How Do I Ensure Good Data from My Polls?


The Helium 10 software, Audience, makes polling groups of Amazon sellers easier than ever. But, to get the best data, it’s important to:

      • ask questions in a straight, unbiased way 
      • carefully design your poll using professional copy and images
      • select appropriate respondents for your specific purpose

This article discusses best practices and related polling challenges that might lead to inconclusive results. It covers how to write your questions, design your images, and filter for the most relevant audience.

1. Keep your question simple, focused and unbiased. 

Don’t pose a question that is asking more than one question. These multi-layered questions are called double-barreled questions, and they generally lead to inconsistent or confused responses. It might seem like you can gather data for two or three questions in one poll, but often, respondents will focus on one aspect, and you’ll end up with incomplete answers that make it difficult to interpret the data. 

For example, imagine you asked 50 Prime Amazon shoppers the following question paired with two pictures of a toy pony in its branded packaging:

Would you buy hand-carved wooden ponies for a toddler’s birthday, from either a company named Bob’s Wild Creatures or a company named Bob’s Organic Animals? 

Respondents would be unsure what you are asking exactly. Some possible interpretations include:

      • Of the two images, which product looks more appealing?
      • Would they buy either product?
      • Would they buy a hand-carved, wooden pony?
      • Which company name seems more appealing?
      • Which company name do they prefer specifically for this type of product?
      • Do they think a pony would be an appropriate birthday gift?
      • Do they think a pony would be an appropriate gift for a toddler?

Some respondents might focus on the product (saying they wouldn’t associate a pony with “creatures”) while others might focus on a word like “organic” (saying they wouldn't associate “wood” with organic), while yet another might consider the gift inappropriate for a toddler, and/or for a birthday gift. Another might like the name Bob’s Wild Animals but doesn’t consider a pony to be a wild animal. 

In the end, you’ll get a wide variety of responses, but you won’t necessarily get the useful data you need to create an optimal listing. Keep the question and task simple when creating a poll. If you want to get answers to all of the embedded questions, create separate polls with individual questions targeting each component, and then run separate polls.

If you do want to include a slightly more detailed question, try italicizing or bolding the most important words to emphasize their priority. For example, you could ask:

Do you prefer the brand name “Bob’s Wild Creatures” or “Bob’s Organic Animals” for a hand-carved wooden pony?


Which toy pony would you prefer for a toddler?

Using unbiased language is also important in polling. If questions are leading--meaning that they impose hidden assumptions or unconsciously lead respondents to answer in a specific way--the polling data will be compromised. 

For example, if you posed the following revised question about toy ponies to an audience of 25-35 year-old, environmentally-conscious parents, you’d have spent money on a poll that gives predictable, less useful, results.

Would you prefer to buy a hand-carved, environmentally-friendly, hemp pony or a machine-manufactured, mass-produced plastic pony?

This question uses leading words like “environmentally-friendly” and “mass-produced” that will subjectively sway the respondents’ positive or negative feelings about the product, even before they look at the accompanying images.

Because Audience’s poll results include respondent feedback as well as their choice of the options presented, it is better to keep your question simple and free from bias. You will get respondents’ reasoning and feedback in their comments. 

2. Keep your images consistent in size, content and quality.

If you pair the above question with images of two different sample toy ponies having inconsistent image quality--one Amazon-compliant, clear and large, the other slightly out of focus, smaller and non-Amazon compliant--you will have created a problematic poll as well. You won’t know if respondents preferred one image because it was larger and easier to see or whether they preferred it because it contains a more appealing-looking product.

Similar to the wording of your questions, you’ll want to keep the differences between the images simple as well. If your poll is testing an image for a brand name, logo, and slogan, for example, don’t show two completely different graphics, names, and slogans. Choose only one element to change, per poll. If you are testing the brand name, the brand logo, and the brand slogan, start with a poll in which the image and slogan are identical and only the brand name is different. Then using the winning brand name, run a poll testing only the logo, then, run a poll with the winning brand name and logo, testing only the slogan. Even though it will take longer and require three separate polls, in the end, you will be confident that the entire package of brand name, logo and slogan are ready for marketing.

3. Identify your ideal shopper and set the conditions of the poll to target those individuals.

As you develop your brand and its products, you probably have a good idea of who your ideal customers will be. If you sell women’s sundresses, you know most of your customers will be women. If you sell facial shaving cream, you know most of your customers will be men. If you sell toddler toys, like the pony in our example above, most of your customers will be parents and grandparents. If you sell gardening tools, you know most of your customers will be gardeners, or aspiring gardeners. Generally, you want to target the people who are most likely to buy your product in order to collect the most relevant data. 

It might seem easier to just open the poll to all Amazon Prime members, but you’ll end up with responses from people who may have no special interest in buying your product. Before you choose your poll audience, think about who you want to appeal to in terms of both your brand and your product. Consider:

      • Gender
      • Age Range
      • Relationship Status
      • Lifestyle
      • Income level
      • Education level
      • Values
      • Hobbies and Interests

Audience lets you refine your targeted respondents to a very specific group of respondents, increasing the value of the data. Keep in mind, the addition of each trait does increase the cost of the poll, as well as the length of time it takes for the requested number of responses to be collected. 

For example, limiting respondents to men between the ages of 45-55, with college degrees, incomes above $100K, who love fishing and, therefore, might be interested in a high tech $800.00 underwater sonar detector, will require more time for responses. But it’s unlikely that someone who has no interest in fishing is going to care about an expensive piece of fishing equipment. 

While the cost of narrowing the pool of respondents may seem higher, in the end, it will provide you with critical listing data that increases sales and makes your product directly appealing to interested shoppers. The type of question will dictate the need for more or less specific audiences.

In summary, keeping your poll simple, both in terms of the questions you ask and the images you show, but cultivating a more selective, specific audience, will result in gathering the best data.


Create a Free Helium 10 Account Today!

 Start with the free extension: Download the Helium 10 Chrome Extension.

Was this article helpful?
5 out of 6 found this helpful



Article is closed for comments.