From: CMO
To: All Staff
Re: Changes in internal data analysis and reporting
Consistent with current efforts to streamline our operations, management team has decided to change the way the company analyses and communicates business data. Recent assessments made clear that we are using just a small fraction of the available data, many decisions are not supported by enough data and we react too slowly to changing market conditions because our monitoring tools are not adequate. The ROI we are getting from the data is definitely too low.
There is much to be done, but it would be ludicrous to assume that we can revamp everything at a moment’s notice. We are aware of the need for better tools and serious training in key areas, but the real roadblock is in our minds.
The new plan outlines several steps towards a data-driven company. This memo discusses the first one, code-named The Awakening. It contains a set of simple and practical guidelines that you are expected to follow from now on. You can read the details in the intranet, together with the rational behind each guideline.
The Awakening was designed to make us all more aware of how we manage the data and how we can get more insights out of it. Again, no special skills or training are needed at this stage, just a few changes in our routines. Here are the guidelines.
Regarding presentations:
- Slides and presenter notes should be made available to all attendees; attendees are expected to read them before the presentation starts;
- A presentation should not need more than 50 slides;
- If you are using PowerPoint, animation and other effects are not allowed;
- Branding should not take more than 5% of slide real estate;
- Each slide must not contain more than a single sentence; if you need more, write them in plain English in the notes;
- If the audience is required to compare images (including charts), place them in the same slide;
- You must be able to refresh the data in your presentation.
Regarding the use of charts:
- The use of 3D effects is strictly forbidden;
- The use of pie charts is strongly dis-encouraged;
- No chart should contain less than five data points;
- There is a standard color code for each of our products and major competitor products: use them;
- If possible, use a time series;
- Don’t label data points but annotate outliers;
- Scales should always start at zero when using bar charts and single-series line charts;
- If a pattern cannot be easily identified consider removing the chart, remove or mute data or use a different chart type;
- Consider using the main takeaway as the chart title.
Regarding the use of Excel:
- Shared Excel files containing business data must be connected to the Oracle database or Access files managed by the IT;
- Use PivotTables instead of formulas when possible;
- A standard chart or report must include all our products and markets, should allow the user to select the product she’s interested in and should be easily maintained and updated;
- John Doe is our new Excel and PowerPoint go-to guy; if you need help implementing any of these guidelines please contact him;
- John Doe will also identify training needs in Excel, PowerPoint, data analysis and data visualization.
(…)
OK, this is a draft of a memo I’d like to see coming from every company management team as a starting point towards a better data analysis and communication strategy. In the spirit of a small-steps approach, what would you add, change or remove from this draft? Add your comments below.
Nice post – if only all directors cared so much about the validity of their company’s data/BI output! I agree with *almost* all points but wasn’t too sure about the first one: I like to make extensive use of notes on slides and always make them available to attendees but choose to do so after the presentation not before. Firstly because we assume that the presenter has a special knowledge or understanding that we do not – it makes sense that the presenter should be able to present their information without the audience having jumped to their own conclusions/ made up their mind beforehand. And secondly if the audience don’t have the slides in their hands during the presentation, the presenter is in complete control – I find use of the “b” and/or “w” keys are invaluable for holding attention! Sure it depends on the content/industry though =D A.
Arthur: there are exceptions, but as a general rule I prefer a two-way communication and it would be fair to give attendees a few minuts to think about the message, since the presenter had hours or even days to think about it. An active role is usually mre rewarding.
Excellent memo! Well detailed about how to act… Is essential that all users have trainings about data visualization and data analysis. How to know exactly what kind of data they will need to get from several sources…
BR
Nuno
Wondering why you think PivotTables are better than formulas. I prefer formulas because they seem to enable auditing and error resolution. Love to hear a different point of view.
Brian: with PivotTables what you see is what you get, so to speak. This means that you don’t have formulas to audit, nothing is hidden. They also forces you to structure your data. They are faster than dozens or even hundreds of formulas. The function GETPIVOTDATA can easily formatted to get exactly the data you want.
# 1 You are forbidden to READ the Slide verbatim –
I would make a few revisions to this excellent memo.
1. Fifty slides! That’s a lot.
2. Pivot tables do not automatically update when underlying data changes (at least, not in the older versions of Excel I’ve worked in – that are still in use at many organizations). So, I know many people swear by them, but the lack of flow-through made me fear errors, so I find formulas (especially the ridicuously named but useful “array formulas”) superior and more transparent. It’s a hot topic, and most of this memo isn’t really debatable, so I’d take the bit about pivot tables out.
3. There are a lot of good reasons to make presentation data static. This one seems too circumstance-specific to have a general rule.
4. Sometimes just a few data points are just the thing. I once spent more than 20 minutes discussing a slide with a three-number bar chart on it (there was a lot behind each of those bars).
Nice post. I particularly appreciate the visualization tips.
I’m a BI guy and the list of requirements for shared Excel files had me thinking about the challenges associated with spreadmarts. If I were the CMO, I would insist that all shared Excel files are hosted and distributed via a SharePoint library or Excel Services. Let’s eliminate multiple versions and variations in data access.
Todd: That’s why there should be no more than one sentence 🙂
One important point is established by example: Certainly do not prepare a presentation when a memo is both sufficient and superior. The proposed memo on data management is well done. It might be perfected by recourse to Minninger’s techniques* but memo-drafting is a separate discipline better addressed in another forum.
I’d advocate a requirement that all charts (or at least, all packages, once at start and close of the presentation) show author/ analyst name, VERSIONING, DOCUMENT CONTROL ID, source of data, chart creation date (or data collection date) and possibly an hyperlink to the internal webpage/portal/data warehouse archive in which the original data reposes.
A model or rubic for presentations and charts usefully might be derived from the University English Composition 101 requirements for term papers, the constituent paragraghs, the in-line citations, and the footnotes or bibliography. Typically such a term paper rubric would require a header with author/class code/date/short title/ page number. A short paper would refer to cited works by asterisks indicating footnotes at the bottom of the page. A longer paper might use an parenthetical reference to an authors, year of publication, and maybe a page number (or figure number!) — such citation expanded at length in the end pages of bibliography. The particular rules implementing “cites” will be new and different. The general principles seem to me more eternal.
(http://abclibrary.wikispaces.com/file/view/Turabian_style_and_formatting.pdf)
A difference between the traditional term paper and a 21st century presentation is the location of navigation aids. I recommend the end of a presentation, at the time reserved for audience Q&A, the final “slide” be a hyperlinked “table of contents” review of the three to five main points or sections of the whole piece. When, as we come to slide 49 and the slow-deep-hard thinking senior member of the audience rumbles out his question about “that bit near the beginning, your third or fourth chart” — there ought to be a way to get there more quickly than repeatedly hitting a mouse button and scrolling backwards through the entire presenation. (such scolling interpersed with commentary: “This one?” “No, earlier.” “How ’bout this?” “No, it was after that.” … )
Along with the note about standard colors (USE THEM!) ought to follow a comparable note that most presentation and data analysis tools come with spellcheckers. USE THEM! (the Turabin example above embarrassingly misspells the word “author” at one point. )
The thing about zero baselines … This is a current sore point for me. Charts should illuminate the significant information. A bar chart comparing, say, student attendance percentages (e.g. 97.2% 96.9%, 95.2%… ) doesn’t help much. A baseline of minimum tolerable value (90%? ) would be more useful. Or, the data could be converted to the inverse (in this case, “absenteeism”: 2.8%, 3.1%, 4.8%… ) to highlight differences (from 0%) that make a difference. Here the point is that the rule about baselines shouldn’t overwhelm the more fundamental rules about using a chart to tell the story.
* Minninger, Joan. The Perfect Memo. New York: Doubleday, 1990.
pouncer: thanks for the tip for better memos… Regarding zero baselines: if you want to compare data like student attendance percentages you probably shouldn’t use bar charts. A dot plot is a much better chart for that be cause you don’t have to start at zero.