Lovers and leavers: when switching employers is the right move

Key Findings

What makes a dream job turn into a nightmare? Can a dissatisfied worker turn it around and love their job again? We analyzed Glassdoor reviews from 268,226 unique individuals who left multiple employer reviews between 2021 and 2025. The workers were evenly split between stayers (52% listed the same employer for both reviews) and leavers (48% switched employers). Forty-eight percent of workers gave the same rating in both reviews while 23% of ratings improved and 29% fell.

Should I stay or should I go?

Among these repeat reviewers, those who switched employers were much more likely to give different ratings (62% vs. 43% of stayers) - but your level of satisfaction with your current employer is a major factor in whether you’ll make a switch. This table shows how the likelihood of switching employers increases from 39% for the most satisfied employees to 71% for the least satisfied.

Workers also tend to be more mobile earlier in their careers. In a similar vein, the longer an employee stays with an employer, the less likely they are to leave. 

Among all subgroups, the least likely to leave (21%) are those who have been with their employer for 10 or more years and give their job a 5-star rating, while the most likely (87%) are those who have been with their employer for less than one year and give it a 1-star rating. Across each tenure or experience level, workers grow more likely to leave as their initial ratings fall. On the other hand, with any initial rating level, workers are less likely to switch employers the more experienced they are or the longer they’ve been with their employer.

Leavers are more likely to find more satisfying work

Workers who had room for improvement in their ratings (1 to 4 stars) are much more likely to see those ratings improve if they switch employers. Dissatisfied workers (1- or 2-star reviews) who switch employers are 89% more likely to give a rating of 3-stars or higher in their next review than those who stay. Workers giving more positive ratings (3- or 4-star) are 2.0x as likely to achieve a 5-star rating when they switch employers.

The table below shows a transition matrix for employees based on their initial rating and whether they switched employers or not between reviews, with the center column showing the likelihood of the rating staying the same between reviews.

The right two columns show the odds of ratings improving by 1 or 2 points, so a worker with a 4-star Glassdoor rating who switches jobs has a 35% chance of their rating rising one point, compared to 21% for one who stays. The odds of improving ratings are higher for job switchers coming from each initial rating category. The odds of a marked improvement - 2 stars or more - are much more pronounced for leavers whose initial ratings were 1, 2, or 3 stars. Workers who gave 4-star ratings were also more likely to give 5-star ratings after switching employers. The chart below visually shows how dissatisfied workers are much more likely to end up at 4 or 5-stars when switching employers than if they stayed after a negative review.

It’s not just about getting out of a situation you are not happy with, however. It also depends where you land. We analyzed workers whose new employers were among Glassdoor’s Best Places to Work in 2026, and found that switching to an employer on the list gave even better odds of an improving rating. Second ratings from workers whose initial ratings left room for improvement (1–4 star ratings) were 90% more likely to improve if they switched employers - but more than twice as likely (+130%) if they switched to one of the Best Places to Work.

The bottom line is that all workers who believe their jobs could be better - but particularly the most dissatisfied - are more likely to improve their ratings when they switch employers.

Staying carries its own risks

In the first two rating columns in the table above, we can identify the downside risk. In the case of 2- and 3-star ratings, the second rating is more likely to be worse when you stay with the same employer. For workers who gave 4-star ratings, the downside risk of switching is somewhat higher: 7% of those who stayed saw ratings fall 2 or more points, vs. 11% of those who switched roles. The combined probability of a falling rating for lovers and leavers is similar after a 4-star rating.

Among the most satisfied with their first ratings, staying is less risky than leaving, particularly in terms of big negative changes. Satisfied workers who switch employers are 2.1x as likely to have their ratings fall by 2 stars or more than those who stay - though the majority of both groups give another 5-star rating in their second review, making this the most stable of any rating.

These trends are similar when workers switched occupations or industries. They also hold for new and seasoned workers alike, though there is one exception: when long-tenured workers who gave 5-star ratings did choose to switch employers, they were about 10 percentage points more likely to give the new job a 5-star rating than lower-tenured groups in the same situation. From the first section, we can see that this group was also the least mobile: only 21% switched employers between groups. This may indicate that they only left for high-quality offers, as they were the most likely of any initial rating-tenure combination to give a 5-star rating in their new roles.

Drivers of rating changes

We showed that initial ratings, tenure, experience, and whether or not the worker switched employers all have a big influence on their second rating. What this doesn’t tell us, however, is what drives these changes. In this section, we make two attempts to answer this question using Glassdoor reviews: additional sub-factor ratings employees can give (1-5 star ratings of career opportunities, senior management, and more), and the free text where reviewers list pros, cons, and advice to management.

To analyze rating subfactors, we calculated the probability of ratings improving or worsening for employees who stayed or left using a logistic regression with changes in sub-factor ratings as explanatory variables. Each subfactor had a positive relationship with the overall rating, so they rose and fell together. The table below shows how a 1 star rise or fall in a sub-factor ratings impacted the odds of the overall rating rising or falling, so a 1 star decrease in the career opportunity rating is associated with a 76% rise in the odds of the overall rating falling.

Career opportunity ratings had a very strong association with overall ratings for workers who stayed. When ratings fell for stayers, ratings of culture & values and senior management also fell. For leavers, overall ratings rose with ratings of senior management, career opportunities, and culture & values. As with stayers, a falling culture and values ratings greatly increased the odds of a falling overall rating. Work-life balance and compensation & benefits ratings had the weakest associations with rising or falling ratings for both stayers and switchers.

To analyze rating content, we built a set of keywords based on themes we encountered for two types of stayers: those whose ratings fell from positive (4- or 5-star) to negative (1- or 2-star), and those that went from negative to positive. As a reference group, we tracked keyword prevalence for workers whose ratings were 3 or 4 for both reviews. The table below shows how much more (or less) prevalent keywords were in the first and second reviews for these “swing” groups as opposed to the reference group.

When ratings fall from positive (4- or 5-star) to negative (1- or 2-star) at the same employer, the reviews are 2.0x as likely to discuss burnout, 3.6x as likely to mention layoffs, 2.5x as likely to mention mergers or acquisitions, and 4.2x as likely to discuss management changes. 

These events can also lead to rebounds, however: reviewers whose ratings go from negative to positive were more likely to mention burnout (1.6x), layoffs (2.2x) and mergers/acquisitions (2.5x) and management changes (4.4x) in their first reviews - consistent with the first review occurring after burnout or a negative event. Our research on layoffs over the same time period found that average ratings for layoff survivors did recover after roughly two years. Management changes remained a common theme when bad ratings turned positive.

Conclusion

By tracking Glassdoor users over multiple reviews, we found that individual worker’s ratings are dynamic over time and when they switch employers. Negative ratings are associated with more employer switches, and workers tend to be more satisfied with their new employers - unless they really loved their old roles.

While our analysis doesn’t differentiate between why these workers switched employers (promotions, pay increases, involuntary terminations due to a layoff), we found evidence that burnout and instability - represented by layoffs, mergers & acquisitions, and leadership changes - can make the most engaged employees start to hate their jobs.

Methodology & Acknowledgements

We identified Glassdoor users who provided multiple reviews in the 2021-2025 window. For each, we chose the most recent two reviews that were at least 6 months apart and no more than 2 years apart. Glassdoor users are limited to one review of their current employer every 12 months, so stayers’ reviews would have been 12-24 months apart. We pulled total relevant years of experience from salary reviews from the same users over the same time period, providing experience estimates for 88% of the users in the sample.

For the logistic regressions described in the “Drivers of rating changes” section, we subset the data to employer stayers or switchers whose ratings could improve or worsen (so dropped ratings of 5 when analyzing rising ratings, and ratings of 1 when analyzing falling ratings). The dependent variables were the changes in rating subfactors, including a control for the level of the prior review since lower ratings were more likely to improve than higher ratings. Since rating subfactors are optional and one or more was missing for 36% of users across the two reviews, we imputed the rating subfactor delta at the grand mean.

This research was prompted by a question from Tim Ballard, PhD, who also provided helpful feedback on preliminary results.