Roaster Cafes Versus Cafe Only Shops in Visibility Data
Short answer: in our census of more than 1,000 shops across six regions, snapshot dated September 15, 2026, shops that roast their own coffee and also run a cafe show a somewhat higher median visibility score than cafe only shops, and roaster only businesses without a cafe score higher still. That is a real, measured pattern in our data. It is not evidence that roasting itself causes a higher score, and we want to be direct about that distinction from the start.
This post walks through how we grouped 1,109 shops into three categories, what the median scores actually look like for each group, and whether that pattern holds up when we split the data by region instead of looking at it nationally. All scores in this post are on our /40 census scale; we are not mixing them with the separate Southeast score bands used in earlier audit work.
How We Actually Built the Three Groups
The roaster_or_cafe field in our census data is free text, filled in during data collection rather than pulled from a clean, standardized list, which meant the first real step was building a consistent mapping rule before any comparison could happen. We defined it plainly: any label starting with roaster and also mentioning cafe or plus was mapped to roaster plus cafe. Any label starting with roaster alone, with no cafe mention, was mapped to roaster only. Any label starting with cafe was mapped to cafe only. A small remaining group that did not cleanly fit any of those patterns was set aside as an other bin.
Applying that rule across all six regional census files, checked September 21, 2026, produced 441 shops mapped to cafe only, 589 mapped to roaster plus cafe, and 62 mapped to roaster only, with 17 shops falling into the other bin due to messy or unusually long text labels. Those four groups total 1,109, matching our full census count exactly, which confirms the mapping rule accounted for every row rather than silently dropping any.
We are naming this mapping process explicitly because the underlying label field is genuinely messy, and a different mapping rule applied to the same raw data could plausibly produce somewhat different group sizes. We built one clear, defensible rule, applied it consistently, and are showing our work here rather than presenting the three group counts as if they arrived pre sorted and beyond question.
The Median Scores by Group
Using that mapping, median score_total, recounted and checked September 21, 2026, comes out to 15 for cafe only shops, 16 for roaster plus cafe shops, and 18 for roaster only shops, all on the same /40 scale. That is a real, if modest, upward step across the three groups, cafe only lowest, roaster plus cafe in the middle, roaster only highest.
It is worth sitting with how modest that gap actually is before drawing any conclusion from it. Three points on a 40 point scale, between the lowest and highest group, is a real difference, but it is not a dramatic one, and there is real overlap in the underlying distributions behind each median. Plenty of individual cafe only shops score above the roaster only median, and plenty of individual roaster only shops score below the cafe only median. A group median describes where the middle of a distribution sits, not a rule that applies to every shop within it.
We also checked blog presence and schema status across the same three groups, since those are two of the components that feed into the overall score. Both follow a similar direction, roaster plus cafe and roaster only shops tend to show a smaller share with zero blog posts and a larger share with schema present compared with cafe only shops, though we are reporting these as directional patterns within the same dataset rather than as separately validated findings, since they are built from the same underlying score components already described above.
Checking Whether One Region Is Driving This
A national median can hide a lot. If one region happened to have both an unusually large roaster population and an unusually high overall visibility level for reasons entirely unrelated to roasting, that single region could be pulling the whole national pattern in a direction that does not hold true everywhere else. So we repeated the same group and median calculation split out by region.
The ordering, cafe only lowest, roaster plus cafe in the middle, roaster only highest, held in the large majority of regions we checked, which is a meaningfully stronger result than the national figure alone would suggest, since it means this is not simply one region's pattern getting amplified into a misleading national average. In the regions where the roaster only group was especially small, sometimes just a handful of shops, we are treating that region's specific ordering with real caution, since a median computed from a very small group can shift substantially based on just one or two shops.
Where a region's roaster only sample was too thin to support its own reliable median, the more defensible approach was pooling it with a neighboring region's data rather than reporting a number built on too few shops, and we did that rather than force a per region breakdown everywhere the underlying sample could not support one. The overall conclusion is that this pattern is broad, appearing consistently across most of our regions, rather than being an artifact of any single one.
What This Pattern Does Not Prove
It is genuinely tempting to read a pattern like this as roasting your own coffee makes your marketing better, and we want to push back on that reading directly, because our data cannot support it. This is an observed association within one dataset, not a controlled comparison, and there are several equally plausible explanations that have nothing to do with roasting causing better visibility.
A shop that invests in its own roasting operation is also, plausibly, a shop that has already made a larger overall investment in its business generally, more capital, more staff, more operational maturity, any of which could independently produce both a roasting operation and a stronger online presence without one causing the other. It is also possible that shops choosing to roast their own coffee skew toward owners with a particular kind of marketing minded temperament from the start, again meaning both things trace back to the same underlying cause rather than one producing the other.
We are also not ranking any individual shop based on this pattern, and we want to say plainly that a cafe only shop with a modest score is not doing anything wrong by choosing not to roast. Roasting is a significant operational undertaking, equipment, sourcing relationships, a different skill set, and plenty of excellent, well regarded shops make a deliberate choice to focus entirely on the cafe side of the business rather than add roasting on top of it. This pattern describes an association in our dataset, nothing more, and treating it as a verdict on any individual shop's choices would be reading far more into these numbers than they can support.
Why This Pattern Is Still Worth Knowing About
Even without a causal claim attached, this pattern is useful context for a few different readers. For a roaster considering whether to add a cafe, or a cafe considering whether to add roasting, it is a real, if modest, data point that the businesses in our census combining both tend to sit somewhat higher on visibility than cafe only shops, alongside all the operational reasons for or against that decision that have nothing to do with visibility scores at all.
For a cafe only shop reading this and wondering whether the gap should worry them, the more useful takeaway is the earlier point about overlap: plenty of individual cafe only shops in our data score above the roaster only median. Whatever group a shop falls into, the actual, specific things that drive a visibility score, a working website, current schema, an active blog, consistent business information, are available to any shop regardless of whether it roasts its own coffee or not.
That is really the honest bottom line of this whole comparison. The group a shop belongs to is a weak predictor at best, given the real overlap in the underlying distributions, and the specific, concrete practices behind a stronger score are the same practices for every shop in our census, roaster, roaster plus cafe, or cafe only alike.
What the Overlap Actually Looks Like in Practice
It helps to think through a concrete, anonymous illustration of what overlap between groups means, since a median alone can make three groups sound more cleanly separated than they really are. Picture a cafe only shop with a excellent website, current schema markup, an active blog with a real posting rhythm, and a strong, consistent Google Business Profile. That shop's score sits well above the roaster only median of 18, despite belonging to the group with the lowest overall median. Picture, on the other side, a roaster only business that never got around to setting up schema, has not touched its website in years, and has no blog at all. That shop can easily sit below the cafe only median of 15, despite belonging to the group with the highest overall median.
Both of those shops are entirely plausible within our actual dataset, not hypothetical extremes invented to make a rhetorical point. That is exactly what real overlap between group distributions looks like on the ground, rather than as an abstract statistical concept. The median tells you about the middle of a distribution; it says nothing definitive about where any one specific shop sits within it.
This is why we keep returning to the same caution throughout this post. A pattern this real and this consistent across regions is worth reporting honestly, and we have done that. It is also modest enough, and overlapping enough, that no individual shop should read their own group membership as either a guarantee of a strong score or a verdict explaining a weak one. The underlying practices behind the score matter far more for any single shop than which of the three groups that shop happens to fall into.
How We Would Suggest Reading a Study Like This
There is a broader habit worth naming here, since this is not the only pattern in our data that could tempt an oversimplified causal reading. Any time a dataset like ours shows two things correlated, a business type and a score, a platform and a score, a region and a score, the responsible next step is asking what else might explain the correlation before assuming the first thing causes the second. We tried to model that habit throughout this post rather than just state the rule abstractly.
In this specific case, we named several plausible alternative explanations, general business investment, operational maturity, owner temperament, precisely because a honest data study should surface the alternatives its own author can think of, not just the explanation that makes the cleanest headline. A pattern that survives that kind of scrutiny and still holds up regionally, as this one does, is more trustworthy than one that has not been examined that closely, but surviving scrutiny is still a long way from proving causation.
We would encourage any reader, owner or otherwise, to bring that same habit to any marketing claim they encounter elsewhere too, inside this post or outside it. A correlation reported honestly, with its limitations stated plainly, is a useful piece of information. The same correlation dressed up as a causal rule is not more useful, it is just less honest, and it tends to lead an owner toward decisions based on a confidence the underlying data never supported.
Related Reading
More from Hillcane on this topic, plus the pages on the site that sit next to the work.
Frequently Asked Questions
Do roaster cafes score higher than cafe only shops in Hillcane's data?
Yes, as a measured pattern. Median score_total, recounted and checked September 21, 2026 across our census of 1,109 shops, is 15 for cafe only, 16 for roaster plus cafe, and 18 for roaster only, all on our /40 scale. The ordering held in the large majority of regions when split out. It is a real but modest difference, three points at most, with real overlap between groups, and not a causal claim.
Does roasting your own coffee cause a higher visibility score?
We cannot say that from this data, and we are not claiming it. This is an association observed within one dataset of 1,109 shops, not a controlled comparison. Shops that roast may also differ in overall business investment, capital, staff, operational maturity, or owner temperament, any of which could independently explain both the roasting and the stronger online presence without roasting itself causing the higher median.
How did you decide which shops count as roaster, roaster plus cafe, or cafe only?
We built a mapping rule from the free text roaster_or_cafe field: labels starting with roaster and mentioning cafe or plus became roaster plus cafe, labels starting with roaster alone became roaster only, and labels starting with cafe became cafe only. A small other bin held labels that did not cleanly fit. This produced 441 cafe only, 589 roaster plus cafe, 62 roaster only, and 17 other, totaling 1,109.
Does this pattern hold up in every region, or is one region driving it?
We recounted the same group medians split by region, and the same ordering, cafe only lowest, roaster plus cafe in the middle, roaster only highest, held in the large majority of regions we checked. Where a region's roaster only sample was too small to trust on its own, we pooled it with a neighboring region rather than report an unreliable number.
Is a three point gap on a 40 point scale a big difference?
It is a real but modest difference, and there is significant overlap in the underlying distributions behind each group's median. Plenty of individual cafe only shops score above the roaster only median, and plenty of roaster only shops score below the cafe only median, so the group a shop falls into is a weak predictor of any individual shop's actual score.
Should a cafe only shop start roasting its own coffee to improve its score?
We would not recommend it based on this pattern alone. Roasting is a significant undertaking, with equipment, sourcing relationships, and a different skill set. The practices that drive a stronger score, a working website, current schema, an active blog, and consistent business information, are open to any shop. Plenty of cafe only shops in the census already score above the roaster only median of 18 on our /40 scale.
Do roaster cafes also show more blog activity and schema presence?
Directionally, yes, within the same dataset. Roaster plus cafe and roaster only shops tend to show a smaller share with zero blog posts and a larger share with schema present compared with cafe only shops. We are reporting this as a directional pattern within the same score components already discussed, not as a separately validated finding.
Are these numbers on the same scale as the Southeast audit posts?
No, and we want to be explicit about that. This post uses our /40 census scale exclusively, from the nationwide snapshot of more than 1,000 shops across six regions dated September 15, 2026. It is never mixed with the separate Southeast score bands used in earlier audit work, since those come from a different scale and a different vintage of data. The medians of 15, 16, and 18 belong to the census alone.
How many roaster only shops are in the dataset?
62 out of 1,109 total, after applying our mapping rule to the free text roaster_or_cafe field, alongside 441 cafe only, 589 roaster plus cafe, and 17 in an other bin. It is the smallest of the three main groups, which is why some regional breakdowns were pooled with a neighboring region rather than reporting a median built on a handful of shops that one or two outliers could swing.
What is the honest takeaway for a shop owner reading this post?
The category a shop falls into, roaster, roaster plus cafe, or cafe only, is a weak predictor of its actual visibility score, given the real overlap between groups. The specific practices that drive a stronger score, an active blog, current schema, a working website, are the same practices available to every shop in our census regardless of which group it belongs to.
Work with Hillcane
Roaster cafes score somewhat higher than cafe only shops in our census. That is a real pattern. It is not proof that roasting makes your marketing better.
If you want to know where your own shop actually stands, regardless of which group it falls into, reach Hillcane at hillcane.co/contact or call (256) 384-2449.
Comments