What is data reliability?
Data reliability is the consistency of your data - that you get the data you expect, when and how you expect it. Data reliability comes down to four key components:
Accuracy - Is it the correct data?
Completeness - Does it include everything it should?
Availability - Can it be accessed by anyone who needs to, when they need it?
Usability - Is it in a useful format?
Let’s take a look at how this translates into different contexts and what to do to ensure that you have reliable data.
Data reliability in the context of marketing
Great digital marketing is all about data. It’s how you know what campaigns are working, what platforms are giving the best results and what copy is resonating with your audience. Bad data leads to bad marketing and good data leads (potentially) to good marketing.
In this context, the most important aspects of data reliability are that it’s accurate and useful. Conversions are a clear example of how even accurate data can be difficult to make useful.
Total Google Analytics
The math doesn’t add up, but then again it does: Some of the converting buyers saw your ads on both channels before buying. This is why both marketing channels report the same conversion. It's important to aggregate the data in a way that you can make comparisons, run experiments and drive results.
Data reliability in research
Data reliability (or data quality) are a bit different in a research context. Researchers look at data reliability as whether a research method can produce the same results, multiple times. In other words, they try to measure and reach consistency of the results.
The goal is to prevent data quality issues and ensure data reliability.
What is a data reliability assessment?
To ensure that data meets quality standards, researchers can implement audits or data reliability assessments. While there are many different approaches these tests can take, they typically fall within one of four categories:
Test-retest reliability - Does retesting with the same group produce the same results?
Parallel forms reliability - Does testing for the same phenomena using different methods produce the same results?
Inter-rater reliability - Do you get the same results using different assessors?
Internal consistency reliability - Does asking the same questions in a different way produce the same results?
This way of testing the data's reliability is focused on experimentation. Putting thought into how you run experiments is a strong determinant of the usefulness of your data. It’s important to keep in mind, but not likely to be where businesses would run into their biggest data problems.
When is a data reliability assessment necessary?
While data reliability assessments should be regularly performed, you may want to implement some ad hoc tests if you suspect that your data is flawed. This may be due to some odd patterns in your dashboard visualizations or a broken link here or there.
While small connection errors or duplicates may not be a huge deal once or twice, they may be a signal of a greater snowballing effect. Just like if you have a persistently sore joint or muscle. It doesn’t hurt to go see a doctor to have it checked out by a professional.
Unreliable data and data driven decisions
The consequences of unreliable data can be significant. For example, a company can invest a lot of money in a marketing campaign that seems to be going well, but in reality yields little.
And it can even influence strategic business decisions, such as divesting an industry because the numbers are bad. In a lot of companies, data analysts collect and analyze data to make informed decisions. Imagine making such a decision based on inaccurate data…
Ensuring data reliability
How to make sure your marketing data is reliable
Now that we have a good understanding of the topic, it is time to discuss what we can do to improve data reliability. Let’s go back to the four pillars of marketing data reliability to see how you ensure that it is reliable.
To check the accuracy of your data you will need to measure and compare it using multiple tools. So, you want to compare data between your advertising platforms, website analytics and perhaps eCommerce platform or CRM tool. Or, after adding a graph to your new data studio dashboard, double check the numbers in the graph with the numbers in the data source (for instance GA).
When you look at your data model you need to see if you’re getting all the metrics and dimensions that interest you and are important for measuring your results. For more on this you might want to take a close look at what to include in your digital marketing report.
All the data in the world doesn’t do much for you if it isn’t available when and where you need it. You want to have all the most useful data from each platform consolidated into a marketing report or dashboard that you can easily update and reference. Those who need to view the reports should be able to access them when needed and in a best-case scenario, all relevant users would also be able to work with the data.
With different definitions and even names for dimensions and metrics (e.g. cost vs spend), it can be hard to compare across channels and make use of the data. The important thing here is to make sure that the data can be easily compared and is usable for optimizing marketing spend and creating reports.
The Funnel way of reliability
At Funnel, we live for data reliability. That’s why we built Funnel as a marketing data hub: a secure place to gather, store, organize and then share your data. Here are some ways Funnel ensures your data is accurate, complete, available and usable:
- Large data library
Funnel allows you to collect data and then share it with the destination of your choice. Instead of setting up multiple pipelines or connectors, you bring all your marketing data together in Funnel. Funnel then functions as a single data source containing exactly the metrics & dimensions you need for your report.
One common headache for marketers who work with marketing data is that the reports are slow to load or even break. This is because most reporting API’s have quotas that timeout if a query has too many granular fields or the date range is too long. In contrast, Funnel downloads data in small batches and takes into account platform specific limitations that are hard for end users to be aware of. For you, this means fewer instances of broken reports, so that your data is ready when you need it.
- Code-free environment
Funnel’s platform is built on a no-code logic that anyone can easily master. This empowers marketers to own their data, from creating campaigns to presenting results without having to rely on developers or data engineers.
- Organize your data
Not all data is presentation ready when it comes from a reporting API. Sometimes you want to add transformation logic, like grouping business regions or add calculations to account for consultancy costs. Doing this work in a visualization tool is limited in functionality and not scalable. In Funnel you can finalize all your mapping and transformation logic at your own pace and send to one or multiple reports, saving you time and energy.
It’s not uncommon for business logic to change or for new marketing channels to be added, and when it does you need to be able to push updates to all data sets quickly instead of having to catalog and update data sets individually. With a central source of truth like Funnel for your marketing data, this can be done with ease.
We hope this blogpost helps you assessing data reliability for your organization. And while it is not the sexiest topic for most people, it is important to invest in data reliability. When you unlock the power of your marketing data, it’s going to drive incredible results.