Ciri asks, “What’s the best way to measure brand awareness? Among all the searching I’ve done, it seems like people have a lot of different takes on measuring brand awareness.”
The gold standard for how to measure brand awareness is unaided recall among your target audience, which typically requires a significant (five to six figures) but worthwhile investment in a market research firm like Edison Research. That said, you can begin to sort out brand awareness from a series of digital metrics that can inform your market research. The metrics to consider are:
- Branded organic search, and the metrics it generates
- Coverage, and the metrics it generates
- Conversation, and the metrics it generates
- Downfunnel metrics and business outcomes
These three categories will generate something on the order of several hundred variables to process, which requires some data science techniques to properly process.
- Ingestion and cleaning
- Centering and scaling
- Variable importance measurement
- Outcome modeling
- Intermediary KPI modeling
Can't see anything? Watch it on YouTube here.
Listen to the audio here:
- Got a question for You Ask, I'll Answer? Submit it here!
- Subscribe to my weekly newsletter for more useful marketing tips.
- Find older episodes of You Ask, I Answer on my YouTube channel.
- Need help with your company's data and analytics? Let me know!
What follows is an AI-generated transcript. The transcript may contain errors and is not a substitute for watching the video.
In today's episode, Ciri asks, What's the best way to measure brand awareness? Among all the searching I've done? That seems like people have a lot of different takes on measuring brand awareness. Absolutely. People have tons of different takes on measuring brand awareness. A lot of those things being done at and put out there are by individual vendors promoting their software. There is there are a couple of different ways to measure brand awareness. The gold standard for measuring brand awareness is unaided recall surveying among your target audience. So if your target audience is Chief Technology officers, you would commission a market research firm to check in with them once a quarter or whatever and say hey
in your experience, please name five vendors who provide I don't know email marketing services and see what these people remember. unprompted unaided. See if they recall Hall your brand and they do great, you're you have brand awareness within your target segment. If it never name your brand at all. Is it okay, well, we're not reaching the target audience. Now, market research. Proper market research requires significant investment. The firm I recommend typically is a company called Edison research. They do top shelf market research. And they do it properly.
There are a lot of companies to call themselves market research companies. But
market research like that should be the my friend Tom Webster who works at Edison research calls it reassuringly expensive,
you should be planning on your mid five figures to low six figures for the budget to do something like that. Because you're you're going to want to check in with these people frequently. And you need somebody if your audiences like senior executives, you're going to need credible market research companies. To get to those executives. You can't just spin up a survey monkey and email them and you'll get like a completely statistically insignificant response, right.
the second way, which is a precursor to the market research is to use some of your digital metrics to start to assemble a score that will inform your market research, it is not a replacement for market research, it is a a prerequisite of the market research to make sure that you're doing the things you need to do in order to get people in the door. And
that digital metric
combination really comes out of four buckets. bucket number one is a branded organic search the number of people who search for you by name over time, and then the conversion metrics that go along with that. So the brand number of brand organic searches, returning users to brand organic searches, etc, etc, etc. That's going to be you know, 2030 variables
if it was a spreadsheet, like 20 columns coverage, so public relations, media relations, influencer relations, whose
writing about you, what are they writing? What's this sentiment? What's the tone? What's the importance? What's the SEO value, what are the number of clicks on articles, social shares, all those metrics around coverage are a second big bucket that's you're gonna you're talking
potentially another spreadsheet of 50 or 60 columns. The third is conversation, people talking about you, and this could be influencers, but it could also be regular people, your target audience, and then all the subsequent metrics that those generate likes, comments, shares,
pro profile clicks, all the works, that's going to be a gigantic spreadsheet. And finally, in the fourth bucket, you're going to need off your down funnel metrics. So you have your awareness sort of top of funnel, then you have web traffic, new users, returning users time on page by segment, goal conversions, and then you get out of web analytics. You go into marketing, online system, your
marketing, qualified leads, sales, qualified leads, opportunities, deals, one deals, loft, etc.
You're going to need to put together the spreadsheet and the spreadsheets going to have
probably several hundred columns,
you will need data science
techniques to properly process this data.
There is no there's no human way to do this, at least not in anything that would take you less than two years are you doing nothing but that
because it is a massive undertaking
the it's a five step process you need to do ingestion and cleaning mean take all the data info the sources, clean it up, fixed, missing, or broken data, remove anomalies, and so on, and so forth. So that's step one. Step two is what's called centering and scaling where you need to normalize the data so that you can do apples to apples comparisons a little more cleanly. For example, if you are looking at branded organic search, and you're looking at social conversation, this is going to be have very different scales. So it's very difficult to do a comparison of those metrics without normalizing them scaling sent to them, scale them, make them more like each other. The third step is doing what's called variable importance identification. And that is that a lot of cases that's going to take actual machine learning to run
every possible combination of those variables against a, a, an outcome, a target, like
and figure out which metrics in combination have a high correlation to the
actual outcome you care about.
We know that, you know, there's, there's a sequence within the funnel, people don't necessarily, you know, they don't follow linearly, but they there is a path from awareness to purchase, people don't make a purchase without awareness. That's a, that's a logical. So the variable importance measurement helps you identify the variables, a mathematically high relationship,
once you've done that you've gotten rid of, you know, 80, 90%
of the variables that don't have any mathematical relationship to the outcome you care about, you'll want to build a couple of models, you're going to build an outcome model which says, Hey, we, if we want more sales, we need to test doing more of these things. And then you'll go back and rerun variable importance to do what's called intermediary KPI modeling.
And this is especially for bigger companies.
There are a lot of dependencies on a sale
problem. I used to have it at a company just to work with was that marketing kept being asked for more and more and more leads every quarter more leads, more leads, more leads, and sales was closing at something like a 1% closing rate. So benchmarking off of sales, as the only outcome meant that a lot of marketing data got thrown out. Because the salespeople were incompetent. They they couldn't have sold fire to a freezing person, and
so intimidated KPI modeling says, okay, for your department, what outcome Do you have responsibility for if you work in corporate communications awareness, maybe the variable to measure for if you're the web guy or the web girl, you know, new traffic to the website is your KPI. And so you'll want to rerun that variable importance for each departmental outcome so that each department understands, hey, these are the things that
we know contribute to the outcome that we care about. And again, build models for that. And then the last step of the process is, once you've got these models, you have to test them. If, for example, tweets on Tuesdays, that contain a poop emoji have the highest mathematical correlation to the outcome you care about. You cannot assume that correlation equals causality, you have to build a testing plan to say, Okay, now let's do five more tweets on Tuesdays and put three poop emoji and the tweeting instead of two and see if that commensurate increase in activity
yields to the command a commensurate increase a proportional increase in outcome. And so there's that testing plan to bring to life that model and and validate that the model either works or does not work it this is the scientific method, just using a lot more math and data, you come up with a hypothesis, you test it, you analyze it, you find your hypothesis, until you're you have a proven model. And that's how you develop a working model, a working measurement model for brand awareness. You can't just throw a bunch of numbers on a spreadsheet, average them and add them all up and call it brand awareness. Because you don't actually know what does and does not contribute. You have to go through this process of testing. And you need to use data science and machine learning if you want the model to be credible and proven and and develop a testing plan that is workable because again, if you've got a spreadsheet with 500 variables, testing each one and then testing each combination of one you have run out a lifetime before you you get through you. One testing machine has to help you do it. So great question is a complex question and it requires data science help. It's not something that you can build a credible model for by yourself just with a spreadsheet. If you have follow up questions, please leave them in the comments. And of course, subscribe to the YouTube channel newsletter
and I'll talk to you soon. Take care
want help solving your company's data analytics and digital marketing problems.
This is trust insights.ai today and let us know how we can help you
You might also enjoy:
- 2020 Rewind: Artificial Intelligence in Sales
- You Ask, I Answer: Creating Content for Search Engines?
- Best Practices for Public Speaking Pages
- How To Set Your Consulting Billing Rates and Fees
- Almost Timely: The 2020 Essays
Want to read more like this from Christopher Penn? Get updates here:
Get your copy of AI For Marketers