Those are small numbers and I am not going to dress them up. Ten orders is ten orders. But the split surprised me enough to write it down, because it is the opposite of how I was told a direct to consumer brand gets started in India.
We built the Instagram. We made the ads. We wrote the copy and shot the creative and picked the audiences. That work produced nothing. Not a poor return, nothing, across the whole period.
Meanwhile people kept arriving on their own, at full price, with no discount and no reviews on the page to reassure them.
How we found out
By asking. After every order I send the same message, which is roughly: thanks, it ships soon, and how did you find us, because the information really helps a small business.
Almost everyone answers. That one question has told me more about my own business than any dashboard I have access to, and I would recommend it to anyone selling anything.
The first customer said they found us through Google, searching for the best honey brands. Then the answers started coming back differently.
I asked him what he had actually typed. His answer was two words.
Not a brand. Not a region. Not a variety. Just real honey, from someone who had been looking for a while and had given up on working it out from labels.
What it says about us
Another customer had asked something more specific, the best honey from Jammu and Kashmir. In that answer we came third in a ranked list, behind a well known name, which was fair enough. But at the end, in the part where it said what it would actually buy, it picked us. The reason it gave was this:
That sentence is the whole thing, and it is worth reading twice. It did not pick us because our honey is better. It has no way to taste honey. It picked us because we appeared checkable.
Everyone in this category writes pure. Pure is not a statement anything can act on. It is an adjective with no test attached. What our pages happen to contain instead are specifics: named laboratories, accreditation, the names of actual tests, pollen analysis, testing tied to a harvest. Those are things a machine can weigh, because they could in principle be checked.
We did not do that as a strategy. We wrote it that way because after reading about how much Indian honey is cut with syrup, vague language felt like part of the problem.
Now the part I would rather not write
The batch report it described does not exist on our website.
It cited a specific batch number and specific results, syrups not detected, acacia pollen dominant. There is no such page. There never was. We had said we test each harvest and would send the report to anyone who asked, and somewhere between that sentence and the answer it produced, a document got invented.
So at least one person bought our honey partly because of a page that was not there.
I could have left that out. Nobody would have known, and it makes a cleaner story without it.
But we sell honey to people who are tired of being told things that are not quite true, and it would be a strange business to build if the first thing I did with a bit of luck was tidy up the details. So: it happened, we did not do it on purpose, and it is being fixed by actually publishing the report rather than by hoping nobody checks.
Why I think it happened
Honestly, partly luck. But some of it looks structural, and none of it was expensive.
| What we did | Why it may have mattered |
|---|---|
| Wrote pages as questions and answers | Our headings are the sentences people type. That makes a page easy to lift into an answer. |
| Used specific technical language | Named tests and named accreditation are checkable. Pure is not. |
| Left the AI crawlers alone | Our robots file welcomes them by name. Many sites block them or never consider it. |
| Answered narrow questions properly | Nobody was competing seriously for what a buyer should ask a honey seller. There was room. |
The last one matters most. We are not outranking anybody on the big terms and will not for a long time. What we did was answer smaller questions carefully, and it turns out that is enough when the alternative answers are thin.
What I do not know
Plenty, and I would rather say so than pretend this is a playbook.
I cannot see any of it. There are no impressions, no ranking, no dashboard, nothing to test against. If this stops next month I will find out because orders stop, and I will never learn why.
I do not know whether it lasts. Models change. What we are being recommended for today is a thing another brand can copy in an afternoon by paying for a lab test.
And ten orders is not a result. It is an anecdote with a small number attached. Treat it as one.
Which is the plan. Publish the harvest report in full. Keep answering the awkward questions properly, including the ones where the honest answer is that a competitor does something better than we do. Keep asking every customer how they found us.
If an AI is going to tell people we can be checked, the least we can do is be checkable.
Single origin acacia honey from Jammu and Kashmir. One flowering, one harvest, nothing blended in.
See the honey →