Why Pub Engagement Scores Are the Most Misleading Metric in Digital Health
By Tom Fairhall | Pub Post Pro | April 2026
There is a number in the 884-pub dataset that looks straightforward and is not. The average engagement score across the entire sample is 3.02 out of 15, roughly 20% of the maximum. Scan the data and you might conclude that this is simply the state of things: pubs do not post, pubs do not engage, end of story.
The trouble is that 3.02 is not primarily a measure of what pubs are doing on social media. It is, to a significant degree, a measure of what the automated scoring system could reliably detect.
This distinction matters enormously for anyone trying to understand the real state of pub digital health, and for any brewery marketing director trying to identify which pubs in their estate are actually worth examining closely, and which ones are genuinely invisible online.
This article explains what the engagement and content data is actually measuring, what it is missing, and why the gap between those two things is the most important opportunity in pub digital marketing right now.
What a Default Score of 3 Actually Means
When an automated audit scores a pub's engagement at 3 out of 15, it is not necessarily saying the pub posted nothing. It is saying the system could not determine with confidence that the pub had posted anything meaningful within the detection window.
Automated social media auditing works by checking for publicly visible signals: recent posts, follower activity, post frequency indicators, and platform metadata. When a Facebook page exists but its posting history is not cleanly accessible to an audit tool, or when posts are visible to followers but not indexable at scale, the fallback score is a default.
That default in this dataset is 3. The same 3 that scores a pub with no posts in six months also scores a pub that posts three times a week but whose account settings limit public visibility of its timeline.
The result is that 3 becomes the most common engagement score in the dataset. Not because most pubs are doing the same thing, but because automated detection compresses a wide range of actual activity into a single undifferentiated number.
The content pillar tells the same story. The average across 884 pubs is 3.07 out of 15, nearly identical to engagement. Content scoring in automated audits typically looks for evidence of blog posts, regular news updates, or structured written content on a website or social feed. Most pubs do not have a blog. Most pub websites are static.
So the content score defaults to 3 across the board, regardless of whether the landlord is writing Facebook posts every day or has not touched the account in four years.
What the Standard Deviation Shows About Real Variance
The within-estate variance data is one of the most revealing pieces of analysis in the dataset, and it becomes even more revealing when you apply the measurement caveat to it.
Harvey's Brewery has a standard deviation of 14.2, the highest in the 23-estate sample by a significant margin. Its nearest competitor, Arkell's, is at 9.8. The industry average is around 5.8.
Harvey's variance exists because some of its landlords are clearly running their own digital marketing, independently, actively, and effectively. The Royal Oak in Southwark scores 76, the single highest score in 884 pubs. Seven Harvey's pubs break the 60-point Band B threshold that no other pub in the entire dataset reaches. At the same time, 20 Harvey's pubs are in Band D, scoring 20-39.
The extreme spread at Harvey's is not primarily a product of the engagement scoring being more accurate there. It is a product of the other pillars (reputation, volume, presence, website) reflecting genuine variance in how individual landlords have built their digital infrastructure.
The engagement and content scores at Harvey's (3.50 and 4.45 respectively) are above the industry average, which suggests that the Harvey's pubs scoring higher have at least some additional engagement signal being picked up.
But for the bulk of the estate data, the 93.2% of pubs clustered in Band C, the engagement and content scores are telling you almost nothing about the actual differences between pubs. A pub scoring 3 on engagement could be dormant or could be posting five times a week. The score does not distinguish between them.
The Implications for Band C
The Band C ceiling of 59 is the most commonly cited finding in the dataset. 93.2% of pubs score between 40 and 59. Zero pubs score above 84. The data describes an industry stuck just below the Band B threshold.
But how much of that ceiling is real, and how much is an artefact of measurement?
Consider the arithmetic. Under the 7-pillar scoring system, a pub that achieves perfect scores on reputation (15), volume (13), presence (13), and website (12), all fully measurable pillars, accumulates 53 points before the engagement and content pillars are even counted. Add the standard defaults of 3 on engagement and 3 on content, and you reach 59.
That is the precise score where the industry ceiling sits.
In other words: the 59 ceiling is what you get when every measurable pillar is performing well and the two unmeasured pillars receive their default values. It is not necessarily a ceiling of performance. It may, in part, be a ceiling of measurement.
This does not mean pubs are doing better than the data suggests. Many pubs scoring 59 are genuinely inactive on social media. But it does mean that the difference between a pub scoring 59 that is posting daily and a pub scoring 59 that last posted in September is invisible in the aggregate data.
When a brewery marketing director looks at an estate average of 56 and says "everyone is performing around the same," they may be looking at a measurement artifact rather than a genuine performance reality.
Why This Matters for Identifying Top Performers
The practical implication of this measurement issue is significant for brewery marketing teams trying to identify which pubs in their estate are the genuine digital standouts.
If engagement and content scores default to 3 for almost everyone, the pubs that are actually posting well do not show up in the data. They score 56 instead of 53, not the 65-70 their activity would justify under accurate measurement.
This has two consequences.
First, the best-performing pubs in the estate are invisible to any analysis based solely on aggregate health scores. A pub with a genuinely active landlord who posts five times a week, runs regular Instagram Stories, and engages with customers in comments will score around 56-59 under this framework, indistinguishable from a pub that has not touched social media since its last landlord left.
Second, the underperformers are also invisible. A pub that is genuinely dormant (nothing posted in months, the Facebook page last active in 2023) also scores around 53-56, pulled up by decent reputation and volume scores that reflect historical rather than current performance.
The engagement gap identified in the aggregate data (passive pillars averaging 68.4% of max, active pillars averaging 20.3%) is real and important. But within that gap, there is a wide range of actual activity that the data cannot currently differentiate.
The breweries that understand this will stop relying solely on automated health scores and start auditing engagement manually: looking at actual post frequency, actual follower growth, and actual review velocity, for at least a sample of their estate.
The Engagement Gap as the Biggest Opportunity
Setting aside the measurement issue for a moment, what does the genuine engagement data suggest?
Even allowing for the fact that some pubs scoring 3 are posting regularly, the balance of evidence suggests most are not. The review data (which is accurately measured) shows that volume scores cluster heavily at the top and bottom, with a bimodal distribution. The presence data shows 80.9% of pubs at score 10, a basic presence but not an active one. The website data shows most pub websites are static rather than regularly updated.
The pattern is consistent: pubs have set up their digital infrastructure and then largely stopped interacting with it.
The defining gap in pub digital health is not ratings (81.3% of pubs score 12 or 13 on reputation, meaning Google ratings are strong). It is not presence either: 80.9% have the basic profiles. The gap is active, consistent, human engagement: posting, responding, creating content, building relationships with customers online.
Line up all seven pillars by their average percentage of maximum and the gap stops being abstract. Reputation averages 79.5% of maximum, Volume 71.2%, Presence 62.7%, Website 60.1%, and Responsive 40.9%. Content and Engagement trail every other pillar at 20.5% and 20.1% respectively. Every other pillar in the seven-pillar system clears 40%. Only Content and Engagement do not.
An estate average of 3 on engagement means that if you can move your pubs to a genuine engagement score of 9 or 10 (achievable through twice-weekly posting with a content calendar and some basic accountability), you are not closing a modest gap. You are doing something almost no estate in the country has managed.
No estate in the 23-estate sample averages above 57. The ceiling is not a technical limit. It is a habit limit. And habits can be changed with the right infrastructure.
What Genuine Measurement Would Show
If engagement and content were measured accurately at scale, through direct API access, authenticated social media auditing, or manual review sampling, the data would almost certainly reveal a more complex picture than the current defaults suggest.
Among the 884 pubs in the dataset, a genuine measurement of posting frequency would likely show a distribution something like this:
A small minority, perhaps 5-10%, posting at least twice a week consistently. These are the pubs run by landlords who are personally invested in social media, who have found their voice, and who treat posting as a professional habit rather than an occasional chore. These pubs are scoring significantly above 59 on a genuinely accurate scale.
A middle group, perhaps 20-30%, posting occasionally: once or twice a month, around events, when something notable happens. These pubs are visible but not active. They are maintaining minimum presence without building a compounding following.
A large majority, perhaps 60-70%, effectively dormant. Posts from months or years ago. Profiles that exist but communicate nothing about what the pub is currently like or why someone should visit.
The 3-out-of-15 default captures the third category accurately. It says nothing useful about the first two.
The Recommendation: Manual Sampling
For any brewery marketing team that wants to understand the real digital performance of their estate, the recommendation is straightforward: supplement the automated health score data with a manual sampling audit.
Select 10-15% of the estate, weighted towards pubs you suspect are either significantly above or significantly below average, and manually review their Facebook and Instagram activity over the previous 90 days. Record post frequency, content type, and engagement (likes, comments, shares per post). Cross-reference with Google review velocity (new reviews per month, not total count).
This 90-day manual sampling will reveal things the automated audit cannot: which landlords are genuinely building their digital presence, which pubs are posting content that generates real community engagement, and which ones have simply never been supported well enough to get started.
That data (real frequency, real engagement patterns, real review building) is the intelligence a brewery marketing team needs to direct support and training where it will have most impact.
The automated health score tells you the state of the estate at a structural level. The manual sample tells you where the real work is.
Further Reading:
Tom Fairhall is the founder of Pub Post Pro, a done-for-you social media service for independent pubs and brewery estates across the UK.