The two speed numbers vendors lead with are a 70% cut in turnaround time and a 90% cut in manual work. Both are "up to" figures, so treat each as a best case until your own records show the same result.
Sharp takeaways
- Treat "up to 70%" and "up to 90%" as ceilings, not typical results.
- Ask each vendor which stage its figure covers: data return, decision, or the full check.
- Pull your own median and slowest days-to-decision from recent hires before any vendor trial.
- Count how many results in your sample need a person to read them, since a fast return does not make those decisions for you.
- Check whether your ATS, HRIS, and payroll platforms are on the vendor's integration list before you assume a workflow gain.
What a 70% turnaround cut actually measures
The most specific speed figure on Checkr's professional services page comes from a single customer story. The page reports that ArchKey unified disconnected legacy systems onto Checkr's automated platform, and that the move cut turnaround times by up to 70%.
That sentence has two parts worth examining before you repeat the number to leadership. The first is the starting point. The gain followed a move away from disconnected systems. The page does not explain which parts of the process drove the gain.
The second is the phrase "up to." It sets a ceiling, not an average. A ceiling tells you what the best described case achieved, which is useful for knowing what is possible but weak for forecasting. Ask the vendor how many checks sat in the reference case, what period turnaround covered, and whether the clock stopped when data came back or when a final decision was logged.
Turnaround also has no fixed definition across hiring teams. Write down the start and stop points your team uses today before you compare the figure with your own numbers.
Judging a return in under one minute
The healthcare page from GoodHire makes a different speed claim. It says: "GoodHire hosts millions of records which allows us to return most employee screens in less than 1 minute." That statement, found on GoodHire's healthcare page, has three parts, and each one limits what it proves.
The first limit is "most." The claim covers most employee screens, not all of them.
The second limit is "employee screens." The minute describes how quickly a screen returns. It says nothing about the time between consent and your final decision, which includes any review your team performs after results arrive. Those are separate clocks, and the healthcare figure starts only one of them.
The third limit is the mechanism the page credits: record volume. A large database helps retrieval, so the speed here describes lookups. That is useful, but it is a narrower promise than end-to-end speed.
Comparing the speed figures side by side
The three speed figures in this post measure different stages, so the table below lines them up by what each one covers.
| Claim | Source page | Figure | What it measures |
|---|---|---|---|
| ArchKey case | Checkr, professional services | Up to 70% | Turnaround time after a platform move |
| Transportation adjudication | Checkr, transportation | Up to 90% | Manual work removed by automated adjudication |
| Healthcare screens | GoodHire, healthcare | Less than 1 minute | Return time for most employee screens |
Read the table as three separate tests, not one scoreboard. Run each as its own test with its own start and stop points, and avoid adding the numbers together or ranking them as if they measured the same thing.
Testing the up-to-90% manual work claim
The claim concerns who handles the file at that step, not how long a database takes to answer.
To test it, define manual work in a way you can count. One usable definition is the number of times a staff member opens, reviews, or logs a decision on a file. Capture one or both for a sample processed before the automation and a sample of similar files processed afterward.
Ask the vendor the questions the headline leaves open. Does manual work include time spent on files flagged for review? Does the reduction apply only to files that clear automatically? If many of your files need a person to look at them, the best-case reduction will not reach those files, and your average gain will be smaller than the headline suggests.
Automated adjudication also runs against criteria. Write down the rules your team applies today to clear, review, and decline results, then confirm that the automated step applies the same rules. A faster decision made under a different standard is a process change, not a speed gain, and a turnaround figure will never reveal the difference.
Which records need a human decision
Not every background check result is a clean yes or no. The criminal background services page from HireRight puts it this way: "While not every record would disqualify an applicant, understanding an applicant's history helps you make more informed decisions." The HireRight page makes the point that not every record would disqualify an applicant, so each record needs a reading before a decision.
That distinction matters for speed testing. A result that contains a record still needs a reading, and reading takes time. If your test counts only clean returns, the average will flatter the process, because the results that take longest are the ones that need judgment.
Split your sample into two groups before you measure: results with no record, and results where a record needs a reading. Time each group separately. You will then know whether a vendor's speed applies to the group that drives your hiring, or only to the simpler majority.
Report the two groups in separate columns with the median time for each. A single blended average hides the split that matters. If the reading group is small, say so, because a small group with long review times can still hold up a requisition that needed to close quickly.
Matching ATS integrations to your workflow
Integration claims are a different kind of promise. The transportation page from Checkr says: "With 200+ integrations across the industry's top ATS, HRIS, and payroll platforms, Checkr makes it easy to initiate, track, and manage background checks without switching workflows." The manufacturing page says: "Minimize manual work with powerful automations, 200+ seamless ATS/HRIS integrations, and our flexible API."
The two Checkr pages describe the integration scope in slightly different terms, as the table shows.
| Claim | Transportation page | Manufacturing page |
|---|---|---|
| Integration count | 200+ integrations | 200+ ATS/HRIS integrations |
| Platforms named | ATS, HRIS, payroll | ATS, HRIS |
| Interface or access | Start, track, and manage checks without switching workflows | Flexible API |
An integration count tells you a connection exists, not that your particular system is connected or that data moves without errors. Check whether your ATS, HRIS, and payroll platforms appear in the categories the page names. The transportation page includes payroll, while the manufacturing page names only ATS and HRIS. If payroll is part of your hiring handoff, that difference affects which workflow you can test.
Each step that stays manual is time the integration cannot save. Then run a sample of requisitions through the integrated path and count how many times a recruiter has to leave the ATS to start, track, or close a check. A low count supports the "without switching workflows" claim. A high count means the integration covers less of your process than the headline implies.
Adverse action across 180+ jurisdictions
Turnaround does not end when a result arrives, especially when a report leads to a negative decision. The healthcare page from GoodHire says: "We manage consent forms and automate the adverse action process at the local level in 180+ states and municipalities." The GoodHire healthcare page makes this a claim about coverage across many local jurisdictions.
Two details matter. List your hiring locations and check each against that coverage before you count on the automation for them.
The second is "automate the adverse action process." The steps after a negative result can take time, and if they are still manual in your process, a fast return can sit in a queue. Measure the time from a negative decision to a closed file, separately for each location, so slow steps show up instead of hiding inside a fast overall average.
Run this as a small test: choose negative results from a few locations, time each stage from decision to close, and note which locations still rely on manual steps.
Judging results against the role
It says: "Results can be evaluated in relation to the role's responsibilities, access to assets or customers, and other relevant factors."
This gives a useful test for judging results against the role, because one turnaround figure hides differences between roles. A warehouse position with no access to money or customer data may be reviewed against different factors than a finance role or a job that visits client sites, so the factors applied to a result may differ by position. If your team reviews both with the same steps, your timing data blends two processes that are not comparable.
Before a vendor trial, write down the factors your team uses for each role family: the duties, the access the role carries, and any other factors your hiring leads treat as relevant. Label each sampled result with its role family, then time evaluation separately for each group.
Keep the criteria fixed during the trial. If reviewers change how they weigh access or duties partway through, a time comparison becomes a comparison of two standards. Record any change and its date so you can exclude or adjust the affected files.
Automation in high-volume manufacturing hiring
The manufacturing page from Checkr makes a broader promise than its transportation page: "Minimize manual work with powerful automations, 200+ seamless ATS/HRIS integrations, and our flexible API." The page attaches no percentage to that promise, so there is no figure to test directly. Your trial has to produce the number.
High-volume hiring makes the question concrete. Count the files in a typical batch, estimate the minutes each takes at each manual step, and multiply.
The phrase "flexible API" needs its own check. Teams with custom systems may connect through the API rather than a prebuilt integration, and that route usually takes engineering time to build and maintain. Count that time as part of the cost. A faster check that takes months to set up may not beat a slightly slower one that runs from the ATS you already use.
Also note what the page leaves out. It describes minimizing manual work, not the decision step or the time to a decision. Ask the vendor which steps the automations cover, then measure those steps in your own batches.
What compliance filters can and cannot prove
The healthcare page from GoodHire makes a claim about trust: "Advanced compliance filters and record verification processes ensure you can trust our results." The GoodHire healthcare page gives no timing figure for this sentence, but it still matters to a speed test. A fast result that turns out to be wrong creates rework, and rework can erase whatever time the vendor saved.
"Ensure" is a strong verb, so it calls for a specific test. The page does not explain what verification involves, and neither tells you whether your team read the result correctly afterward. Ask the vendor what the filters screen for, what verification involves in practice, and whether results show which checks were applied.
To test accuracy, pull a sample of results and compare each against the source record your team can access. Count the mismatches. Also count results that needed correction after your team acted on them, since those corrections cost time a return figure does not capture. A low error count in a small sample is a reason to test more, not proof of reliability.
Keep the compliance test separate from the speed test. If you combine them, a fast vendor with errors can look better than a slower one that is correct. Measure accuracy first, then measure speed on the results that pass.
Screening volunteers without slowing the queue
The nonprofit page from GoodHire puts speed and compliance in one sentence: "Quickly and easily screen volunteers and employees with GoodHire's fast, reliable background checks—while also ensuring your organization stays compliant with the law." The GoodHire nonprofit page names two groups, volunteers and employees, who often move through different queues in the same organization.
That split is the first thing to test. Volunteer screening may start from a sign-up form, while employee screening may start from an offer. If both feed one queue, a slow employee check can hold up a volunteer, or the reverse. Separate the two in your timing data, record the trigger for each, and measure the time from trigger to a cleared result.
The page's two promises need separate tests. "Stays compliant with the law" concerns accuracy and process, so review a sample against your own compliance steps rather than the vendor's description.
If your compliance check requires a person to confirm local requirements, record that step as manual work even if the vendor counts the screen as automated. Volunteer intake often arrives in irregular bursts, so run the trial over several weeks of real requests rather than one batch.
Which one should you pick?
Pick by the bottleneck you actually have, not by the largest number on a page. The table pairs each claim with the scope its page states, so you can match it to the part of your process it covers.
| Vendor and page | Figure on the cited page | Scope of that figure |
|---|---|---|
| Checkr, transportation | Up to 90% less manual work | Automated adjudication |
| Checkr, professional services | Up to 70% shorter turnaround | ArchKey after unifying its systems |
| GoodHire, healthcare | Under 1 minute for most employee screens | Return of employee screens |
| GoodHire, healthcare | 180+ states and municipalities | Consent forms and adverse action |
If manual decisions cause most of your delay, start with the Checkr transportation claim and test the decision step. If the wait sits in the return stage, test GoodHire's one-minute claim across several weeks of requests. If adverse action across several states slows closure, check GoodHire's coverage against your locations. If many of your records need a reading, weigh HireRight's framing, since its page focuses on evaluation rather than speed. Run the same sample through each finalist so the comparison uses one set of requisitions, one clock, and one definition of done.
Related guides: Vendor hiring-tool claims: 5 checks before you buy · AI screening agents: 4 vendor figures to check · Background check turnaround times: 3 hiring ranges.
Questions readers ask
Does the 70% turnaround cut come from a process change?
The professional services page from Checkr says ArchKey unified disconnected legacy systems onto Checkr's automated platform, and afterward it simplified background checks and cut turnaround times by up to 70%. The page ties the gain to that move away from disconnected systems, so the figure describes one customer's before-and-after change. It is a ceiling for that case, not a promised result for every team. Source: Checkr professional services page
How much manual work can automated adjudication remove?
The transportation page from Checkr says automated adjudication can reduce manual work by up to 90%. "Up to" marks a best case, so your share depends on how many of your files clear automatically and how many still need a person. Measure your manual steps before and after a trial to see your own result. Source: Checkr transportation page
Does a 200+ integration count mean my ATS is covered?
The transportation page from Checkr says it offers 200+ integrations across the industry's top ATS, HRIS, and payroll platforms. That count describes the platforms it connects with, but it does not confirm that your specific system is among them. Check your ATS, HRIS, and payroll tools against the list, and test whether you can start, track, and manage checks without leaving your system. Source: Checkr transportation page
Can I connect Checkr to a custom system?
The manufacturing page from Checkr mentions 200+ seamless ATS/HRIS integrations and a flexible API. Teams with systems outside those integrations may be able to connect through the API, though the page does not say how long that takes. Plan for engineering time to build and maintain any custom connection. The page does not name a payroll platform, so confirm that connection separately if you need it. Source: Checkr manufacturing page
Does every criminal record mean I should reject the applicant?
The criminal background services page from HireRight says that while not every record would disqualify an applicant, understanding the applicant's history helps you make more informed decisions. Separate records that need reading in your timing data so your average does not hide the review step. Source: HireRight criminal background services page
What factors should I weigh when reviewing a background result?
The same page from HireRight says results can be evaluated in relation to the role's responsibilities, access to assets or customers, and other relevant factors. The same record may call for different review depending on the job. Write down the duties and access for each role family before you start, then apply those factors consistently. Source: HireRight criminal background services page
How fast do GoodHire's employee screens come back?
The healthcare page from GoodHire says it hosts millions of records, which allows the company to return most employee screens in less than 1 minute. "Most" means some screens will take longer, so track the share that return within a minute rather than relying on a typical time. The minute covers the return of the screen, not the full time from consent to your final decision. Source: GoodHire healthcare page