I spent 2 years doing audience research manually. Now I'm trying to turn it into a business

I’d test agencies first, but sell a repeatable decision artifact rather than hours of research. Ask one agency to run the same brief across three clients and measure time saved plus how often findings change a campaign. If the output only informs, it gets cut; if it changes the next action, it becomes part of delivery.

One thing that helped me with a small $27 digital product was treating distribution as part of the product definition, not a separate channel. I’d test one narrow job with one buyer first—e.g. an agency using the research to answer a specific client question—before choosing the broader agency vs. brand path.

For each pilot, I’d ask what decision changed (offer, landing page, or ad angle) and keep the raw quote/source beside the recommendation. If nobody can point to a changed decision, the research may be interesting but not urgent; that signal seems more useful than general willingness to pay.

How did you decide this was worth building in the first place?

Thanks for sharing the numbers, that makes it much easier to follow.

Solid lesson. Which channel has worked best for you so far?

Good write-up. What would you do differently if you started again?

Curious how long it took before you saw the first real results?

SIGNAL: Distribution being harder than the research process is the familiar trap — and the agency-vs-brand fork is where a lot of research products stall. Both segments will say “interesting”; only one will name a near-term decision and pay.

GAP: “Audience research” is still an abstract label. Brands and agencies buy differently: a brand buys one decision unlocked (pricing objection, churn reason, competitive claim), while an agency buys a reusable method. If the intake doesn’t force a concrete decision for the next ~2 weeks, you get curiosity conversations, not paid work.

ACTION: Before you pick a segment, lock one falsifiable deliverable tied to a decision — e.g. “top 10 pricing objections in category X with dated quotes + source URLs + frequency, delivered in 7 days.” Put that same SKU in front of both. Whoever pays first tells you where urgency is. Keep evidence→interpretation→action labeled, and mark uncertainty (“one loud thread” ≠ “category consensus”).

Curious: across the magnesium/protein work, which single finding actually changed a client’s next move — and would that fit on a one-page decision brief?

— Francisco / TrixellaIQ — competitive intelligence for D2C brands
https://trixellaiq.com (sample on the site)

Across your past research projects, what decision did clients actually change after reading the findings?

I’d sell the first few to whoever pays fastest, not whoever sounds like a better long-term bet.

Agencies look nicer on paper (one process, many clients), but they also drag you into discounts and “can you white-label this?” pretty quick. A DTC brand with a live launch or a churn problem will usually just buy the report.

If I were you I’d package one tight deliverable — top objections + quotes + sources, done in a week — and run that past both. First paid job tells you more than another week of guessing.

The part about distribution being harder than building really resonates.

I’m experimenting with a related problem from the AI-search side. I built a stateless platform where you enter a brand and its competitors, get prompts to run in AI tools, then paste the responses back and analyze whether the brand is actually appearing in those answers and how it compares with competitors.

One thing I’m increasingly interested in is the connection between the two: what customers are saying publicly is one signal, but what AI models are actually surfacing when someone asks about a category or product is another.

It makes me wonder whether there’s eventually a useful feedback loop between customer research and AI visibility research — especially for brands trying to understand how they’re being represented to potential customers.

Really interesting to see you productizing something you spent two years doing manually.

Picking the wrong segment is less risky than both segments saying "interesting" and nobody paying. So I would let the first paid engagement choose: put a fixed-scope, fixed-fee teardown in front of both — say, 10 pricing objections in the category with quotes, source links, and frequency counts, delivered in 7 days — and give it a hard deadline. Whoever pays first tells you where the urgency actually is. Everything before that is a hypothesis.

One thing to watch with agencies, since a lot of people are pointing you that way: the one-method-many-clients math is real, but it also means agencies push hardest on per-project price and white-labeling. A brand buys one project at full margin; an agency buys the same method ten times and wants a discount on all of them. So track two numbers per segment: days from first conversation to payment, and revenue per hour you actually spent. If agencies take 3x longer to close for half the margin per hour, the scale argument may not survive contact with the spreadsheet.

And on distribution, I agree with the others: your existing research outputs are the inbound play. Publish the magnesium and protein findings, anonymized, as category reports. The brands that find those pages are your customer list arriving on their own.

The part I'd protect most is the jump from a conversation to a conclusion. Keep the raw excerpt, platform, date, and source URL next to every pattern you find. That makes client reviews more useful and helps you tell a recurring pain from one loud thread. For agencies, that evidence trail is probably part of the product, not just an internal quality check.

I'd start with agencies, but sell a single falsifiable deliverable, not "audience research." Something like: "10 pricing objections in category X with quotes + source URLs + how often each showed up." Brands buy a story; agencies buy a reusable method they can put on the next brief next week.

The distribution problem you named is real — the people who need this aren't searching for "audience research tool." They're searching for answers to category questions. Publishing anonymized findings (the magnesium/protein patterns you already have) might pull better than the service page.

One question: when a brand and an agency both say yes, whose timeline actually gets you paid in under 30 days?

Agencies seem like the easier first wedge because one win can repeat across clients, but the buying process may be slower. Have you noticed a meaningful difference in urgency or willingness to pay between the two?

Really interesting point about distribution being harder than building the actual research system.

I’ve been working on a somewhat related problem with Revencast. We’re trying to make the early-stage research/validation process easier for founders who want to answer questions like:

• “Is there actually a market for this idea?”
• “Who are the competitors?”
• “What are people saying about similar products?”
• “What should I potentially charge?”
• “And is this worth pursuing before I invest more time and money?”

What I’ve noticed is that the information is often already out there - the difficult part is bringing the different signals together and turning them into something a founder can actually use to make a decision.

I also think the agency vs direct-to-brand question is really interesting. Agencies could potentially have much higher repeat usage, while selling directly to founders/brands might make it easier to get closer to the actual problem and learn faster.

Curious to see where you take Insightios. I’m building in a similar space, so definitely following along.

Have early conversations with brands and agencies shown a meaningful difference in urgency or willingness to pay, or is the agency advantage still mainly a distribution hypothesis?

Once the tool gets usage, what behavior will tell you you've attracted a RevPages prospect rather than just a useful free-tool visitor—returning proposals, creating an account, or exploring the paid product?

The interesting product question is how you weight signal vs noise. If someone says they dislike a competitor's pricing on Reddit once vs seeing 50 Quora posts asking the same pricing question - that's a huge diff in conviction. What's your framework for turning raw feedback into prioritized insight? Like, do you have a scoring system, or is it more pattern recognition on volume and consistency?

i'd try agencies first, but with one very specific research question rather than a broad "audience insights" pitch. an agency can put it to work on the next client brief and tell you quickly whether the finding changed an ad, offer, or landing page. what have the DTC agencies you've talked to actually asked you for so far?

You know, what you're doing is real value — automation just doesn't give you that. But two years of manually reading through Reddit, YouTube, and Quora is an incredibly time-consuming process. When we built similar mention-monitoring across platforms for clients, the hard part was never finding the posts — it was filtering them: the same recurring question over and over, bots, irrelevant threads under the same keyword. Again, the problem isn't volume, it's the noise in the text.

Curious how you're currently sourcing candidate threads — we wrote up how we approach this kind of cross-platform monitoring at scale: https://data-ox.com/resources/blog/brand-monitoring/

This resonates a lot — how long did it take before you saw any real signal on it?

Agencies, for two reasons. First, a brand buys one research project and an agency buys the same methodology across every client. Your unit economics improve because the process is the same even when the category changes — you have already done magnesium and protein powders, so the framework clearly works across DTC verticals.

Second, agencies already have the distribution you are missing. One agency relationship gives you access to their entire client portfolio without finding each brand individually.

We face the same distribution bottleneck at UtilitySEO (SEO scanner). DA 3, four visits a month, the product works — discovery is the entire constraint. The people who need your research are not searching for "audience research tool." They are searching for answers to specific category questions, which means your existing research outputs are probably your best inbound asset. Publish the findings, not just the service page.