Every few months, the same question appears in a recruiting Slack channel: what does hiring look like at other SaaS companies? How many interviews do they run? How quickly do they move? How often do their interviewers submit scorecards?
The question is understandable. In Metaview's 2026 AI & Hiring Alignment Report, 67% of the 505 recruiting leaders and hiring managers surveyed said their teams lose qualified candidates every month to competitors that move faster.¹ When speed feels like the problem, an external benchmark sounds like the obvious place to start.
Metaview's interview corpus cannot provide a trustworthy SaaS benchmark. The canonical dataset does not contain a validated industry field or a reliable company-size field that can be used for that comparison. Calling any cut of it “SaaS” would require adding outside company data and publishing how that classification was made.
That limitation does not make the corpus useless. It changes the question it can answer. The data can describe documentation patterns inside the interviews Metaview captured. It can also give individual teams a reference for auditing their own process across roles, stages, interviewers, and time periods.
One boundary applies to every finding below. The corpus contains no post-hire performance, retention, tenure, attrition, or quality-of-hire outcomes. It cannot tell us whether a documented decision produced a better hire. It can only show what was recorded during the hiring process and what was missing from the measured record.
The benchmark you're looking for isn't in the data
A valid industry benchmark needs a reliable way to assign each company to an industry. Metaview's canonical interview dataset does not include that field. It also does not include a dependable company-size band that could separate early-stage startups from large employers.
You could create those fields by joining the interview records to outside company data. That may be a reasonable research project, but it would introduce new decisions: which taxonomy defines SaaS, how mixed business models are classified, how subsidiaries are handled, when headcount is measured, and what happens when the external data is missing or wrong.
None of those choices exists in the current corpus. The honest conclusion is limited to this dataset: it cannot tell us how many rounds SaaS companies usually run, how quickly SaaS companies hire, or how their scorecard behavior compares with another industry.
There is another reason not to treat the corpus as a market benchmark. Every company represented chose to capture interviews with Metaview. That makes the dataset a view of recorded hiring among Metaview users. It is not a representative sample of every employer. Even with a perfect industry label, the results would still need to be described as findings from participating companies rather than a universal norm.
The figures below are therefore reference points. They are not targets. They describe what appears in the measured corpus. They do not tell your team what a good rate should be.
What the record shows when a decision gets made
The corpus does allow a narrower analysis: how often a captured interview has a scorecard attached, how often a scorecard is submitted, and what those submitted or created scorecards contain.
Across 5,207,185 captured candidate interviews, 1,624,988 had at least one scorecard attached. That is 31.21%.² The unit matters. This is a count of interviews with an attached scorecard, not the share of candidates with complete feedback, the share of scorecards submitted, or the share of hiring decisions supported by written evidence.
A separate analysis looked at candidates who advanced. Among 296,555 advancing candidates, 41.9% had no submitted scorecard in the measured record when the advancement happened.² This is not the same metric as scorecard attachment. It uses candidates rather than interviews, and it checks submission at a decision point.
The finding does not prove that nobody wrote anything. A transcript, Metaview notes, ATS comments, recruiter notes, or debrief notes may have existed. What it shows is that the expected structured evaluation was not submitted in the measured record before the candidate moved forward.
Where scorecards were created, completeness also varied. In a quality-control sample of 13,373 created scorecards, 25.6% were completely empty and 32.1% had every field completed.² The remaining scorecards sat somewhere between those two endpoints. An empty scorecard is not evidence that the interview was poor. It is evidence that the form existed without a completed written evaluation.
The comparison between manual and Metaview-drafted scorecards is also about documentation. In a capped sample of 120,000 scorecards, 50.3% of scorecards where Metaview generated the first draft were submitted, compared with 28.6% of scorecards written from scratch.² The groups contained 93,502 and 26,498 scorecards respectively.
A separate capped sample of 80,000 found an average of 7.85 completed fields in Metaview-drafted scorecards and 2.61 in manually drafted scorecards.² These were different samples, so the submission and field-count figures should not be merged into one result.
Both analyses are observational. Teams and interviewers were not randomly assigned to draft types. A prefilled draft also starts with more content than a blank form, which is part of what the feature is designed to do. The data does not show that the Metaview-drafted scorecards were more accurate, fairer, or better reasoned. It shows that they were submitted more often and contained more completed fields in these capped samples.
Metaview's current product guidance is consistent with that boundary. Notes summarize and organize factual information from the conversation. With supported ATS workflows, Metaview can fill objective or factual scorecard fields. It leaves subjective evaluations, ratings, recommendations, and final submission to the interviewer.³
It helps us not only think more critically about conversations, but it also makes it easier to partner with hiring managers. You can readily output a series of notes from your calls that day and say 'this is what we talked about, this is why I think they're a great candidate, and here's the evidence.'”
Most scorecards that do get submitted arrive quickly
Among the 811,298 scorecards that were submitted, the median submission time was 2.32 hours after the interview ended. Almost three quarters, 74.7%, arrived within 24 hours. The mean was much higher at 36.16 hours because a smaller group took far longer.²
This tells us when submitted scorecards tend to arrive. It does not tell us why other interviews never received a submitted scorecard. The missing group could include forgotten forms, unclear ownership, ATS matching problems, scorecards that were never created, forms that were opened and abandoned, or interview stages where the team did not expect a scorecard.
That distinction changes how a team should investigate the problem. Reminders may help when interviewers intend to submit but forget. They will not fix a stage with no assigned owner, a form that does not match the interview, or an integration that failed to connect the conversation to the correct record.
Start by separating the failure modes. Check whether the scorecard was created, whether it matched the right interview, whether an interviewer was assigned, whether it was opened, and whether it was submitted. A single completion percentage hides all of those steps.
The same gap shows up at kickoff
The scorecard corpus does not measure kickoff alignment, and the Alignment Report does not measure scorecard completion. They are separate sources. The practical connection is simpler: both sourcing and interviewing become harder to audit when the team's criteria or reasoning remain implicit.
In the survey, 79% of respondents in the combined group with an excellent recruiter and hiring-manager relationship and high alignment on requirements said their teams exceeded business goals. The comparable figure was 36% among respondents in the combined group with a fair or poor relationship and low alignment.¹
That 43-point gap should not be treated as proof that writing a better kickoff document causes a team to exceed its goals. Each segment combines two conditions, relationship quality and alignment. The outcome was self-reported, and all measures came from the same cross-sectional survey. Stronger teams may rate both their relationships and their results more positively, or differ in many other ways.
The survey supports a modest conclusion: respondents who described stronger working relationships and better alignment also reported better business results. It does not identify which process change produced that association.
A team can still make one concrete improvement at kickoff. Put the role criteria where recruiters and interviewers can use them later. Record the capabilities that matter, what evidence would support each one, which requirements are firm, which are preferences, who owns each assessment, and how disagreements will be resolved.
Metaview's Application Review can help turn role documents into a draft Ideal Candidate Profile. The team reviews and edits that profile before using it. Candidates are then assessed against the agreed criteria with written reasoning and a match category. A person decides who progresses or is rejected, and Metaview never auto-rejects candidates.⁵
The comparison you can run
A SaaS-wide target is not available from this corpus. Your own process still gives you useful comparisons, provided you keep the definitions consistent.
Start with four questions:
- Of the candidates who advanced last quarter, how many had a submitted scorecard at the time of advancement?
- Which roles and interview stages have the lowest scorecard attachment and submission rates?
- Where a scorecard was created, how often was it empty, partly completed, or fully completed?
- How quickly were submitted scorecards completed, and where does the long tail come from?
Run the same definitions over time. A movement from one quarter to the next is easier to interpret than a comparison with a corpus drawn from different companies, roles, workflows, and integration setups.
Metaview Reports can help teams search and filter captured conversations using calendar and event information, topics or keywords, ATS properties, and custom properties. Reports can also add structured AI columns, create charts, and export data to CSV.⁴ The exact questions you can answer depend on the fields available in your workspace and ATS integration.
Use those tools to find the gap, then inspect a sample of the underlying records. A low submission rate may come from one stage, one form, one integration, or one expectation that was never made explicit. The aggregate number tells you where to look. The records tell you what to fix.
Metaview's corpus stops before post-hire outcomes. A company that separately defines and governs post-hire measures could study how interview assessments relate to later results, but that would be a different dataset and a different analysis.
Write it down where a second person can read it
The absence of a SaaS benchmark is not a reason to stop measuring. It is a reason to use the right comparison.
Look at whether your own criteria are visible before the search begins, whether interviewers know what they are assessing, whether structured feedback is submitted before a candidate advances, and whether another person can understand the basis for that decision later.
A submitted scorecard does not prove the judgment was correct. An attached scorecard does not prove it was complete. More completed fields do not automatically mean better evidence. Those limits should stay visible whenever the figures are used.
The useful standard is more practical: when a candidate moves forward, can the next person see the relevant observations and the interviewer's judgment without reconstructing the decision from memory?
If the answer is no, the team does not need an industry benchmark to know what to improve.
Read your own hiring record first
See how your process documents decisions across roles, stages, interviewers, and time periods.
Frequently asked questions
Is there a SaaS hiring benchmark in Metaview's interview data?
No. Metaview's canonical corpus does not contain a validated industry field or dependable company-size field, so it cannot support a SaaS-specific benchmark. Outside company metadata could be joined to the records in a separate study, but the classification method and selection limitations would need to be published.
Does a completed scorecard mean a better hiring decision?
Not on the evidence available here. Metaview's corpus contains no post-hire performance, retention, tenure, attrition, or quality-of-hire outcomes. A submitted scorecard shows that a structured evaluation was recorded. It does not establish that the evaluation was accurate or that the candidate became a successful hire.
What do the Metaview-drafted scorecard figures measure?
They measure submission and field completion in two capped observational samples. Scorecards where Metaview generated the first draft were submitted more often in the 120,000-scorecard sample and had more completed fields in the separate 80,000-scorecard sample. The analyses do not measure accuracy, fairness, decision quality, or post-hire outcomes.
Does Metaview integrate with our ATS?
Metaview connects with a range of ATS platforms, but available data, permissions, scorecard workflows, and write-back actions vary by system and configuration. Check the current integrations documentation for the capabilities supported in your setup.
Sources
¹ Metaview 2026 AI & Hiring Alignment Report
² Metaview canonical aggregate data, July 22, 2026
³ Metaview Help Center: Best practices for Notes
⁴ Metaview Help Center: Reports Overview
⁵ Metaview Help Center: Application Review Overview
