with your own figure.

Six returns from a better CRO comparison, and the three this calculator leaves out

Work out your own yearly value from selections, award size and comparison hours

A sponsor placing one or two CRO awards a year probably cannot justify software on hours alone. For that sponsor, a clear assumptions sheet and a careful spreadsheet checked by a second person may be the right answer.

Numbers on our own solution: about 6 minutes to a like-for-like grid of four bids and a 24-day median from proposals in to award are design targets, measured in the live solution. A 3 to 8% better price is our own estimate. All three are unpublished.

Your outsourcing team's readiness in eight answers: selection volume, past awards, written rules, sign-off

In shortAnswer as your team works now. Eight questions on selection volume, RFPs, past awards, rules and sign-off score 0 to 2, and a ninth is there to think over. Your running total sits in the meter, beside why each question matters and what a strong answer sounds like.

Your score decides the next move: fix the assumptions sheet, replay a selection or plan the rollout

The total can flatter. A 0 on owner and budget, or on quality and AI approval, holds you at a pilot whatever else you scored, because either one can stop a rollout alone.

Testing a bid analysis vendor on sources, re-pricing, rival prices and who signs

In shortPut the same questions to every vendor, us included. The answers show where each number comes from, whose rates re-price a bid, how a benchmark is kept apart from a price, how proposals are kept private, who decides, and how you could test all of it on a finished selection.

We found no published independent test of any bid analysis vendor, as of November 2026. A replay on your own finished selection is the fairest comparison, and the same test measures your current process.

Starting with a past CRO award, then shadowing a live selection: a twelve-week test decided on three numbers

In shortMark your change orders, write down your rules, and give a tool a past selection with the bids as they arrived. Then run alongside a live one. Decide with three numbers: differences found against a checked list, hours to a like-for-like grid, and how many past change orders the comparison would have raised. None of these steps commits you to a vendor.

1. Mark your last five awards' change orders: was this work in the RFP? You will learn something whichever way you go.
2. Write your like-for-like rules on one page: own rates, RFP estimates for gaps, how risks are shown, one exchange rate.
3. Choose two finished selections with the RFP, the proposals and workbooks as they arrived, and build a checked list of differences by hand for one.
4. Run alongside a live selection for one full cycle, with your lead sending the questions and signing as usual.
5. Decide with three numbers: differences found against missed, hours to a grid, and past change orders it would have raised.

Scoring the bid analysis pilot against what your lead found: misses, hours, citations

Bid grids, pass-through costs, change orders, bid defenses: common CRO bid analysis questions

In shortThirteen questions clinical outsourcing teams and buyers search for, each answered on its own.