I work with a fairly large team across departments. One of my coworkers, whom I've never met in person, is polite and responsive. However, the data that they send to me for analysis is rarely accurate. The problems include out-of-date spreadsheets, data for the wrong projects, and most often data that is just wrong (numbers aren't correct for whatever reason).

I generally like this person but their requests take me days instead of hours since I have to correct my results so often - twice on average and on this task 3-4 times. I'm not worried about budgeting since their project budgets are high but re-doing my work constantly is irritating and takes me away from other tasks.

How can I politely ask this coworker to ensure that they're sending me the correct data?

Edit: I should add that I don't have any way of verifying this data. Think of someone reporting rainfall from their back yard - but they don't tell me where their yard is. So the only way that I know something is wrong is when they tell me that they made a mistake and I need to do the analysis again.

Edit2: I requested a double-check of their data and they assured me that they triple-checked it. However, they sent me an email this morning correcting their triple-checked results.

  • Why can't you just highlight errors and send it back for correction/clarification?
    – Kilisi
    Jul 13, 2021 at 19:53
  • Good question. I don't know where the errors are. I get a list of numbers that represent chemical concentrations from specific wells, so I don't have a way of verifying if data is correct. Jul 13, 2021 at 19:54
  • Then how do you know the data is wrong? Out of scope?
    – Kilisi
    Jul 13, 2021 at 19:56
  • 1
    @DanielR.Collins that's a good question and I have no idea. The datasets are based on elevations, which are determined from values obtained in the field. So I don't know where the changes are coming from. Jul 16, 2021 at 19:32
  • 1
    Thanks. You (maybe in cooperation with your manager) should ask. Jul 16, 2021 at 19:36

3 Answers 3


Suggest that you want to learn more about their process, and ask if they could walk you through it a few times. They may find their errors more easily if they have to explain their process to someone.

Think "rubber duck debugging"… they may become more aware of what they are doing and thus catch the errors.


Just request that they doublecheck their data before submitting it to you. Rinse and repeat every time. This gives you some metrics to cover your back and show what is happening.

If it's a persistent problem I'd ask my manager what should be done. Theoretically the manager will then sort it out, or at least give a guideline.

You shouldn't engage in a potential conflict with a colleague, it can be handled easily enough at a higher level.


The problems include out-of-date spreadsheets, data for the wrong projects, and most often data that is just wrong (numbers aren't correct for whatever reason).

Is it clear to them what data they should be using? In a past job, it was quite common for someone on the finance team to ask a dev to write some SQL for a report. The problem was that they were not always clear on what data the report should be run. For example, the request "please get me summary data for city usage" could be interpreted to mean any number of things, many of which hinged on how you defined "city."

  • Did they mean the city, as in the municipal government?
  • Did they mean the city, as in the municipal public works business unit inside the city (if often meant this). I worked for the city, but not for the city as a business unit.
  • Did they mean the city, as in all the units in the city?
  • Which city did they mean? Sometimes it referred to all the cities and they wanted a summary of each city we operated in.
  • Often the Finance team had updated data. The problem was that we needed to be informed of that for the reporting to work. It did not always happen.

It is hard to say if this is the problem you face, but in my example it would have been no use telling the engineer to check his work, as the problem was what work should be done and which data sources should be used, not anything wrong with the code itself.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .