Internal — enter password
Internal — enter password
Brand health · Functional coffee category · fielded July 2026
What this is. Zappi asked 400 Americans what they think of us and ten competitors. It follows the How Brands Grow framework, so it measures what Byron Sharp says drives growth. Who knows you. Who would consider you. Who buys you. What you are known for.
Who they asked. People who don’t reject coffee alternatives. Mushroom coffee, functional drinks, adaptogenic teas. Not our customer list. Half men, half women, spread across the country. One wave, so there is no before-and-after here. Every number means “of Americans open to this category,” not “of Americans.”
The short version. We convert about as well as Ryze. We are just smaller. And the buyers we win don’t drink us much, because we are known for things this category doesn’t buy on.
Ryze is known by 39.5% of the category. We are at 22.7%. But of the people who know us, 46.9% have bought. Ryze’s number is 46.4%. Below awareness, nothing is broken.
Buyers say the product works. 92.7% of them say it meets their needs. But people who only know us like us less than almost anyone: 2nd-least liked of nine. And of the buyers we win, 12.9% drink us daily. Last of nine.
Rank every brand on whether it is known for what buyers say drives their choice. We come last of nine. We own “innovative”, “grounded energy” and “ethical”. Buyers rank those 12th, 9th and 11th out of 13.
Taste is the #1 reason people pick a brand here. It is our weakest association of thirteen. It is also the weakest for every brand we looked at. The thing buyers care most about is sitting unclaimed.
76% say they stick with what works instead of thinking about it. 69% only switch when something goes wrong. The biggest place they shop is Walmart, at 39%, and we are not there. Read the sample caveat in §2 before sizing that.
Seven actions follow in §7. Method, base sizes, and the weak data we kept out of the findings, are in the appendix.
01 — Where we stand
Everything in this report sits downstream of these two charts, so start here.
Prompted awareness is recognition. You show someone a list and ask which brands they have heard of. Top-of-mind awareness is memory. You ask what comes to mind with no list at all, and code what people type. The second is much harder to earn and much more valuable, because it is the shortlist people actually shop from.
Zappi codes top-of-mind answers from free text and only reports brands that clear a volume threshold. Five of the nine never clear it, which is why they are marked not coded rather than zero. Absence of coding is not absence of awareness, but it does mean almost nobody typed their name.
We are a strong second on the measure that matters and a distant fourth on the measure that is easier to buy. That is an unusual shape, and it is the reason Finding 03 matters: we have earned real memory, and we have filled it with the wrong things.
02 — How to read this report
Three things will mislead you if you read the raw Zappi dashboard. Two minutes here saves you from all of them.
Consideration and appeal only get asked of people who already know the brand. That sounds harmless. It isn’t.
Say five people have heard of a brand and four of them bought it. It scores 80%. A brand millions of people know can never do that, because its “aware” group includes everyone who shrugged.
So the number doesn’t tell you how good a brand is. It tells you how unusual the people who know it are.
Across the brands shown, the two move in opposite directions at r = −0.62.
The fix is simple. Stop dividing by “people who know the brand” and divide by all 400. Multiply awareness by consideration and you get a plainer number: how many people in the market are open to you. Do that and the relationship flips to +0.99. Brands more people know are brands more people would buy. Nothing was corrected. Only the denominator changed.
The whole study is 400 people. Sub-questions cut that down fast. Every chart carries one of these badges, and anything below “solid” draws its error bars right on the marks so you can see the doubt instead of reading past it.
This is the sharpest question anyone has asked of this data. Online panels don’t look like the country. They are built from people willing to answer questions for a small payment, which skews suburban and rural, less coastal, and less affluent than the customer base of a premium DTC brand.
There is a tell inside the data. Only 14.6% of these respondents say they have ever bought functional coffee from a brand’s own website. Across every channel mention in the survey, brand websites account for 8%.
That is not what this category looks like. Ryze, Everyday Dose and we all built our businesses direct-to-consumer. A sample that under-counts DTC buyers this badly is almost certainly over-counting mass-retail shoppers.
So Walmart’s 39.1% is right in direction and probably too big in size. What survives the doubt: this category shops shelves far more than our own customers do, and it runs on habit. What doesn’t survive: any precise number for the Walmart opportunity.
Two ways to settle it. Ask Zappi for the region and income breakdown, which isn’t in these exports. Then hold their channel mix against our own Shopify and retail split. If our business is 85% direct and the survey says the category is 15% direct, one of those is wrong about who the category is, and it matters which.
DIRTEA and Spacegoods are excluded by default. Both are UK brands with almost no US business, so their scores describe a few dozen unusual people rather than a market — detail in the appendix. Use the filter at the top of the page to add them back, or to narrow to any set of brands you want to compare.
Fund reach. The gap is 17 points of awareness, and mass channels are where that gets won. But don’t call the funnel healthy. Closing even half the gap to Bulletproof on aware-to-bought would add volume for less than awareness costs.
Two different jobs, and the survey separates them cleanly. Before trial, the problem is desire, not proof — we lead with efficacy and score last on liking, so more functional claims will not fix it. That is what Findings 03 and 04 are about. After trial, the problem is frequency, and it is the most commercially valuable line in this report: a point of daily-use conversion compounds on customers we have already paid to acquire. Occasion cues, morning-ritual creative, subscription cadence, flavour variety so the third cup of the week is not a chore. Validate the size of it against Recharge and Shopify before funding — at n=42 the survey can raise this but not size it.
Stop putting paid creative behind innovation and mission. We already lead the category on both, in territory buyers rank near the bottom. That budget is buying ground nobody is shopping for. Move the mission work to owned and organic, where it deepens loyalty instead of chasing trial.
This is what makes Finding 01 worth spending on. Buy reach while still pointed the wrong way and you just buy more of the wrong association.
Read these for shape, not height — each panel is scaled to its own brand’s range, because absolute ownership differs too much for a shared axis (Ryze sits around 24–30% on every attribute, Everyday Dose 13–15%, Four Sigmatic 8–10%). Those levels are in each subhead.
Between them these carry category importance ranks 1, 6, 7 and 8. Each panel shows what share of that attribute’s associations in the whole category belongs to each brand.
It is true, and it is the most interesting structural fact about this competitive set. Ryze, Everyday Dose, Four Sigmatic, Bulletproof, Javvy and Vitacup are all coffee. We are not. Buyers were asked about “mushroom and functional coffee beverages,” so they judged us inside a frame we don’t sit in.
The tempting conclusion is that our taste score is unfair. People expected coffee, got cacao and chai, and marked us down. The data doesn’t support it.
Rasa is also a coffee alternative with no coffee in it. Taste is Rasa’s 4th-strongest association. Ours is 13th of 13, and we are the only brand in the study where taste finishes last. Being a non-coffee brand does not push taste to the bottom. Rasa proves it is possible to be an alternative and still be known as a nice thing to drink. Our taste position is ours, not the category’s.
What the frame really tells us is commercial rather than defensive. Buyers already think about this category as coffee, and we are about to sell coffee. The new product enters a frame where we already hold 22.7% awareness and the 3rd-largest share of category associations. We are not starting cold, and it removes the one objection no message can answer: that we are not what you were looking for.
Pick taste. Commit to it for a year. Hero assets, packaging, landing pages, product innovation, subject lines.
Launch the coffee product on taste, not on function. It is the one product we have that competes inside the frame buyers are already using, and taste is the axis they judge on. Leading it with mushrooms and adaptogens would repeat the mistake Finding 03 describes.
It is the category’s top driver, our worst association, and nobody else owns it. You almost never get all three at once.
Pair it with proof on trust: third-party testing, visible dosages, review volume. Taste earns the trial and the second cup. Trust removes the reason to say no. Neither needs a new product.
This changes the order of everything above it. In a category where people don’t deliberate, being available beats being persuasive. That is the How Brands Grow argument showing up in our own data.
Distribution is probably worth more than any message change on this page, and it is the one lever creative cannot substitute for. It also has a clock on it. Habit compounds for whoever is already on the shelf, so the gap to Ryze gets harder to close the longer we wait.
Size it against our own numbers before committing capital. The survey says the category shops mass retail. It is not a reliable guide to how much.
07 — What to do
Two things this report cannot settle, listed so nobody acts on them prematurely: whether price is genuinely our biggest barrier (n=32 — test with Intelligems before touching price architecture), and which brands we actually lose customers to (n=26). Both are in the appendix as hypotheses.
08 — Appendix
Barriers and substitution are asked only of a brand’s own rejectors or loyalists, which cuts the base to a few dozen people. Both are shown with error bars so the uncertainty is visible. Treat everything here as a hypothesis to test.
base: 26 people — indicative only
bases: 4–84 people — indicative only
The asymmetry worth testing: 22.2% of our loyalists name Ryze as their fallback, while only 12.1% of Ryze loyalists name us. If real, we leak to Ryze about twice as fast as it leaks to us. Four Sigmatic (32.4%) and Vitacup (22.8%) loyalists name us most often — the two conquest pools where we already have permission.
DIRTEA takes 89% of its traffic from the UK and 1.25% from the US — about 1,800 US visits a month — and returns no US Google Trends signal at all. Spacegoods is 75% UK, 5.3% US. If almost nobody in America can buy a brand, the few respondents who know it are not a small version of the market; they are an unusual group who travelled or went looking. Of course a high share of them bought and liked it. That is what seeking something out looks like, not desirability.
Of the eleven brands, DIRTEA is the only one Google returns no US number for — its American search volume sits below Google’s reporting threshold, verified on two separate queries.
Read the last column only. Google Trends matches literal search strings, so a brand written several ways gets split across them: mudwtr alone returns 4.6 against Ryze’s 100, which is not credible — people also type “mud wtr” and “mud water”. The fix is Google’s topic entity, which merges every spelling, but this connector only accepts strings. Ranking brands on these numbers would be the same like-with-unlike mistake §2 is about.
Brit got to this data first and wrote it up in Notion. She and I worked separately, which makes the overlap worth something: where two people reach the same conclusion from the same numbers by different routes, the conclusion is probably not an artifact of either method.
Where we landed in the same place, independently.
Where she was ahead of me.
Where we disagree.
One flag for her. Her retailer argument uses Ryze’s first choice at 21.1%, which is more than 3× ours. The direction is safe. The number isn’t: 21.1% exceeds Ryze’s own reported purchase of 18.4%, which is impossible, so those two questions must use different reference periods. Use “more than three times” and don’t put the decimal in a retailer deck.
1.96 × √(p(1−p)/n) on Zappi’s
weighted bases. Weighting inflates true variance, so treat every whisker as a floor.1 − 6Σd²/(n(n²−1)).