44% Can Show a Performance Management Metric.

The confidence and the process are not the same measurement. HR Dive’s October 13, 2026 piece reports a 2026 Betterworks survey: close to 85% of HR leaders said they were very or extremely confident that performance management can help with organizational priorities, and only 44% said they could provide business metrics to back that up. The poll, conducted in August, covered 934 HR and business leaders. I could not retrieve the full article body. I am using the figures the search extract printed, and I am not adding sentences the extract did not contain. EvalFlow’s performance-review cost page, which the search index dates to October 9 and which the page itself says was last reviewed October 8, cites Talent Strategy Group’s 2026 Performance Management Report. That report surveyed more than 250 organizations. 91.6% of them had a formal performance-management process. 91.6% and 44% are not a before-and-after. They are two surveys, two sample units, and two questions.

We already had a remote-work edge that a later measurement did not keep. The remote edge vanished, and the boss did not. We already had a four-day week that was a review, not a Friday off. That review is a different calendar from a performance cycle. We already had an AI time-save that a second clock took back. Eleven hours saved, 6.4 hours spent botsitting. This week’s pair is about whether a performance process can show its work. I am not going to tell you to buy software. The usable part is which percentage belongs to which page.

Performance management confidence is 85%. The metric is 44%.

The Betterworks figures, as HR Dive’s extract states them, are a gap inside one poll. Close to 85% were very or extremely confident that performance management can help with organizational priorities. 44% could provide business metrics to back up their position. “Close to 85%” is not 85.0. I am not going to drop the “close to” and print a tenth. 44% is printed as 44%. The sample is 934 HR and business leaders, polled in August. The page does not, in the extract, say 934 organizations. Leaders and organizations are different units. EvalFlow’s Talent Strategy Group citation is organizations. If a slide says “934 companies,” it has changed the unit HR Dive printed.

The extract also says the findings suggest HR should buckle down on data, and that workplace experts urged HR not to lose the human element while chasing metrics. I am not going to turn “buckle down on data” into a dashboard spec. The sentence I can use is the gap: confidence is common in this poll, and a business metric to support that confidence is not, at 44%. A team that reports the 85% and skips the 44% has kept the mood and dropped the constraint. A team that reports the 44% as “most HR leaders cannot measure anything” has edited “business metrics to back up their position” into a broader failure the extract did not claim.

Jill Barth’s October 14 piece at HR Executive does not print either percentage. The page I could retrieve is a stub: ServiceNow’s Jacqui Canney urges HR to redesign work for AI, Hilton’s Laura Fuentes focuses AI resources on priority business problems, and both call on leaders to make learning part of work. Those are three bullets. They are not a sample size. I am not going to attach 44% to Canney or Fuentes. Their names are on a different article, published the next day, and the body beyond the bullets did not come back. A quote that puts the Betterworks gap in their mouths has merged two October pages.

91.6% have a process. That is not 44%, and it is not 85%.

Talent Strategy Group’s figures, as EvalFlow cites them, start with a different question. More than 250 organizations. 91.6% had a formal performance-management process. Having a process is not the same as being confident it helps priorities, and it is not the same as being able to show a business metric. You can have the form and still be in the 56% that, in the other poll, could not produce the metric. You can also be outside both samples. I am not going to subtract 44 from 91.6 and call the difference a finding. The samples were not the same people.

The rest of the Talent Strategy Group list is about how the process is built, and it is more useful than the headline percentage if you are trying to see where time goes. 75.7% used one process across the organization rather than multiple processes. Organizations using a single process reported 17% stronger individual-performance outcomes in the study’s comparative analysis, plus 14% stronger alignment, 8% stronger feedback and coaching, and 14% stronger assessment outcomes. EvalFlow’s citation says these are survey associations, not causal estimates. I am leaving that label on. A manager who treats 17% as the return on collapsing two forms into one has deleted the sentence that says the study did not estimate a cause.

91.1% included objective setting. 84.8% set objectives annually. 6.6% did so twice per year, and 6.6% more than twice. 84.8 plus 6.6 plus 6.6 is 98.0, not 100. I am not going to invent the missing 2 points. The page printed three slices. It did not print a residual. 33.6% targeted one month to complete objective setting, and 32.9% targeted two months. In practice, 23.6% took three months, 10.4% took four months, and 8.3% took five months or more. The target column and the practice column are not the same question. A one-month target that became a three-month cycle is the kind of gap the page is pointing at, and it is not a gap you can see if you only record the target.

80.7% required employees to use technology for performance management. 16.7% made technology optional. 2.7% reported no technology use. 80.7 plus 16.7 plus 2.7 is 100.1. I am not going to “fix” that to 100. Rounding on the source page is the source page’s. The operational sentence is that a required tool is the common case in this survey, and a no-tool case is 2.7%, not a lifestyle brand.

The conversation count and the review count are different clocks

Feedback is where the intended calendar and the actual calendar split inside the same citation. 47% of organizations expected three or four conversations per year. 68.5% reported that one or two feedback conversations actually occurred. Expecting three or four and getting one or two is the productivity fact, if you are the person who has to hold the meetings. A process document that says “quarterly check-ins” has written the 47% intention. It has not written the 68.5% result. I would put both numbers on the card before I added a fifth meeting to a year that is already missing the third.

93.6% included a formal performance review. Among those organizations, 56.3% reviewed annually and 36.1% reviewed twice per year. 56.3 plus 36.1 is 92.4, not 100 of the organizations that review. Again, I am not filling the residual. 92.4% assigned performance ratings. That 92.4 is a different 92.4 from the sum I just did. Same digits, different sentence. I am not going to treat them as one statistic because the number matches.

84.7% used performance calibration. Mandatory manager training was reported by only 21.2% for objective setting, 17.9% for feedback, and 19.2% for performance reviews. Calibration without a required training hour is the common shape in this citation. A rollout that announces calibration and assumes managers were trained has assumed the minority case. 21.2% is the objective-setting training figure. It is not the Betterworks 44%. One is a share of organizations that require training. The other is a share of leaders who can produce a business metric. I am not going to put them in one sentence without the nouns.

The dollar example is a model. The Deloitte range is a citation.

EvalFlow prints a worked example and labels it as one. 800 eligible employees, 120 managers, one annual cycle. Illustrative loaded rates of $45 for employees, $70 for managers, and $60 for HR. Employee time 1.5 hours, so 800 times 1.5 times $45 is $54,000. Manager time 3 hours per employee, so 800 times 3 times $70 is $168,000. HR administration 180 hours times $60 is $10,800. Calibration 45 participants times 3 hours times $70 is $9,450. The page says that at 800 employees the example equals about $303 per eligible employee, and that the figure does not include software, executive approvals, external advice, delayed decisions, or the value of better conversations.

I added the four labor lines. 54,000 plus 168,000 is 222,000. Plus 10,800 is 232,800. Plus 9,450 is 242,250. 242,250 divided by 800 is 302.8125, which rounds to the “$303” the page already prints. The arithmetic checks. It does not become a survey. Talent Strategy Group did not say the median company spends $303 per employee. EvalFlow said an example with those inputs does. If your company has 80 employees, or your loaded manager rate is not $70, the $303 is the wrong output even before you argue with the method. The page’s own advice is to baseline one completed cycle with calendar data and system activity, not memory, and to value time conservatively with an adoption assumption below 100%. That is a method. It is not a benchmark I am willing to paste onto the Betterworks 44%.

The other dollar figures on the page are older citations, and EvalFlow labels the chain. Deloitte cites Gallup projections that the lost-hours cost of inefficient performance evaluations can range from $2.4 million to $35 million per year for a company with 10,000 employees. That is a range for a 10,000-person company, attributed through Deloitte to Gallup. It is not the 800-person example, and it is not the 250-organization survey. Deloitte’s own redesign, EvalFlow says, was prompted in part by a finding that the organization was spending approximately two million hours a year on performance management, a figure EvalFlow attributes to Harvard Business Review. Two million hours at Deloitte is Deloitte’s history. It is not your cycle, and it is not the $242,250 example. A post that says “research shows performance reviews cost $2.4 million to $35 million, or $303 a head” has stacked three objects and deleted the sample sizes.

What to count before you add another meeting

If the only number you take from this week is 44%, take the noun with it. In the Betterworks poll HR Dive reported, 44% of the leaders could provide a business metric for a claim that close to 85% were confident about. If the only number you take is 91.6%, take that noun too. In the Talent Strategy Group survey EvalFlow cited, 91.6% of organizations had a formal process. A formal process that cannot show a business metric is exactly the gap between the two pages, and it is a gap you can only claim if you remember they are different samples.

The number I would actually put on a team card, because it is about time rather than confidence, is the feedback split. 47% expected three or four conversations. 68.5% got one or two. That is a calendar you can audit without buying a suite. The training figures are the other audit. If calibration is in use, the citation says 84.7% of organizations use it, and mandatory training for the review itself is 19.2%. Those two can be true in the same organization. They are a bad surprise only if the rollout assumed the training was required.

I am not going to close this with a better process. The October pages did not publish one. They published a confidence gap, a process census, a conversation gap, and a labeled example. The work is keeping the nouns attached to the percentages when someone tries to average them.