Advertising can feel like a guessing game. You create a campaign you think will resonate with your target market, send it off into the world and hope for the best. You can improve your odds with market research, but surveys and keyword trends can only tell you what happened in the past — and those trends change fast.
That leaves you back where you started: in the dark as to how your campaign might land with your audience and whether you’ll get the results you want. It’s no wonder 23% of businesses don’t have a clear market research strategy. Where do you even start?
At WebMarketers, we’ve cracked the code.
We’ve adopted askpolly, an innovative AI-based tool that takes the guesswork out of advertising by generating up-to-the-minute insights based on what real people are talking about online. So your campaign delivers what your customers are actually looking for, boosting engagement and ROI.
Contents
- Traditional Market Research Doesn’t Get The Clicks
- AI-Powered Market Research Delivers Real Results
- How Does askpolly Work
- askpolly in Action
- Looking Forward to The Future of AI in Advertising
1. Traditional Market Research Doesn’t Get The Clicks
Maximizing your marketing ROI means maximizing engagement. But how do you do that if you don’t know what people are struggling with or what they want? The answer is market research, but the traditional ways of doing that leave a lot to be desired.
Google trends and keyword research isn’t enough
While data from Google can give you some useful insights, that data is based on what people have searched over the last year and is updated only on a monthly basis. In other words, it won’t include what people were searching for last week or earlier today.
An estimated 15 percent of Google searches are brand new queries no one has ever asked before. With about 6.3 million Google searches happening every minute, that’s a lot of search terms that won’t be covered in the trend data.
The sheer volume of worldwide searches also means the data is often inaccurate, especially for smaller, more niche subject areas. Google’s keyword data might say a term you’re interested in has a search volume of zero, even though your own website analytics are showing dozens or hundreds of searches for that term every month.
For the internet as a whole, 100 searches is so insignificant that they’ll go uncaptured by Google. For a small business, it’s a totally different story. Even if only a small percentage of those 100 searches converts, that can represent a noticeable chunk of the month’s revenue.
Social media tracking data is constantly changing
Your social media profiles can show you which posts followers engaged with most or which ones were seen by non-followers. But the ever-shifting algorithms used by these platforms make it impossible to accurately compare data or develop effective strategies going forward. On top of that, social media is full of influencers, bots and fake accounts that can skew the data.
Official statistics are outdated by the time they’re published
Statistics Canada, Statista, international census departments and other formal sources of statistics can provide a range of industry-related data, but most of it isn’t specific or current enough to be of any real use.
Formal market research takes too long
Another option is to hire a market research firm (like Gartner, Forrester or IDC) to conduct custom surveys or focus groups. While this can deliver detailed, hyper-specific information, it’s expensive and takes time, and it relies on self-reported data. Participants know they’re participating, so they may not be completely forthright in their responses.
Ultimately, all these methods are open to interpretation. Every agency or person who looks at the data might draw different conclusions from it. And that means they don’t offer a really clear path forward to a strategy you know will work.
2. AI-Powered Market Research Delivers Real Results
In 2023, the Denturist Association of Ontario came to us looking to raise the profile of its members and encourage people to consider consulting a denturist. We learned from real conversations happening online that a lot of people with missing teeth had trouble eating their favourite foods. Others felt self-conscious about speaking or smiling around other people.
We created a video that spoke directly to these concerns, showing a person enjoying a dinner with friends thanks to the intervention of a denturist. Our analysis showed campaign results including:
- Almost 700,000 impressions on Google
- More than 4 million impressions on Meta
- A 265% increase in the number of people forecast to use a denturist in the future
Building on this success, we planned additional campaigns focusing on other areas of concern identified in our research, including playing hockey and snoring.
The key to getting these results was askpolly.
By looking at the unprompted conversations actual people are having in near-real time, askpolly can identify trends and determine what’s important to people right now. That means we’re not just guessing what our clients’ target audiences care about or are struggling with. We know what matters to them, so we can create messaging that resonates and encourages them to engage.
3. How Does askpolly Work
As a starting point, askpolly works similarly to tailored market research, with a set of questions you want to ask the target audience. For example:
- What’s your favourite vegetarian restaurant?
- How much would you pay for a particular type of product or service?
- Do you feel safe using your city’s public transit system?
From there, the process is quite different. When you run the program, it starts crawling the web and various social media platforms, using natural language processing to scan online discourse for talk that’s relevant to the questions it’s researching.
Correcting for bias
One of the hurdles to getting accurate results is the need to avoid bias in the sample. You don’t want your results skewed by bots, paid promoters or groups with agendas. askpolly corrects for this issue by assessing the linkages between people and collecting input only from independent (i.e., unrelated) people.
In doing so, askpolly eliminates:
- Bots, because most are connected to large networks of other bots and can be discounted
- Influencers, typically identifiable by their large followings, and their followers
- Groups of related people actively trying to raise the profile of something, like multiple employees of a company all posting about a new product
Delivering real insights
Once the data is collected, askpolly uses AI to analyze the findings and deliver usable insights, presented through easy-to-use visual dashboards featuring graphs, maps and more.
These insights can then be used not only to inform advertising campaigns for major brands, but also to give politicians, religious organizations or others a better understanding of what people are saying about them, their offerings or anything else relevant to their interests.
This approach is quite different from the most talked-about AI tools available today, which are based on generative AI.
GenAI (e.g., ChatGPT) uses a large language model to predict what words should come next based on how often those combinations of words appear in its data set. It doesn’t actually understand the relationships between the words or what the words even mean, so it can’t generate any real insight.
askpolly is a different form of AI. It’s a proprietary system designed specifically for this purpose, using classification to do sophisticated analysis on the data it collects, rather than just blindly putting words together because others have put them together previously. That enables it to draw new conclusions, producing new information that builds on, rather than just reproducing, the information fed into it.
4. askpolly in Action
Here are a few more examples of the kind of insights and results we’ve achieved for our clients thanks to askpolly:
Understanding what matters most for translation
Our client was a translation service that combined the efficiency of AI with the expertise of human translators. They wanted to promote the benefits of this offering and overcome any potential concerns potential clients might have about the use of machine translation. We used askpolly to gain a better understanding of those concerns so we could ensure our campaign addressed them.
We also identified regional differences in translation priorities. For example, the translation market in Ottawa is largely driven by government and therefore security and trustworthiness are critical. In Toronto, the market is focused more on large businesses, so cultural accuracy is prized.
We used these insights to build a campaign that addressed fears about using machine translation and highlighted the benefits of our client’s approach, tailored to each market.
Meeting audiences where they are
When the producers of a new deck of oracle cards wanted an assist to make sure they hit their crowdfunding goal, we needed to know more about the online media habits of their likely customers. Using askpolly, we identified several Facebook groups where members of the target audience were likely to be found. This let us be ultra-specific about where ads for the cards should be shown for the greatest impact.
And it worked. Within five days of our targeting efforts, the client’s pageviews were up 450,000% (yes, you read that right) and they had reached 110% of their crowdfunding goal. The hyper-precise targeting also drove a high conversion rate, so the whole campaign came in under budget.
5. Looking Forward to The Future of AI in Advertising
From Alan Turing’s seminal work in the 1950s all the way to recent developments like deep learning, neural nets and generative AI, artificial intelligence has come a long way. askpolly is one of the many forms of AI that exist today, enabling us to take bias out of the marketing equation, so we can build campaigns based on real evidence rather than just gut feeling.
We were thrilled to learn about this tool and jumped at the chance to become one of the first marketing agencies to use it. We’re blown away by what it’s enabled us to do — and we’re excited about the possibilities that AI will support going forward.
