The New Facebook Insights — One More Analyst’s Take

By on in , with 19 Comments

Facebook released its latest version of Facebook Insights last week, and that’s kicked off a slew of chatter and posts about the newly available metrics. Count this as another one of those. It’s partly an effort to visually represent the new metrics (which highlights some of the subtleties that are a little unpleasant, although, in the end, not a big deal), and it’s partly an effort to push back against the holy-shit-Facebook-has-new-metrics-so-I’m-going-to-combine-the-new-ones-and-say-we’ve-now-achieved-measurement-nirvana-without-putting-some-rigorous-thought-into-it posts (not linked to here, because I don’t really want to pick a fight).

Basically…We’re Moving in a Good Direction!

At the core of the release is a shift away from “Likes” and “Impressions” and more to “exposed and engaged people.” There are now a slew of metrics available at both the page level and the individual post level that are “unique people” counts. That…is very fine indeed! It’s progress!

Visually Explaining the New Metrics

As I sifted through the new Facebook Page Insights product guide (kudos to Facebook for upping the quality of their documentation over the past year!) with some co-workers, it occurred to me that a visual representation of some of the new terms might be useful. I settled on a Venn diagram format, with one diagram for the main page-level metrics and one for the main post-level metrics.

Starting with page-level metrics:

Defining the different metrics — heavily cribbed from the Facebook documentation:

  • Page Likes — The number of unique people who have liked the page; this metric is publicly available (and always has been) on any brand’s Facebook page.
  • Total Reach — The number of unique people who have seen any content associated with a brand’s page. They don’t have to like the page for this, as they can see content from the page show up in their ticker or feed because one of their friends “talked about it” (see below).
  • People Talking About This — The number of unique people who have created a story about a page. Creating a story includes any action that generates a News Feed or Ticker post (i.e. shares, comments, Likes, answered questions, tagged the page in a post/photo/video). This number is publicly available (it’s the “unique people who have talked about this page in the last 7 days”) on any brand’s Facebook page.
  • Consumers — The number of unique people who clicked on any of your content without generating a story.

A couple of things to note here that are a little odd (and likely to be largely inconsequential), but which are based on a strict reading of the Facebook documentation:

  • A person can be counted in the Total Reach metric without being counted in the Page Likes metric (this one isn’t actually odd — it’s just important to recognize)
  • A person can be counted as Talking About This without being included in the Reach metric. As I understand it, if I tag a page in a status update or photo, I will be counted as “talking about” the page, and I can do that without being a fan of the page and without having been reached by any of the page’s content. In practice, this is probably pretty rare (or rare enough that it’s noise).
  • Consumers can also be counted as People Talking About This (the documentation is a little murky on this, but I’ve read it a dozen times: “The number of people who clicked on any of your content without generating a story.” Someone could certainly click on content — view a photo, say — and then move on about their business, which would absolutely make them a Consumer who did not Talk About the page. But, a person could also click on a photo and view it…and then like it (or share it, or comment on the page, etc.), in which case it appears they would be both a Consumer and a Person Talking About This.
  • A person cannot be Consumer without also being Reached…but they can be a Consumer without being a Page Like.

Okay, so that’s page-level metrics. Let’s look at a similar diagram for post-level metrics:

It’s a little simpler, because there isn’t the “overall Likes” concept (well…there is…but that’s just a subset of Talking About, so it’s conceptually a very, very different animal than the Page Likes metric).

Let’s run through the definitions:

  • Reach — The number of unique people who have seen the post
  • Talking About — The number of unique people who have created a story about the post by sharing, commenting, or liking it; this is publicly available for any post, as Facebook now shows total comments, total likes, and total shares for each post, and Talking About is simply the sum of those three numbers
  • Engaged Users — The number of unique people who clicked on anything in the post, regardless of whether it was a story-generating click

And, there is a separate metric called Virality which is a simple combination of two of the metrics above:

That’s not a bad metric at all, as it’s a measure of, for all the people who were exposed to the post, what percent of them actively engaged with it to the point that their interaction “generated a story.”

The Reach and Talking About metrics are direct parallels of each other between the page-level metrics and the post-level metrics. However (again, based on a close reading of the limited documentation), Consumers (page-level) and Engaged Users (post-level) are not analogous. At the post-level, Talking About is a subset of Engaged Users. It would have made sense, in my mind, if, at the page-level Talking About was a pure subset of Consumers…but that does not appear to be the case.

KPIs That I Think Will Likely “Matter” for a Brand

There have been several posts that have jumped on the new metrics and proposed that we can now measure “engagement” by dividing People Talking About by Page Likes. The nice thing about that is you can go to all of your competitors’ pages and get a snapshot of that metric, so it’s handy to benchmark against. I don’t think that’s a sufficiently good reason to recommend as an approach (but I’ll get back to it — stick with me to the end of this post!).

Below are what I think are some metrics that should be seriously considered (this is coming out of some internal discussion at my day job, but it isn’t by any means a full, company-approved recommendation at this point).

We’ll start with the easy one:

This is a metric that is directly available from Facebook Insights. It’s a drastic improvement over the old Active Users metric, but, essentially, that’s what it’s replacing. If you want to know how many unique people are receiving any sort of message spawned from your Facebook page, Total Reach is a pretty good crack at it. Oh, and, if you look on page 176 of John Lovett’s Social Media Metrics Secrets book…you’ll see Reach is one of his recommended KPIs for an objective of “gaining exposure” (I don’t quite follow his pseudo-formula for Reach, but maybe he’ll explain it to me one of these days and tell me if I’m putting erroneous words in his mouth by seeing the new Facebook measure as being a good match for his recommended Reach KPI).

Another possible social media objective that John proposes is “fostering dialogue,” and one of his recommended KPIs for that is “Audience Engagement.” Adhering pretty closely to his formula there, we can now get at that measure for a Facebook page:

Now, I’m calling it Page Virality because, if you look up earlier in this post, you’ll see that Facebook has already defined a post-level metric called Virality that is this exact formula using the post-level metrics. The two are tightly, tightly related. If you increase your post Virality starting tomorrow by publishing more “engage-able” posts (posts that people who see it are more like to like, comment, or share), then your Page Virality will increase.

There’s a subtle (but important this time) reason for using Total Reach in the denominator rather than Page Likes. If you have a huge fan base, but you’ve done a poor job of engaging with those fans in the past, your EdgeRank is likely going to be pretty low on new posts in the near term, which means your Reach-to-Likes ratio is going to be low (keep reading…we’ll get to that). To measure the engage-ability of a post, you should only count against the number of people who saw the post (which is why Facebook got the Virality measure right), and the same holds true for the page.

Key Point: Page Virality can be impacted in the short-term; it’s a “speedboat measure” in that it is highly responsive to actions a brand takes with the content they publish

This is all a setup for another measure that I think is likely important (but which doesn’t have a reference in John’s book — it’s a pretty Facebook-centric measure, though, so I’m going to tell myself that’s okay):

I’m not in love with the name for this (feel free to recommend alternatives!). This metric is a measure (or a very, very close approximation — see the messy Venn diagram at the start of this post) of what percent of your “Facebook house list” (the people who like your page) are actually receiving messages from you when you post a status update. If this number is low, you’ve probably been doing a lousy job of providing engaging content in the past, and your EdgeRank is low for new posts.

Key Point: Reach Penetration will change more sluggishly than Page Virality; it’s an “aircraft carrier measure” in that it requires a series of more engaging posts to meaningfully impact it

(I should probably admit here that this is all in theory. It’s going to take some time to really see if things play out this way).

Those are the core metrics I like when it comes to gaining exposure and fostering dialogue. But, there’s one other slick little nuance…

Talking About / Page Likes

Remember Talking About / Page Likes? That’s the metric that is, effectively, publicly available (as a point in time) for any Facebook page. That makes it appealing. Well, two of the metrics I proposed above are, really, just deconstructing that metric:

This is tangentially reminiscent of doing a DuPont Analysis when breaking down a company’s ROE. In theory, two pages could have identical “Talking About / Page Likes” values…with two very fundamentally different drivers going on behind the scenes. One page could be reaching only a small percentage of its total fans (due to poor historical engagement), but has recently started publishing much more engaging content. The other page could have historically engaged pretty well (leading to higher reach penetration), but, of late, has slacked off (low page virality). Cool, huh?

What do you think? Off my rocker, or well-reasoned (if verbose)?

Similar Posts:

19 Comments


  1. Nice post, Tim. It got me thinking about work stuff (and posting a reply!) on a Friday night, which is about the highest compliment I can give you.

    I like Talking About / Page Likes a lot more than you do, and I think it can be a useful benchmark. I think the key will be to look at that number over time. It should be done weekly, because the public Talking About metric is the 7 day number.

    If a page’s Talking About % is consistently high, then they are consistently delivering engaging content – which is likely driving both their reach penetration and page virality. And if that’s the case, do I really need to know the exact breakdown between Reach Penetration and Page Virality? If they show volatility in their Talking About %, then I can be pretty sure that their variance is being driven by variation in their page virality. After all, as you pointed out Reach Penetration is an aircraft carrier metric.

    So it seems to me that looking at Talking About % over time will help me determine which pages are delivering engaging content, and may also help pinpoint content or campaigns that are delivering abnormal engagement.

  2. Wow. And I thought *I* was lame for writing the damn thing on a Friday night!

    The more I think about it, the more I feel like it’s going to be a “time will tell” metric. I am curious as to: 1) how noisy that metric (and the ones I proposed, for that matter) turns out to be, and 2) whether brands will be able to legitimately and intentionally influence the movement of the metric. At least for the pages for which we have Facebook Insights data, we’ll be able to compare the “public” metric to the “underlying component” metrics and see what happens. I tmay be that we can use the “Talking About / Likes” as a KPI, and then use the componentization of it behind the scenes to get a better sense of what needs to happen to improve it (maybe…although I think only one of the components I propose is directly actionable).

    My concern is that there have been social media metrics proposed in the past that are unduly swayed one way or another by fairly innocuous events, which makes them hard to use as a performance measurement tool. It may be that only time will tell if that’s the case here.

  3. Pingback Facebook Measurement: Impressions from Status Updates | Gilligan on Data by Tim Wilson

  4. Great post Tom. Thanks for sharing your thoughts about this.
    Instead of “Reach penetration”, maybe we can call it “Reach rate” or “Activation rate” ?

    I agree with you. I’m not sure if the “Page talking about this” can be used as an insightful benchmark metric either, considering total fans is a lousy indication of how many unique users each page reach.

    Amer as a point though. Maybe you can see trends overtime anyway.

  5. I like the new stats but I wish they hadn’t just wiped out the old ones. I’d like to have some continuity in what I report, but some of the old ones don’t correspond at all to any of the new ones.

  6. @Steve I like those suggestions. It’s been interesting to noodle around with the data, as well as watch as some additional measures have gotten rolled out. I’ll need to do an update to this post and a follow-up one of these days, I suspect.

    @Chris You’re right. I’ve actually been meaning to write a new post. Interestingly, although the wording is slightly different “Reach” and “Active Users” really seem to be the same thing. However, when we’ve done comparisons of those two metrics during the overlapping period, they don’t match up, and they’re off as much as 10%. I assume that is because they improved their method for calculating the metric, and Reach is “more accurate,” but it does make for some difficulties in trending.

  7. Pingback Understanding Facebook Insights Terminology Redux | Gilligan on Data by Tim Wilson

  8. Pingback Facebook Reach - What You Need to Know | Social Amateur

  9. Hi! Thanks for the article, I found it while researching the new Insights for our small business page. I have a very simple question I can’t seem
    to find an answer to (hence why we are fabulous florists but very amateur social media users 🙂 haha) –

    Why if we have approx 700 likers do our posts only reach
    For example 142 of these?

    Thanks anyway for a great read. Cheers from Australia!

  10. @Sally — What you are seeing is typical, unfortunately. This is a symptom of two things: 1) Some of your fans, presumably, are not logging into Facebook with high enough frequency for all of your posts to be recent enough for them to see them, and 2) Facebook’s EdgeRank/GraphRank algorithm determines that certain of your fans have other content they would rather see. The latter is the real bugaboo, and what you’re seeing is in line with a study that AllFacebook.com undertook a few weeks ago. The average they saw across 4,000 pages was 17%, whereas you are seeing 20%.

    You might want to do an export of the Page-Level data and look at the Weekly Reach and 28-Day Reach for your page — you’re probably reaching more of your fans than 20% over the course of a month (you’re just not reaching them with ever post). From that same export, you can also see, of all the people you reach, how often you are reaching them. I hope this helps.

  11. Great article, Gilligan. This has really helped me understand more about how to analyze my social media metrics.

    Quick question— what tool did you use to make these circle diagrams?

  12. Thanks, Daniel! I wish I could say it was some sort of fancy diagramming tool…but it was just PowerPoint — the blessing and curse of many a working professional.

  13. Hi Tim,
    If I am not wrong, the Total Reach here is the ‘potential total reach’. If yes then is there a way to find out ‘most likely reach’ for the page during given period of time?
    Thanks much.
    Alakh

  14. Total Reach is actually looking backwards and is the “actual reach” of the page (for the past day, the past week, or the past 28 days). We typically use that historical data as the basis for projecting forward (which is where Facebook’s sporadic algorithm changes are a bear). Things get super-messy when Facebook paid media is involved, as the Paid Reach almost always dwarfs the Organic Reach and Viral Reach. It’s a messy answer. But, if we keep posting frequency and content relatively constant, and we’re not running campaigns, then the future reach remains fairly consistent.

  15. Pingback Novas métricas do Facebook estão mudando o jogo para os marketers | Especialistas em SEO - Curitiba

  16. Tim,

    Is there a way to capture the reach of a post over time? For example, a new post will have 100 likes on the first day, 125 on the second, 150 on the third day, and so on. Another post may, for purpose of comparison, have 50 likes the first day, 75 the second, etc.

    It seems that a FB admin can differentiate fairly early on how quickly a post is increasing in reach. There’s probably a fairly standard curve that any post follows, but a more effective post should stand out with perhaps hours of its posting. Some posts may extend theiir reach over a significantly longer time period and that would be useful to know as well.

    I would find it useful to be able to graph posts and identify those outliers that are either overperforming or underperforming.in terms of “reach/time.”

    This blog article is excellent and thanks for sharing your knowledge and insights.

  17. Hi Michael,

    That’s a good question. I don’t know of a way to get that information from Facebook Insights. I suspect it’s available through the graph API in some way, but that might take some effort — pulling the post ID and then pulling each action taken on it (and I’m not sure if that is actually doable for likes).

    In my experience, it’s a rare post that has an organic life of more than a couple of days, FWIW.

    Tim

  18. Tim, thanks for your feedback. Good point you make about the lifespan of a post – by the time we would get around to analyzing it and trying to react, its organic life would have already “passed.”

Leave your Comment


« »