Picture this: you're updating salary bands for Senior Software Engineers in Germany. One reputable dataset puts median base salary at €65,000, while another says €75,000.
That's a substantial difference when you're using market data to inform salary bands, hiring offers and compensation reviews. Using a poorly matched benchmark to guide these decisions could lead to pay outcomes that don't reflect the market you're competing in.
The goal isn’t necessarily to find out which benchmark is “correct”. Both could be valid for the markets they represent. Instead, it’s about understanding why the benchmarks differ, which source best reflects your talent market and how you'll justify your decision. Keep reading to learn what that process looks like in practice.
Why do salary benchmarking sources disagree?
A salary benchmark is an estimate drawn from a particular dataset, not a universal price tag for a job. NFP's guidance on using multiple data sources identifies differences in survey participants, timing and job matching as common reasons for variation.
Differences in job matching and levelling
A Senior Software Engineer in one survey may be an experienced individual contributor. In another, the match might include technical leads with wider responsibility for architecture, mentoring or delivery. Even where providers use similar titles, their job families and levelling frameworks may not line up perfectly.
This gets even harder with hybrid roles. A person who manages a small team of engineers but spends much of their week coding may not fit cleanly into either a pure engineering or management benchmark.
Different comparison groups and talent markets
One dataset might draw heavily on large technology employers, while another includes companies across many sectors and sizes. The pay levels can differ because the employers represented are competing in different markets.
Geography also plays a huge role. A Germany-wide figure may not reflect the same hiring market as a Berlin-focused one. For remote roles, the distinction between the employee's location and the employer's can add another complication.
Differences in data collection and methodology
Depending on the source you use, the headline figure might mean different things. One provider may show base salary while another reports total cash compensation, including incentives. One may report the median and another a different percentile. Survey periods, sample sizes, data validation and the treatment of unusual values can also vary.
It’s also important to look at the date each data source was collected, as market movement can have an impact on accuracy. However, newer doesn't always mean better: an older survey might be a better fit if it has stronger job matches or more relevant participants.
How to handle conflicting salary benchmarks in 3 steps
Without clear criteria for choosing between benchmarks, it's easy to fall into the trap of selecting whichever figure supports the decision you already want to make.
In our example above, a hiring manager pushing for a higher offer might favour the €75,000 figure, while someone trying to contain salary costs might prefer €65,000. Both can point to credible data, but neither has established which benchmark actually reflects the relevant talent market.
The point isn’t to prove that one provider has the right answer. You just need to find a market reference that makes sense for your company and that you can justify to others. Here's how to approach it.
- First, check that you're comparing like with like
Start with the underlying survey entries, not the two headline salaries. A quick check can reveal that the figures were never directly comparable. For example, look carefully at:
- Role and level: Do both matches cover the same responsibilities, decision-making scope and seniority?
- Pay measure: Are both figures annual gross base salary, in the same currency and for the same working-time basis?
- Statistic and location: Are you comparing the same percentile and geographic market?
- Reference date: Are the figures from the same period? If not, should you 'age' the older data to a common date by applying a documented estimate of salary movement?
Take our German engineering example. One possibility is that the €75,000 match covers a technical lead, while the €65,000 match describes a senior individual contributor. In that case, these aren't competing estimates for the same job. Correct the match before deciding which data to use.
For an unusual role, speak to the hiring manager about what the person will actually be accountable for. As WorldatWork's article on difficult job benchmarks points out, clearer job documentation can be more useful than another round of data gathering.
- Evaluate the relevance and reliability of each source
Next, look at the population behind each result. Which employers contributed? Are they comparable in size, sector or operating model? Does the survey cover the organisations you're competing with for that talent?
Alternatively, both German engineering benchmarks might describe the same senior individual contributor role, at the same percentile and reference date. The €75,000 dataset draws mainly on technology employers, while the €65,000 dataset covers a broader range of industries. Both could be reliable estimates for the markets they represent. The question is which employer group better reflects where the company recruits its engineers.
Ask the provider about sample sufficiency, participant concentration, collection methods and validation. Pay attention to whether a figure is based on actual incumbent salaries, advertised ranges or self-reported estimates. WTW's guide to choosing compensation data makes the point that these sources answer different questions and have different limitations.
💡Is more data always better?
A larger sample can improve statistical reliability, but only if it represents the market you're trying to measure. A smaller, well-matched dataset may give a more useful reference than a much larger, less relevant one.
- Decide whether to prioritise one source or combine them
By this point, you should have a clearer picture of why your two benchmarks differ and how well each one represents your talent market. You can use that assessment to decide whether one source deserves more weight, or whether it makes sense to combine them.
If one benchmark is a substantially better match for the role and relevant talent market, use it as your primary reference. The other can provide additional context.
For our German engineering role, a company competing mainly with technology firms might choose €75,000 as its primary market reference, retaining €65,000 as broader context. An employer recruiting engineers across different sectors could reasonably favour the €65,000 benchmark instead. The choice should reflect the market the company is recruiting from, not a preference for the higher or lower figure.
If both sources are credible and genuinely comparable, consider whether a combined reference would be useful. You might give greater weight to a survey with more relevant participants or closer job matches, while using the other for broader context. Document the reasoning behind that weighting, rather than choosing percentages to produce an appealing number.
You should also check for overlapping survey participants. If there's significant overlap, avoid treating the datasets as independent when combining them, as you risk counting the same employees twice. And remember that averaging two published medians doesn't give you the median of all the underlying salaries, particularly if the datasets differ in size. Treat the result as an estimate, not a statistically calculated market median.
The table below can help you decide what to do in different situations.
💡Tip: Don't automatically average conflicting benchmarks.
The midpoint between €65,000 and €75,000 is €70,000, but that doesn’t make it a sensible market reference. An equal-weight average gives a weak match the same influence as a strong one. Combining sources is a deliberate methodological choice, not a way to split the difference.
What if neither benchmark is convincing?
Of course, all of this assumes you have at least one reasonably good benchmark to work with. That's not always the case. Emerging technical roles, unusual combinations of responsibilities or small local markets can all leave compensation teams with thin evidence at best.
If you find yourself in this situation, start by checking whether a better-matched survey or adjacent role can provide context. You might also look at relevant recruitment evidence, such as accepted offers, candidate expectations or advertised ranges. Treat these as supporting signals rather than direct measures of market pay.
Sometimes the most accurate conclusion is that the available data doesn't support a precise number. Record the assumptions behind the provisional reference, flag the uncertainty and revisit it when new evidence is available. A tidy figure with no credible explanation is not much help.
Making benchmarking part of your compensation process
Whether you've settled on a strong market reference or found a way to work with less-than-perfect data, the reasoning behind your decision needs to be clear. Particularly when different managers or HR teams are using benchmarking data across the business.
Start by establishing some basic rules for how your team selects which market data to use. For example, you might use one benchmarking source as the default for corporate roles in Germany, but a specialist technology survey for certain engineering positions. When someone departs from the usual approach, document which sources were considered and why the exception makes sense.
This doesn't need to become a lengthy approval process. A short record of the reasoning used helps other compensation managers apply similar logic in the future.
Then there's the question of what you do with the market reference you've selected. A market median doesn't automatically tell you where to set a salary band or what to pay an individual employee. Those decisions depend on your compensation philosophy, and existing salary structure.
Internal equity is another consideration. For example, if a new benchmark suggests substantially higher pay for a role, you'll want to consider the implications for current employees before adjusting salary bands or hiring offers.
Making better use of multiple benchmarking sources
Conflicting benchmarks are easier to manage when you can see what each dataset represents, compare the evidence and explain why a particular reference was chosen.
Figures makes it possible to bring together different benchmarking sources, including Figures' own data, Mercer and your company's surveys, in one platform. You can switch between sources across your compensation workflows, with a clear view of where each figure comes from. That makes it easier to work with the data most relevant to each decision, while keeping your approach consistent and explainable.






