User First, Algorithm Second: Researching Demand Beyond Search Volume

One industry, 5.6 million queries, 100,000 reviews.

A keyword tool can show zero search volume while people are asking for help in a Facebook group. A review can describe a problem that deserves a product feature. An earnings call can reveal why that problem matters to the business. We need a way to bring those signals together.

That was the thinking behind my brightonSEO San Diego talk, “User First, Algorithm Second.” I shared research spanning roughly 5.6 million queries and more than 100,000 reviews in the telecom industry. The work connected industry-level demand to specific customer frustrations and, in one example, a product idea worth testing.

The most useful slide in that story is a map of twenty demand sources. It asks us to follow people across the places where they look for answers, compare options, and describe what happened afterward. Understanding those differences is how research becomes a better content brief or a better customer experience.

Presenting User First, Algorithm Second at brightonSEO San Diego 2026. Photo supplied by Venkata Pagadala.

Twenty demand sources for understanding people

Search demand occupies one box in my framework. The other nineteen broaden the investigation to customer conversations, lived experiences, business decisions, and expert knowledge. Together, they help explain what people need and what prevents them from getting it.

The original twenty-source framework from the presentation. Search volume is one view of demand; the other sources add context, experience, and evidence.

I organize the twenty sources into five groups, each with a different job:

Search and language: Search demand and wikis help map the questions, entities, and terminology around a topic. Queries show what people type; reference material helps interpret what they mean.

Social conversations: YouTube, X, Reddit, TikTok, Facebook groups, and Instagram reveal questions, explanations, disagreements, and the experiences people choose to share.

Customer experience: Google Business Profile reviews, marketplaces, and app reviews show what customers valued, what disappointed them, and how they describe the gap.

Market context: Competitors, public data, earnings calls, job postings, and regulations help explain pricing, commercial priorities, investment, and constraints.

Expert knowledge: Conferences, newsletters, podcasts, and research papers contribute specialist interpretation, emerging questions, and evidence to investigate.

The twenty-source map is a prompt to investigate broadly. A company statement and a customer complaint carry different kinds of evidence. The value comes from understanding what each can tell us, then checking whether other sources support the same explanation.

Social listening reveals the context behind demand

Social listening deserves a central role in this work. People often explain their situation before they know the right search query. They describe what they have already tried, why an answer did not help, and which trade-off they cannot resolve. That context can disappear when we reduce the conversation to a keyword.

My own car-buying experience made this concrete. As a new father researching a Toyota Grand Highlander Hybrid, I found value in Facebook owner groups. I wanted to understand how the vehicle would fit my life. Other owners could explain experiences that specifications alone could not settle.

For a researcher, the follow-up questions in a Reddit thread or Facebook discussion can be as useful as the original post. Which detail changed the recommendation? What did several people misunderstand? Why did one answer earn trust? Those clues help define the information a customer actually needs.

A zero-volume estimate should therefore trigger another question: where else might this need be expressed? It is evidence about a tool’s estimate for a query, market, and period. It cannot establish that nobody has the underlying problem. Equally, one lively thread cannot establish the size of a market. Look for recurrence, check the context, and test the opportunity.

I want the research to follow the need across sources. If it appears in queries, community discussions, video comments, and reviews, those perspectives should inform the same decision. Separating them into departmental reports makes it harder to understand the customer’s full experience.

Why the platform changes the answer people want

People choose platforms partly because of the experience they want at that moment. Someone may want to watch a demonstration, compare detailed owner experiences, or ask a trusted friend. The same person can want all three at different stages of a decision.

TikTok makes the visual example especially relevant. A short demonstration can help someone understand a product or process quickly. TikTok’s own Creator Search Insights surfaces searched topics and content gaps where relatively few videos address frequent searches. That gives researchers a platform-specific view of questions that deserve attention.

Snapchat adds another perspective beyond the original twenty sources. Snap describes its origins around photo and video messages shared with close friends and family. For some users, the appeal may be the relationship and the ease of sharing a moment. Understanding that motivation requires asking the audience, rather than assuming every platform serves the same purpose.

Private conversations are not a public listening feed. Interviews and voluntary feedback can help explain what role those conversations play in a decision. The research question remains practical: who does this person trust, what are they trying to understand, and which format helps them act?

That answer should shape the content. A visual setup problem may need a demonstration. A complicated comparison may need a detailed explanation with trade-offs. A customer trying to arrange a visit may need an accurate appointment page and a working booking flow. Publishing the same article everywhere misses those differences.

Connecting social and visual discovery data

Google Search Console adds another useful view. Its platform properties let creators examine the queries, clicks, and impressions bringing people from Google Search to their Instagram, TikTok, X, and YouTube content. This measures discovery through Google; engagement within those platforms still needs its own analysis.

Adding Instagram, TikTok, X, or YouTube accounts to examine how people discover their content through Google Search. Screenshot supplied by Venkata Pagadala.

Example YouTube Insights report showing Google traffic by discovery surface. The 17.8K clicks are illustrative figures in the supplied screenshot, not results from my research or native YouTube engagement.

Visual discovery adds a different dimension. On September 24, 2026, Google announced multimodal search reporting, covering searches made with images through Lens, Circle to Search, image uploads, and Chrome’s Search this image feature. The screenshot below shows that filter. A person can start with something they see, even when they do not know what to type.

Multimodal search reporting in Google Search Console. This filter covers image-led web searches and is separate from social platform reporting. Screenshot supplied by Venkata Pagadala.

I would combine this performance data with the twenty-source research map. Use Search Console to see which published assets people discover through Google. Use social listening and reviews to understand the questions those assets still leave unanswered. Keep the source and the customer’s context attached to each finding, then segment the content around the task.

Earnings calls connect customer needs to business priorities

Earnings calls belong in a demand-research process because they offer a business perspective on the same market. I would examine prepared remarks and analyst questions for recurring discussion of retention, service costs, investment, or customer experience, then compare those priorities with what customers report.

Consider a hypothetical telecom example. Management emphasizes retention while store reviews describe long waits. That combination gives a team a focused question to investigate: could service friction be contributing to customers leaving? It does not establish causation, but it helps connect a customer problem to a business decision.

Public data, job postings, and competitor offerings can add context. A hiring pattern may suggest investment; it cannot confirm a launch. Management’s account should be checked against customer evidence and operational data. The useful output is a better question and a testable response.

Segment the evidence before creating the content

Collecting more sources only helps when we can organize what they mean. I would segment findings by the person’s situation, the task they are trying to complete, their stage in the decision, and the obstacle they describe. Keep the original language, source, and date so the interpretation can be checked.

In telecom, “switching providers” could involve an individual worried about losing a number, a family comparing total costs, or a business coordinating multiple lines. One broad topic label hides very different needs. Each segment may require a different explanation, proof point, or service.

This is where the industry map becomes useful. The presentation shows 5.56 million raw queries, rounded to 5.6 million, alongside 1.78 million keywords sized by volume and 37,712 topics mapped into communities. Zero-volume terms were included. I built the view in Omniscite, my own research platform, to illustrate the method.

The original industry map distinguishes raw queries, keywords sized by volume, and topics. It also reports 105.4K reviews across 465 verified businesses.

For each segment, the brief should specify the customer problem, supporting evidence, appropriate format, and next action. AI can help classify a large dataset, but the categories need checking against the underlying examples. Otherwise, a tidy map can conceal a misunderstanding at scale.

How review analysis revealed a product opportunity

The second part of the research shows why this approach can matter beyond content production. I analyzed more than 100,000 reviews of telecom stores across Texas. Long wait times emerged as a recurring frustration, pointing to an experience the business could improve.

That finding led to a product idea: digital check-ins and in-app virtual appointments. A customer might reserve a place, understand when help will be available, or complete an appropriate service remotely. Those options could address the uncertainty and wasted time described in reviews.

The separate theme-analysis view shown in the talk contains 47,485 reviews and highlights long waits. It is distinct from the 105.4K-review industry total. Digital check-ins and virtual appointments are proposed responses to test.

Reviews identify a reason to investigate; they do not tell us whether staffing, scheduling, or the service process caused the wait. I would check those possibilities with store teams and operational records before choosing a solution. A booking feature is useful only if the business can deliver the appointment experience it promises.

Turning the idea into revenue and useful engagement

The commercial hypothesis is straightforward. If an easier appointment process helps more customers complete their visit, it could reduce abandonment and preserve sales that would otherwise be lost. If it makes service easier to access, it could support retention. Those are potential outcomes to test, not results established by this analysis.

I would begin with a limited pilot and compare participating stores with similar stores over the same period. Measure actual wait times, completed visits, abandoned visits, completed sales, and repeat service contacts. Check whether any improvement holds after accounting for staffing and other changes, and whether the benefit justifies the cost.

Engagement should mean a customer successfully booking, arriving, or resolving a problem. More time spent in an app could simply mean the process is confusing. A useful experience may help someone finish faster.

There is also a content and organic-discovery opportunity. The review themes can inform useful pages about appointments, preparation, service availability, and what to expect. Those pages can answer real questions and connect people to the improved service. This approach fits Google’s guidance on people-first content, which emphasizes helping an intended audience accomplish its goal.

Organic traffic is a separate outcome to measure. Track whether those pages gain relevant search visibility and bring qualified visitors into completed appointments. Increased app engagement alone is not evidence of a ranking benefit. The product idea, customer outcome, and organic contribution each need their own evidence.

My prediction for voice in an agent to agent economy

My prediction is that voice use will grow rapidly as humans orchestrate their personal AI agents. I call this voice, rather than voice search, because the person is directing work: book a flight, order groceries, make a reservation, or help buy a car.

In the future I envision, a personal agent represents the individual’s goals and preferences. It communicates with agents representing airlines, retailers, restaurants, or car brands to explore options, exchange offers and availability, and coordinate the task within the person’s delegated authority.

Imagine saying, “Book a flight that arrives before noon, includes a checked bag, and stays within $500.” Your personal agent could compare airline agents’ offers and arrange an authorized booking and payment. For groceries, it could coordinate budget, delivery, and substitution preferences with a retailer’s agent.

The user remains the orchestrator. They define success, set boundaries, and choose how much authority to delegate. Voice is the human interface; agents exchange information through software protocols.

The foundations are emerging. The Agent2Agent protocol supports communication and collaboration between independent agents, while Visa’s agentic commerce framework describes checkout based on consented instructions and consumer-authorized limits. These developments support the direction of my prediction; they do not establish that every brand already supports this experience.

Understanding user behavior becomes more valuable in that model. A cheap flight can miss an important meeting; an affordable grocery substitution can fail a household’s needs. Communities, reviews, and the other demand sources help businesses understand the human whose preferences the personal agent represents.

Start with the person and follow the evidence

The twenty sources give us more ways to understand a need. Segmentation helps us decide whose problem we can solve. The review example shows how that process can lead to an idea for a better service, with content helping people discover and use it.

That is the habit I wanted to share at brightonSEO: keep asking why, connect the evidence, and make the next experiment useful. Start with one customer problem. Investigate it across the relevant sources. Then give the finding to the team that can act on it and agree how to measure the result.

Real SEO starts when you stop looking at everything through an SEO lens. Understand the person well enough to recommend something worth building.

For more perspectives from the conference, explore Previsible’s brightonSEO articles.