FAQs
How do I know if my results are statistically significant?
Statistical significance testing is not compatible with MRP (bayesian models), so we are not able to currently offer this feature.
However, instead, we provide you with margins of error readings - this is the compatible model to identify error bounds within the data. Using MoE readings (which are provided directly in the dashboard on every data point, including segments), you are able to identify the accuracy and confidence levels in the dataset and provide reliability in the data readings across waves and by segments.
Q. What is the difference between campaign impact measurement and brand health tracking?
Campaign impact measurement and brand health tracking are two different types of tracking.
Whilst your brand tracker is designed to track more KPIs over a longer period of time to enable you to see directional trends, with campaign impact tracking, we are primarily interested in measuring the uplift in awareness, the simplicity in the way we structure our surveys is to ensure reliable data wave over wave.
Q. Is Google Trends data comparable with Latana’s results?
In short, the answer is no. Google Trends data source and Latana data source are not comparable primarily due to the underlying audiences they both capture. Google Trends is measuring active interest within a market, while Latana is assessing brand health within the category of content uploaders:
Google Trends data is essentially a combination of Unaided Awareness + Brand Consideration - it is detecting unprompted interest in your brand.
This is a very different indicator of consumer drive to purchase, due to the fact people only actively search for a brand that they intend to either research or actually buy from. Ultimately, anyone who searches for a brand online already knows that brand, and furthermore has an active and unprompted interest in it, which limits a wider understanding of brand share and potential growth opportunities.
At Latana we are assessing your brand’s wider brand health (such as awareness levels and perception) within a population that is more representative of the category itself (screened for content uploaders only).
There is no pre-disposition to knowing the brand or having any former interest in it when giving an opinion on it. Here we are trying to capture: how many people within the category know your brand, how many would consider your brand, what do people think of your brand and measure this growth over time.
What is the difference of incidence rate (IR) between country/region/city?
Generally, IR's will vary greatly by region, but as you can imagine the general rule is that the IR is highest by country level, and becomes narrower the smaller the region you wish to target. E.g: USA = Whole country (population is 328.2m = 100%) USA, Northeast = Region (population is 55.9m = 17%) USA, New York City = City (population is 8.42m = 2.6%)
Can I have a sample size of less than n=500?
A sample size lower than n=500 would not provide reliable and representative data and will be at risk of high fluctuations across waves, therefore we consider a minimum sample size of n=500 per wave to be a robust sample size to ensure a reliable reading of data.
However, this is also dependent on the level of segmentation required within the data.
If only standard measures of age, gender, education, region and income are required for analysis of the data (no custom segmentations), then n=500 per wave is the minimum we accept. For any custom segmentation requirements, we will require a wider sample size to accommodate for this and ensure reliability of the segmented data.