Surveys hear what people say when asked. Search shows what they do unprompted. A brand tracker gets into trouble when it treats one as the other.
What is brand tracking?
Brand tracking is the repeated measurement of how people know, see and use a brand. Shopify's guide describes it as measuring a brand's value over time, through metrics such as awareness, preference and usage. The repetition is the point, because one reading has nothing to compare with.
Search and other behavioural data now offer a second view beside surveys. They answer different questions, so the choice is rarely one or the other. Keep it apart from brand monitoring, which Shopify describes as reactive and focused on current mentions.
What does a survey measure that search cannot?
A survey can ask people what they think, including people who never search for you. That gives it four strengths that search data does not have.
- Awareness among non-buyers. You can ask a defined population if they know the brand at all.
- Associations. You can ask what people link to the brand, and how that compares with rivals.
- Preference and intent. You can ask which brand they would choose, and why.
- Who answered. You know the age, region and buying habits of your sample, within the limits of its quality.
Awareness is the classic example. One survey guide separates aided awareness, where people recognise a brand from a list. In unaided awareness, they name it without a prompt. Search has no equivalent to the first. Ask the unaided question first, because a list gives away the answer.
Unaided: "When you think of athletic apparel, which brands come to mind?" People type their answer in an open box.
Aided: "Which of the following bottled water brands have you heard of?" People select from a list that includes your brand.
The same guide adds that the two together show who recognises a brand and who defaults to it.
Surveys also have known weaknesses. People who buy a brand more are more likely to link associations to it. A score can therefore reflect past buying as well as perception. Always split scores by users and non-users.
Romaniuk and Nenycz-Thiel merged purchase panel data with consumer surveys. Consumers who bought a brand more often were more likely to give associations for it.
A higher share of category requirements was the stronger driver. The authors read this as competitor use dampening associations more than repeat buying reinforces them.
What can search signals show that surveys cannot?
Search shows what people do without a questionnaire, and it reaches back before your tracker began. A survey cannot be run on last year's market. The IPA notes that Google Trends data is free, generally goes back to 2004 and is available weekly.
That history covers rivals as well as you. You can look at a competitor's launch from three years ago today, or read a market before paying for fieldwork.
There is evidence of a link to outcomes. Share of search correlated with market share in the three categories Les Binet tested. Search remains thin on opinion.
Google states that Trends is not a scientific poll. A spike does not mean something is popular, and the data should be one input among several.
The IPA adds that not all search activity is positive. Trends also shows nothing about who searched, beyond the place and time.
A 2025 industry analysis reports that share of search correlates with salience and sales, citing Binet and its own validation. Its publisher is itself a research vendor.
How do surveys and search compare?
Surveys are stronger on opinion and on who answered, while search is stronger on history and cost. The table sets them side by side, so each question goes to the source that can answer it.
| Question | Survey | Search signal |
|---|---|---|
| Do people know us? | Yes, aided and unaided | Only indirectly |
| What do they think of us? | Yes, with associations and preference | No |
| Why did something change? | Partly, if you asked the right question | No, it shows only the movement |
| What happened before we started? | No | Yes, often for many years |
| How do rivals move week by week? | Rarely affordable | Yes, for brands with enough searches |
| Who exactly was measured? | Your defined sample | Unknown searchers in a place |
| Main source of error | Sampling, wording and respondent quality | Ambiguous terms and noise |
| Cost of one more reading | A new round of fieldwork | Low |
How does Demand Intelligence compare with other brand data?
Demand Intelligence is structured understanding of how market demand forms, changes and is expected to develop. It adds to other brand data and replaces none of it. Trendata publishes this guide, so the last column is its own view.
| Source type | What it does well | What Trendata adds |
|---|---|---|
| Market research and panels | In-depth answer to a question set in advance | A continuous view, including developments you had no question about yet |
| Social listening | Conversations and sentiment | Where demand forms and how it develops, at five levels |
| Retail and sales data | Realised sales by channel or product | Demand and needs before they become sales, also outside your own assortment |
| Internal BI and performance data | Shows exactly what happens inside the company | The outside view, 2-6 months earlier |
The 2-6 months compares with internal performance data: marketing, sales, CRM and finance. It is no guaranteed lead over competitors.
The five levels are Market, Category, Type, Specification and Attribute. For illustration: small kitchen appliances in the Netherlands, air fryers, double air fryers, 8 litres, dishwasher safe.
Which brand health metrics should you track?
Track a few metrics that each lead to a different action. Shopify lists seven common ones, including awareness, associations, preference, usage and market share. A short list is easier to keep stable than a long one.
| Metric | Main source | What a change suggests |
|---|---|---|
| Unaided awareness | Survey | The brand comes to mind more or less |
| Aided awareness | Survey | Recognition is growing or fading |
| Brand associations | Survey, split by users and non-users | Positioning is landing or drifting |
| Preference | Survey, read beside market share | Choice is shifting between brands |
| Share of search | Search data | Interest moves relative to rivals |
| Share of voice | Media or mention data | Your presence changed relative to rivals |
| Market share | Sales or syndicated data | The outcome, after the purchase |
Keep outcomes apart from indicators. Market share tells you what happened, while the others may tell you earlier.
How often should you run a brand tracker?
Measure as often as the metric can truly move and your sample can detect it. Search data can be read weekly at little extra cost. Each survey wave costs money.
No rhythm fits every brand. Shopify's example is a quarterly customer survey. Sample size should set the limit.
Pew Research explains this with election polls. A simple random sample of 1,067 cases gives about plus or minus 3 points on one figure.
A gap between two candidates has roughly double that margin. In Pew's example, a 3-point change in a lead between two polls carries a margin of about 8 points.
Subgroups are worse, because they have fewer cases. Pew also notes that trends across several polls are more reliable than two polls alone. The same logic applies to a brand tracker.
A sensible default is a monthly read of search signals and less frequent survey waves. Use the survey to explain what moved.
What does a wrong reading cost?
A wrong reading costs either a reaction you did not need or a shift you did not see. Which is worse depends on the decision attached, so name that decision before you set the threshold.
| Error | What happens | Cost | Guard |
|---|---|---|---|
| False alarm | Noise is read as a real change | Budget and attention move for nothing | Set thresholds, require two readings |
| Missed shift | A slow decline stays inside the noise | You react after sales have moved | Watch long averages and a second source |
| Broken series | Wording, sample or supplier changes | Old and new readings stop being comparable | Run both versions in parallel for a period |
| Bad sample | Insincere or unsuitable respondents answer | Scores look precise and are wrong | Quality checks, panel audits, trap questions |
| Wrong term | Search counts another meaning of the name | Interest appears to rise or fall | Review the search terms themselves |
Sample quality deserves real attention, because the problem can stay out of sight. A poor panel still produces clean charts.
Pew Research describes bogus respondents, who rush through online opt-in surveys for a reward without answering sincerely.
In a February 2022 opt-in survey, 12% of adults under 30 said they were licensed to operate a nuclear submarine. The real share rounds to 0%.
On probability-based surveys, very few respondents gave such implausible answers. Ask your supplier how the sample is recruited and checked.
Search has its own version of this overconfidence. Lazer and colleagues studied Google Flu Trends in 2014 and named it big data hubris: treating big data as a substitute for traditional methods, rather than a supplement.
Trendata publishes this guide and sits on the behavioural side, so it has that side's weakness. It does not tell you why people feel as they do.
Trendata measures aggregated orientation moments: moments when people look into products, needs and brands, explore alternatives and prepare choices. Brand position then shows how demand for a brand develops against the market and competitors.
Its demand forecasts run up to 12 months ahead, at category, type and specification level. That scope does not automatically cover brand position, and no forecast replaces a survey on perception.
The data covers 48 months of history and is updated monthly. It uses aggregated demand signals only, with no personal data and no tracking on your site.
A search line can tell you when to ask. A survey can tell you what people answer. Neither should do the other's job.
For rivals, see competitor analysis. For motives, see consumer insights.
Frequently asked questions
Can you do brand tracking without surveys?
Partly. Search and sales data can track interest and outcomes without a questionnaire. They cannot show associations, preference or awareness among non-searchers, so those questions still need a survey.
What is the difference between brand tracking and brand monitoring?
Tracking follows stable metrics over a long period. Monitoring watches current mentions in social media, news and reviews, so it helps with quick response rather than long-term position.
How large should the sample be?
That depends on the smallest change you need to detect. Remember that the margin on a change is larger than on one figure, and larger again for subgroups.
Does share of search replace brand tracking?
No. It is one useful metric within a tracker. It shows relative interest and can give an early signal, but it says nothing about what people think.
Sources
- Brand Tracking: 4 Methods and Key Metrics (Shopify, 17 Jan 2025)
- Aided vs. unaided brand awareness survey questions (SurveyMonkey)
- 5 key things to know about the margin of error in election polls (Andrew Mercer, Pew Research Center, 8 Sep 2016)
- Online opt-in polls can produce misleading results (Mercer, Kennedy and Keeter, Pew Research Center, 5 Mar 2024)
- Behavioral brand loyalty and consumer brand associations (Romaniuk and Nenycz-Thiel, Journal of Business Research, 2013)
- Binet presents fast, cheap, predictive Share of Search metric (IPA, 2020)
- Demystifying share of search (Achaibar and Kodu, Kantar, 11 Aug 2025)
- FAQ about Google Trends data (Google)
- The Parable of Google Flu: Traps in Big Data Analysis (Lazer et al., Science, 2014)