Last week’s article covered some high-level tips for better marketing and communications data collection. This week I’m continuing the theme with nine things to consider when it comes to data analysis and storage - the second phase of the data life cycle.
Data analysis and management is sort of like the forgotten cousin of the data life cycle. It isn’t interesting and invigorating like data collection, or fun and sexy like data presentation.
But it’s part of the family nonetheless, and you need to give it a bit of love and attention to make your data-based project a success.
Recap: The data life cycle
In my previous article, I explained the three basic phases of the data life cycle:
Collection: The phase of work where you collect the raw data, e.g. running a survey or conducting interviews and focus groups.
Analysis and management: Crunching and giving context to the raw data (i.e. turning it into information) and storage of the data - something which is often done poorly.
Presentation: Working with a writer and/or designer to bring the research to life, usually in the form of a white paper, eBook or report.
Hopefully the importance of proper data analysis is obvious to you - without it, data is meaningless at best, and misleading at worst.
The topic of data management might seem to be more of a tedious piece of admistrivia, but doing it correctly is essential to build credibility as someone who collects and publishes data for marketing or brand-building purposes.
Plus, when data is managed correctly, it offers far more ‘bang for your buck’. For example, you'll be able to come back to the raw data one, five or 10 years down the track and find something new to communicate to the market, even if it’s just comparing ‘now’ to ‘then’.
Disclaimer: I’m not a privacy or legal expert, so it’s always best to work closely with your organisation’s experts in this area when you’re running any sort of survey or data collection exercise.
Tip 1: Don’t get too creative
As marketers and communicators we need to be really careful with interpreting the data - we are (mostly) creative beings, after all.
Is the data really saying what you think it is saying? We’ve all got things to sell, but avoid the temptation to twist the data to extreme and dishonest lengths to make your case.
Ultimately, content marketing is about building trust and manipulating the data in a dishonest way is a clear breach of that. Moreover, it’s highly likely it will be noticed by someone, at some point. This could damage your credibility, as the study’s owner or author; as well as that of your organisation, as the sponsor or publisher.
Tip 2: Get help if you need it
If numbers and data are a weak point for you, I highly recommend enlisting outside help with the number crunching process. Statistics ain’t easy, baby!
If you think you have enough numerical nous to do it yourself, make sure you familiarise yourself with the basic statistical concepts you’ll need to interpret the data.
SurveyMonkey has a great introductory article that will help you with some of this.
Tip 3: Look for soundbites… but use your judgement
When you are going through your qualitative responses, you should be looking for pithy quotes and soundbites to use as pullouts in the survey report, and in supporting collateral such as the media release and client emails. Look for quotes that not only read well, but are also an accurate reflection of your dataset.
As an ex-journalist myself, I’ve often found the ability to identify a really catchy soundbite can be a weakness as well as a strength. It can be tempting to go overboard when something reads well and give it more prominence than it deserves.
If the quote isn’t an accurate reflection of the overall sentiment of your responses, don’t try to present it as such. In this case, you can still use the quote, but contextualise it as an outlier rather than as a representative of your sample.
Tip 4: Keep a copy of everything
Whenever you’re handling data, you should always be prepared for the possibility you’ll be asked to provide evidence of compliance with legal requirement A, or obligation B. So open up a new file on your shared drive and get ready to do some serious filing!
It’s important that you keep a copy of everything used to gather your data - from the covering emails and social media posts directing people to the survey, to the script for any verbal instructions provided during qualitative interviews, to the privacy and/or consent forms used during the study.
And, of course, a copy of the survey or study itself. More on that in the next tip...
Tip 5: Make a code book and matrix - not as interesting as it sounds
Most online survey tools such as SurveyMonkey automatically keep copies of all your surveys and individual responses in your account. However, you should also keep a version of your raw data in the special survey folder you’ve made on your organisation’s network.
There are several reasons for this, the main one being ownership of your data. You don’t own SurveyMonkey or any other online survey platform, and therefore have no real control over what happens with your account and data. The terms and conditions of use could change at some point in the future, or information could be lost or deleted due to circumstances outside your control.
The neatest solution is to put together something called a data matrix for each survey or study you conduct, where rows represent participant responses and columns represent questions. This is fairly easy to do in Excel or Google Sheets.
It’s also useful to have a second document to contextualise your matrix, often referred to as a code book.
The code book should explain each question and the range of possible responses, including when a participant did not answer (also known as a missing value). More information on code books can be found here.
This can also be done in Excel or Google Sheets.
Tip 6: Store personal data for as long as you need it… not a minute longer
There are several legitimate reasons for requesting personal data from your survey respondents. For example, you might be running a prize draw or other incentive to respond. Or you may give respondents the opportunity to request a callback post survey about your company’s products and services.
Once you’ve fulfilled whatever reason you have for using this personal data, delete it from your files. It definitely shouldn’t make it into your matrix and code book file for longer term storage.
This is especially important if your marketing is targeted at residents of any European Union countries, which means you need to be GDPR compliant. More on that here.
Tip 7: Store responses for 10 years
It’s best practice to store raw survey data for at least 10 years - de-identified, of course (see previous tip). After all, data is like virtual gold and you’ve done all that hard work to mine it - why would you just throw it away?
Your matrix and code book can live quite happily on your company’s network for this period of time - just make sure survey files are part of your handover notes if you decide to move on from an organisation.
Tip 8: Sharing your raw data
If you’ve conducted a survey or study of any significance or interest in your industry, you’re almost guaranteed that you’ll be questioned about at least one data point. Whether that be how it was collected, or your interpretation of it.
Take it as a compliment. It means people are paying attention!
For transparency, you might want to consider sharing the raw data (de-identified, in the form of the matrix and code book you created at the time of collection) with whoever is doing the questioning and allow them to draw their own conclusions. Even if their interpretation is slightly different to your own, it shows that:
Your data is genuine, and
You’re open, trustworthy and committed to building collective knowledge of a market or topic.
Of course, this approach isn’t recommended if the person doing the questioning works for one of your direct competitors.
Tip 9: Don’t forget your sources
Last but not least, you should also add to your survey file any copies of external sources you used as part of the survey or study. For example, perhaps you referenced someone else’s work or another article to contextualise your own data in the final report.
Things come and go from the internet all the time, and articles can be easily changed. If you’ve referred or linked to another piece of research or commentary, always take a screenshot and make a note of the date you took the screenshot.
This way you’ll be covered if the source changes or is taken offline, and someone questions your work on this basis. It’s okay when this happens, by the way - people often change their conclusions based on new thinking or available data. It’s part of the scientific research process!
The important thing is making sure you have a clear audit trail of your own work, so you can show your thinking was genuine and correct based on the available information at the time of publication.
Wrap up and coming next fortnight
There’s so much more to survey data analysis and management than what I’ve covered here - but hopefully it’s enough to get you thinking before you kick off your next data-based marketing or communications project!
Next fortnight I’ll be wrapping up this short series on data for marketing and communications with some tips on the final phase of the life cycle - presentation.
Comments