Protobi can help your team save time in analysis and provide new deliverables for your clients. Here are a few tips to save you time and get results to clients faster.
1. Number questions simply
Protobi's Autogroup arranges questions alphanumerically, so use prefixes to suggest a structure.
For instance, if your survey has sections like "Screener" and "Attitudes", then:
- Screener questions might be S1, S2, S3a, S3b, S4,...
- Attitude questions might be A1, A2, A3,...
If Q101 is a series of similar questions: Q101_1, Q101_2, Q101_3,...
Protobi's autogroup feature will recognize Q3a and Q3b as part of a group Q3. So if you have a question that's not really part of Q3, try not to call it Q3a but maybe Q4 or Q49 instead.
2. Retain question numbers across waves
For tracking studies and other surveys are fielded over multiple waves, it's common for the survey to evolve as your clients' understanding of the market increases.
Just keep question numbers consistent if possible. If you drop a question, just retire the question number rather than re-use it. If you add a new question, use a new number (even it's not in numeric order, see above). That way Q1 in one wave always means the same as Q1 in another wave. Ditto for response options.
The same principle applies for related surveys that are “almost identical" but differ by market.
3. Start your project with a SPSS file
SPSS files contain useful metadata, such as question titles and value formats.
Survey configuration file
You can even ask your survey programmer for the configuration file from the survey engine, such as Confirmit, Decipher, Qualtrics. The XML contains even more data about your survey sections, skip patterns, and other things that can give you an advanced start.
4. Start from raw data
We recommend starting each Protobi project with data straight from the survey engine.
5. Clean data in processes, not in Excel
It's pretty common that the raw survey responses need to be changed before they're ready for analysis and reporting, for instance:
- Create and apply weights
- Temper data
- Date and time calculations
- Remove respondents
- Sum variables
- Ratio calculations
It is tempting to just download the data, modify it in Excel and upload it back.
But your life will be easier (and your internal team hours lower) if you make the changes programmatically in a data process. This way the changes are documented clearly, you can change the rules later if you need, and you can reapply the rules as new cases come in.
If your team does make changes to the raw survey data prior to uploading into Protobi, let our team know. We recommend applying business rules and data cleaning within the tool so that every change in the data is documented within the project.
6. Setup QC checks in Protobi
Even while your data is still in field you can setup quality control checks that flag curious responses or respondents.
One approach is to use the QC dialog (shortcut X C) for a selected element. This will create criteria for flagging curious responses.
Another approach is to setup such rules in a data process, and you can use this to correct or
- Correct or remove certain responses from a question
- Filter entire rows out of the data
Identify “high risk” questions such as client KPIs for QC and mark these elements with Protobi notes or in a change log.
7. Keep element “keys” simple
Protobi allows any unique string as the “key” to identify an element. But your life will be a little easier if you choose keys that would be valid variable names in other programs like SPSS or R.
Typically start with a letter, and stick to letters, numbers and underscores after that. Avoid special characters like spaces, dots, dashes, hashes, plus signs, and other punctuation.
This way it's easier to export data and use it in other programs. For instance "Q3_sum" as the key, instead of "Q3 (sum)".
We recommend being consistent about upper-lower case within a project. You can mix case, just be consistent, following the convention set by your survey numbers or other pattern. Protobi is case sensitive, but other programs like SPSS are not so it's best to have either a Q3 or a q3 but not both in the same project if you can avoid it.
8. Use display keys for readable titles
The element “key” is a unique identifier for bookkeeping, but for display, it's fine, even recommended to choose a prettier display key for humans to read. So Q3_sum might be displayed as "Q3 (sum)",
9. Decide how to treat missing values
Missing or blank values can arise for a variety of reasons, typically skip patterns, and you'll need to decide how they should be displayed in your survey.
Missing values are displayed in Protobi as [NA], and are visible by default. How you choose to handle them will depend on each specific question and your analysis goals.
Align with your team on how to treat missing values early on so the base for each question is clear and the element is showing the numbers you want to report.
10. Identify project leads on each side
Identify the point people on your team and Protobi, and make sure everyone is aware of project changes and progress.
Identify task owners
Clearly define which project edits are for Protobi to make versus which ones your team will handle.
During team transitions
Modern market research surveys can get complicated, and over the first weeks or waves, a lot of details accumulate. These details can get lost in transition, so where one person created sophisticated weights, the next analyst might not remember to apply them.
- Schedule work sessions and tutorial sessions via screen-share
- Create a project intro wiki page with key project details
- Document changes in questions and data cleaning methods
- Contact our project support team (email@example.com)
11. Single channel to log changes
Choose a primary channel for communicating changes and tracking progress, so that we all know whether and when to look to the the notes feature in Protobi, Google Sheet, and/or email.
Each element in Protobi has a notes feature that lets you keep track of details about the details. Early in the project you might use notes to mark changes you'd like to see made, note questions for your colleagues or Protobi support, or possible insights you might draw.
Later, as those early comments are cleared out, you might replace them with additional details about the chart that are useful to know or have on hand, but avoid cluttering the main view.
Google change log
If you prefer to log project changes in a spreadsheet, we can create a Google Sheet and link it right in the project.
12. Upload related documents
Beyond securely storing your survey response data, you can upload other related documents too:
- Survey document
- Analysis plan
- PowerPoint report
This way they're handy for your team when you need them later.
Protobi's permission rules apply to these documents too – only colleagues to whom you've granted admin access to the project can see or download them.
13. Use Protobi “wiki” pages
Protobi Wiki pages allow you to create special pages you can use to accompany the project. It's literally a blank canvas for your creativity, and can include text, images, files, even live elements:
- Add survey context, expectations and a project timeline
- Document project changes such as question additions and question removals between waves
- Document details like weighting or tempering applied to the project
- Create a companion “About this research” page for end clients
- Make it the actual final report
14. Polish for client delivery
Follow our tips to make neat and engaging projects that you'll be excited to share with clients.
15. Lean on your Protobi team!
We're here for you as needed for project-specific analytic support, with hourly rates, and but we're also available for technical support to teach you how to use Protobi efficiently.
Particularly when you're new, lean on our support team. Use the Questions widget to schedule screen share work sessions as needed. We let you know before our time would be billable.
It just pains us to hear a story like “We didn't want to bother you on a Friday so we spent the weekend doing [some time consuming thing you could totally have streamlined]”. Don't do that.
Early on it might seem like we invest a lot of time getting your team up to speed, but we find experienced clients can do a lot with very little support, and love getting you to that point.
16. Identify your in-house Protobi experts
Research professionals are a diverse group of people. But in each firm there's almost always at least a couple analysts who go way up the learning curve to master Protobi's features. (And we're happy to foster that, see tip #16 above)
You may not be able to actually staff an analyst from another team on your project, but it can be very helpful to identify the go-to people within your firm as internal experts.