15 Must-Follow Rules for Useful Analysis from a Former BI Analyst

Your vision is to have solid analysis reported to you in various areas of the business.

Often, you find reports are not quite hitting the mark on providing actionable results and insights but maybe you can’t quite put your finger on what a ‘standard’ should be of items to ensure are included. It could be you or your team that you are trying to get to provide better analysis and reports of the business. The goal of this article is to provide a list of standards, rules, or components needed for every report that is run so that you or your employees are providing accurate results, actionable insights, and the inputs and methodology to get the results are crystal clear. The rules are geared towards the members of your team that will be running the analysis.

The rules to follow to be successful in analysis

Back in my first analyst job, there were some rules if you will, established at the very start that the team followed. Those rules stuck with me the last decade as I witnessed the importance of following them and how senior management trusted these people I was working with for major decisions worth tens of millions of dollars as a result of the analysis by the team. Numbers were never questioned, but that comes after trust is built. These rules have to be followed to provide a complete analysis that can be trusted.

Rule 1 of analysis: Check, double-check, triple-check, and recheck again.

The greatest fear as an analyst should be reporting a wrong number and someone else letting you know during a presentation or after a decision was made. Once you are questioned and errors found one time, or you can’t back up how you got to a number, you’re done. Your credibility may or may not recover. The latter if a major decision was made based on it. It doesn’t matter what people say to your face, that seed is always there.

Rule 2 of analysis: Spot-check.

This is in line with number 1 but make sure your numbers throughout the reporting process, not just at the end, make sense so you can gut check yourself to know things seem within reason.

Example

If you know your product portfolio is 6,000 SKUs, you run a data pull and you did a sum and it’s showing 4,000, you know something went wrong right away before spending more time working with wrong numbers. At a minimum, you should expect 6,000 to know you pulled the entire portfolio before beginning. Something is either wrong in your math, or you pulled a list of SKUs created in the last two years instead of five or something like that.

Rule 3 of analysis: Keep quiet until you really have something valuable to add.

Looking back, it’s somewhat funny, but I wasn’t actually allowed to speak in meetings for close to six months upon beginning the job. The reason for this was to ensure I was prepared to speak intelligently and be able to take action when asked a question. It worked out well as I was able to learn the business and by the time I was presenting, I had respect and I was viewed as valuable to the team, and eventually, it led to a promotion. The point is, take time to learn your data upfront before making suggestions on topics you don’t really know. Let people teach you and show you what they do. This helps build the relationships that you are going to need.

Rule 4 of analysis: Ask questions

At the time, I felt I was driving people crazy, but I wanted to know every detail of the business and as more and more reports were run, it put a light in dark corners and caused more questions. In order to be successful, you have to be comfortable building relationships with people all over the business so you can learn from diverse viewpoints but also how it all comes together. I’ve also found that if you have a question or something that doesn’t make sense, don’t be afraid to ask because the vast majority of the time, the other people in the room have the exact same question but don’t want to look dumb by asking. Every one of them end up relieved once you ask.

Rule 5 of analysis: Never just go with it

There are times things don’t make sense. However, if you’re in a position to be reporting numbers, you can never, ever, ignore when they don’t make sense and let things just slide in your reports without having an absolute solid understanding of why they are there, what it means, etc. Aside from being unable to explain your report, which dings your credibility, you may actually be looking at numbers incorrectly that you can’t explain in every little detail.

Example

Let’s say you ran a data dump from your phone system and there are all types of calls on there. You have to understand the definition of every single call type and what it means before moving forward. If you try to just report on it and say we had 100 of this type and 50 of this one, but you don’t know the definition of those, I guarantee this will burn you every time. It all goes back to number 4 and if things aren’t clear for one person, it’s likely unclear for others. You have to get in the habit of tracing back every type of data in your report and having a solid understanding of what it represents.

Rule 6 of analysis: Be able to answer who, what, why, where, when

When someone opens a spreadsheet, generally, it is a lot of numbers and things to look at. That person opening needs to quickly understand who this is for, who needs to take action, what the numbers and findings mean in terms of results and action to take next, why it was done, where the numbers and information came from, when it was run, and the timeframe the numbers represent. Questions remaining in any of those areas will frustrate the viewer and possibly make them question credibility.

Example

A quick way to convey this information is in the naming convention. If you name it something like “Sales Database Pull Jan 2022 to Jun 2022” you are calling out the what, where and timeframe right there.

Rule 7 of analysis: Labels

This is one of the easiest ways to make reports easier to deal with and it starts with labeling the tabs in a spreadsheet along with headings. Bold them, use color, and more to make them stand out.

Example

If you label your data tab as such and your summaries and visual tabs as such and incorporate dates or months, it helps immensely in the audience to go right to where they need to. Hide any unnecessary tabs.

Rule 8 of analysis: Notes and assumptions section

It’s rare to run into reports where absolutely no assumptions were made and there are no notes needed. Make a section or tab about these. If for nothing else, it will assist you in recalling exactly what was done, definitions, why, and what the findings were.

Example

If you’re pulling information from a tool and you have to make assumptions on the type of customer behavior you are reporting on, you need to make that clear so that audience has context. Let’s say you are assuming all customers use an app vs. a desktop computer and so your data excludes desktop users. This is especially important if you are intentionally excluding certain pieces of information as in this example of the desktop users. The reason for what and why is needed.

Rule 9 of analysis: Summary page

Most often, the audience for the spreadsheet does not want to see all of the information you had to put together to get the requested result. Your boss or team may need to as a peer review, but no one else cares. So that means you need a roll-up or summary page of some sort on a tab that is clear, easy to read, and gives exactly the result asked for.

Example:

You could think of this as financial statements where you see a balance sheet for example if you look up a company like Walmart, but you don’t see all of the backend details that it took to get to all of those totals. The important part for most people are the totals on the actual one-page balance sheet.

Rule 10 of analysis: Visuals

Again, most people aren’t excited to look at spreadsheets so anytime it’s feasible, a graph or visual of some sort is going to help you get more attention span on the topic. A graph or chart makes it easy to immediately pick up on what the results are showing without having to decipher formulas and read numbers.

Example

If you’re reporting on customer sales by region, you could easily include a pie chart broken up by region rather than people having to read for that information. They can quickly see the largest region, smallest, whatever it may be. Make it simple.

Rule 11 of analysis: Timeframe comparison

You can report all day long on something like sales or website traffic but if you aren’t showing a comparison of previous months, it’s not useful. You need to determine which timeframe makes the most sense, but for these two examples, usually, month over month is ideal. The audience needs context or it means nothing.

Example

Looking at sales, going back at least 3 months is necessary, and it needs to be consecutive months for the most part. You don’t want to forget about previous months just because the calendar year changed for example. If you’re in January and running sales, you need January, December, and November at a minimum. Often, it’s useful to see year over year or quarter over quarter as well so you can get insights as to what may have changed and how the trend is looking.

Rule 12 of analysis: Don't short your averages

If you’re looking to find your average purchase per customer as an example, you want to make sure you’re going back again at least 3 months or 90 days. Don’t make the mistake of looking at just one month. You may need to account for seasonality as well.

Example

If you report average sales for this January, listing January is fine, but if the question is more about your average sales in any given month, make sure you go back a few months to get the average or possibly go back a year so you have enough data points to be confident your results aren’t skewed by something like the season.

Rule 13 of analysis: Sufficient data points

Have you heard the term ‘statistically significant’? This means the likelihood that there is a relationship between two or more variables that is not coincidental, but that it’s actually caused by the factor in question. This is how you can be confident in the results you are providing. There are ways to figure this exact number needed for a particular use, but as a general rule, you want to think when you’re reporting results, do we have enough data points for this to be representative of the sample we are looking at?

Example:

If you have 100,000 customers, basing the average purchase price or purchase amount on 10 customers isn’t really representative of your customer base, would you agree? Ideally, you would have 10% as a general rule of thumb, but you don’t need more than 1,000 even if your customer base grows. So in this example, 1,000 inputs would be ideal. Funds, time, your own experience of knowing how diverse your customers are or are not, and other factors will affect how many data points you can get but it’s something to think about when running numbers. Make sure you can communicate this when asked.

Rule 14 of analysis: Tell the audience the next steps

It’s a good idea to add an area to notes and assumptions about the next steps. You deliver a spreadsheet with a lot of information but now what? What do you expect the takeaway to be? Is it to run the numbers again in 90 days, or to take an action now?

Example

Let’s say you ran a report about website traffic. You have the numbers, and you have month over month, and it’s trending down. So now what? What are your recommendations to do or where should you be looking next? You don’t want to just leave it at “well numbers are trending down…” and no further suggestion.

Rule 15 of analysis: Comparison of actuals or post mortem

This is one of the most frequently missed pieces in reporting. Too often things are moving fast but you need to go back and compare actual results to estimates or it could be last month to the next month.

Example

If it’s something like a forecast or follow-up if results were looked at during a particular time frame, make sure the report is set up properly to be able to add actuals, or next month or next year to it. Ideally, you do the initial report once, and you can keep adding to it as needed to start to build a baseline to look at trends.

In summary, performing a useful analysis requires a lot of thought. It’s not just about the immediate numbers, but also about having a solid understanding of your numbers, communicating with others, and coming up with insights into what actions are to be taken. If you incorporate these steps into your upcoming analysis’ I guarantee there will be an improvement.

Written by Nicole Hullihen, January 23rd, 2022