Most social media firms struggle with assessing the market performance of their social media campaigns — often referred to as ROI. In other words, how do you assess your social media performance, which is the starting point for justifying your expense and improving performance?
I know I spent X on this social media marketing campaign, but what did I get back?
Over 70% of firms ask this question and only 15% of respondents in a Duke study felt they had a good handle on the ROI of their social media marketing. In another study reported in Forbes, 44% of businesses can’t assess their social media performance. As a result, CFOs and others are tightening the screws on digital managers in an attempt to force a results-oriented approach to social media marketing.
In an era when IT specialists and BI (business intelligence) analysts decry the nearly 2 zettabytes of data available (the equivalent of 47 YEARS of HD TV programming), why do we still wonder WHERE’S THE ROI?
Why assess social media performance
Establishing a consistent program for assessing your social media outcomes serves several purposes.
First, assessing the ROI of any action within an organization assures management and other stakeholders that their precious resources deliver results rather than represent waste. In the past, assessing the performance of your marketing campaigns was challenging because you didn’t have good metrics in the days when traditional marketing efforts dominated the landscape. For instance, when you run a TV or radio ad, it’s not clear how these efforts contributed to sales lift because a variety of factors in addition to your advertising might impact sales in both directions. With the increased use of social media and other channels, you now have tools to help you assess performance. For instance, you can track a click from social media to your website, even using tags to determine which post generated the click. Once on the website, you can track the click to determine whether it resulted in a conversion or other signals of purchase intention, such as accessing the local address of your physical store.
Second, you need to assess performance to determine which content delivers the highest results. Again, tags really help you assess which posts deliver and determine which social platforms produce the highest results. Using this information, you can work toward creating fresh content based on elements of this high-performing content.
Too many tools, too few insights
In part, the problem of assessing performance is a function of TOO MUCH data — everything from # of Facebook likes, to unstructured Tweets, to # of shares …
How do you make sense of this variety of data coming at you so fast that you can’t analyze it before more comes your way? The tendency is to try to assess everything, which isn’t possible with the velocity of data heading your way. Moreover, not all data you assess has the same impact on performance, so you face an apples and oranges problem. As an example, consider the conversion funnel shown below. Is attracting a new prospective customer more important than converting a visitor?
The answer is NO! The more prospective buyers you load into the funnel, the more flow out as conversions. You must develop and implement strategies for ALL stages in the funnel and metrics that assess these stages are different. Hence, you must develop a strategy to assess performance that considers the entire funnel.
Too many tools exist claiming to provide insights to help determine your ROI. For instance, I’ve spent a lot of time with IBM and their social media intelligence product, CoreMetrics. CoreMetrics is a great tool for monitoring visitors as they move through your website to buy (or not buy) your brand.
- Where did they come from?
- What did they do once they got to your website?
- What problems did they encounter trying to buy your products?
- Which social platforms brought a higher percentage of buyers? Which platform brought the biggest spenders to your site?
Very cool stuff and you DEFINITELY want to have this understanding. But is that all there is to ROI? What happens BEFORE folks get to your site? What creates the awareness, interest, and desire that drive consumer buying (we mean both end-users and business consumers)? Aren’t these critical precursors to buying (rhetorical question — we KNOW they are)? Plus, this tool does nothing to assess social media except as it relates to driving visitors to your website.
Plus, analytics tools are expensive, forcing you into a cost/ benefit analysis to guide the purchase of individual tools.
Down the rabbit hole
And these questions lead us right down the rabbit hole — drowning in information we really can’t process into business insights. Do we measure likes or shares, or …? How do these translate into buying intentions? What DO we pay attention to? Notice I’m very good at asking questions today and not giving you many answers (LOL). Let’s fix that.
The four-factor model to assess social media performance
I found a model that we might adapt to help make sense of all this STUFF — data without a clear analysis plan. Originally, the four-factor model of performance came from salesforce evaluation — where there are both input factors (such as # of sales calls and new leads) and output factors (like sales and AOS (average order size)).
The four-factor model of sales performance looks like this (according to the Tanner, et al. Sales Management Book):
Sales = Days worked X call rate X batting average (close rate) X AOS
Coolness. We now have a model that helps us understand how our actions translate to sales volume. Improving any of these factors should improve our sales.
So, now let’s think about how we might build a model — four factors or otherwise — that helps us assess social media performance! What factors do we include? I built metrics to assess each stage in the conversion funnel to get us started.
1. Amplification (awareness)
Well, let’s start by calculating amplification — the number of shares of our content since the more folks aware of our product, the more folks who might BUY our brand — I mean, face it, you can’t buy something if you don’t know it exists. In social media parlance, we assess amplification using several metrics; engagement that shares your content based on user behaviors such as commenting and impressions resulting when users see your content organically. UGC (user-generated content) involves users creating their own posts related to your brand and offering another amplification avenue. You can figure out your own way to combine these three metrics into a single amplification metric.
Maybe we weigh this raw amplification number by something related to the content posted/ shared. I did this for a client once and it worked out pretty well. For instance, you might weigh content you created a 1 and content created by others a 2 (you can even make it log-linear by using a larger number so that your content is a 2 and others’ content about you is 2X2).
We now have part of our four-factor model:
Sales = amplification (weighted or not)
2. Sentiment (interest)
We know folks won’t buy your brand if they think it’s crap (technical term there). So, we probably want some assessment of whether folks like your brand or not. Now, many tools simply use as much data as they can to score sentiment. I’m not crazy about this because you assume every positive (and every negative) is equal. But we know the statement I don’t like X brand is nowhere near as negative as X brand sucks! So why treat them as if the valence of the sentiment is meaningless. That’s a topic for another day.
Another problem with assessing sentiment is that most are based on NLP (natural language processing), which takes essentially unstructured data such as words and images to melt them down into a number or, more likely, sort them into positive, negative, and neutral. But, language isn’t that straightforward and, without context and non-verbal elements like inflection and facial expressions, it’s difficult to assess sentiment accurately.



For today’s discussion, let’s assume we have some measure of sentiment we feel pretty good about. Let’s add that to our model:
sales = amplification X sentiment
We’re making progress.
3. Marketing intensity
We know that it takes a certain number of exposures to a brand message before your market takes action or develops attitudes toward your brand. The exact number depends on which expert you follow, but most agree the number is 6-8 impressions. We’ll call this marketing intensity.
Let’s add stuff like how frequently we post on a social platform, the number of social platforms we use that reach the same market, our spending on digital advertising, the frequency of our online newsletter, plus our traditional media spend since folks don’t JUST live online. We’ll ignore the scale issue for these elements, but we’ll work that out so each element is equal (or weighted as we see fit).
sales = amplification X sentiment X marketing intensity
4. Close rate
Now we can add in our close rate. Using a tool like CoreMetrics or Google Analytics, we can calculate the percentage of folks who buy based on our site’s number of unique visitors. We can even fancy it up a little by separating new customers from return customers and/ or first-time visitors from repeat visitors.
sales = amplification X sentiment X marketing intensity X close rate
Feel free to add some additional elements you’d like to add to this 4-factor model in the comments below, but I think we’ve got a pretty great model here, both in simplicity and adherence to theory. The major disadvantage of the model is that some of the metrics are pretty challenging to collect.
Using the four-factor model to assess social media performance
Now that we have our 4-factor model, what do we do with it?
Well, first off, we can now compare our performance across various forms of content, publication times, and social networks to determine which content and networks provide the best performance? Since it takes time and effort (plus often some money) to create content and share it across different platforms, we should focus first on those actions with the highest returns.
We can experiment. The 4-factor model doesn’t include any weighting across the factors, but that’s probably not real. In practice, you may find that marketing intensity or amplification drive sales more than other factors. We can test changes in our strategy such as what happens when we spend more to increase the marketing intensity? Or what happens if we modify the landing page or conversion process to increase the close rate (taking a single click out of the sales process dramatically increases sales)? Or maybe we increase amplification by creating some content with killer headlines and images. What does that do to sales?
With these experiments complete, we now know WHERE and HOW to focus our social media budget to create the greatest impact on sales.
Need marketing help to support business growth?
We welcome the opportunity to show you how we can make your marketing SIZZLE with our data-driven, results-oriented marketing strategies. Sign up for our FREE newsletter, get our FREE guide to creating an awesome website, or contact us for more information on hiring us.
Hausman and Associates, the publisher of MKT Maven, is a full-service marketing agency operating at the intersection of marketing and digital media. Check out our full range of services.
[…] Using regression to build a predictive model is probably the best way to understand complex buying situations and support the ROI of digital marketing content. Alternatively, you can use an econometric model for similar types of situations rather than building the model from scratch. I’ve built a predictive model for understanding the overall contributions of your digital marketing to the bottom line. […]