Brand tracking and funnel studies
Awareness, consideration and purchase for your brand and its real competitors, asked in one grid so conversion rates compare. A baseline, or repeated waves.
Market research · Vienna
Market research that says what the sample can carry.
We design, field and analyse brand and consumer studies for companies in Austria and the DACH region. Every number comes with its confidence interval, its base and its limits, so you can decide on it.
- Brand tracking and funnel studies
- Cross-country designs
- Dashboards and reporting
What we do
From the research question to a decision you can defend
Four services that fit together. Take one, or let us run the whole study.
Questionnaire design and programming
A tested instrument: logic, randomisation, panel redirects and automated import checks, before the first interview is fielded.
Analysis and driver models
Weighting, confidence intervals, driver models and segment profiles. We tell you what moves a metric, and what the data cannot show.
Dashboards and reporting
Interactive dashboards that suppress thin subgroups, a plain-language management deck and a field report that documents every exclusion.
How we work
Five steps, each with something you can check
Brief and hypotheses
We start from the decision the result must inform, and write down what would change your mind.
Instrument
Questionnaire, brand grid and quotas, tested in the survey platform before launch.
Fieldwork and quality
Attention checks and speeder rules run in the field. Exclusions are counted and reported.
Analysis
Weighting, intervals and models with the multiple-comparison corrections the question needs.
Delivery
Deck, dashboard, field report and re-runnable scripts, so every number can be traced.
Showcase
A two-country brand study, start to finish
ZINTO is a fictional sugar-reduced snack bar. We use it to show the full workflow, from design and field quality to weighting, analysis and reporting.
Fictional brand, synthetic data. The figures were generated, then analysed with our standard pipeline, so the case study shows real method without exposing a client. Hover or tab to any bar for its interval and base, or switch a chart to its table.
completed interviews in Austria after quality control
median length of the interview
excluded for failed attention checks or speeding
design effect of the weighting
brands asked in one grid, so rates are comparable
Aided awareness across all brands
Share of respondents who recognise each brand when shown the list. Whiskers are 95% confidence intervals.
of Austrians recognise the brand
in the overall ranking, behind the established mass-market brands
as well known as (), the best-known brand below it
From awareness to purchase
Where the brand loses people, set against the median of brands with at least 30 aware respondents.
The gap is at the first gate: of those aware would consider the brand, against in the category. Once considered, it converts as well as its competitors ( against ).
Who knows the brand
Age and gender matter, education does not. They were tested together in one model, so one cannot masquerade as another.
Awareness falls from among 18 to 29-year-olds to among those over 60, and is higher among women ( against ). Education, income and region show no effect once age and gender are accounted for.
Where the brand is noticed
Only respondents who know the brand were asked. Channels with fewer than 30 respondents are not shown.
The shelf is the main point of contact (), social media comes second. This is an association, not an effect: the order of contacts was not recorded.
What the brand team takes away
The four findings the rest of the case study supports, in the order we would present them.
- Consideration is the bottleneck, not awareness. of those aware would consider the brand, against in the category.
- The product holds up once it is considered. of considerers also bought, in line with the category ().
- Aim at age and gender. Awareness runs from to across age bands; education, income and region add nothing.
- The 2026 gain is real. After removing the effect of the survey change, awareness rose by .
Tracking over time: real growth versus a change of method
Between the 2022 and 2026 waves the awareness question changed from a forced-choice matrix to a tick-list. That alone lowered awareness for almost every brand, so the raw change cannot be read as brand growth.
raw change in awareness of the brand
median change of the comparison brands, which is the size of the method effect
method-adjusted gain (): without a real gain
Austria, base in 2022 and in 2026, both weighted. Intervals for the adjusted gain are bootstrapped; p-values for the single brands are corrected for multiple comparisons (Holm).
Knowing what the sample can carry
The same questionnaire ran in two countries. The brand was far less known in Germany, and that decides what we are allowed to report.
Austria
completed interviews, of which know the brand
- Funnel conversion by stage
- Brand image, drivers and segments
- Contact channels
Reported in full
Germany
completed interviews, but only know the brand ( would consider it, bought it)
- Aided awareness, with its interval
- Funnel conversion, drivers, segments and channels
- Pooling with Austria, which has a different population and quota plan
Brand modules not reported: base below minimum
What you receive
Interactive dashboard
Filters for country, wave, age, gender and contact channel act on every figure at once. Subgroups with fewer than 12 respondents are never shown, and the countries are never added together. The demo has a German interface and synthetic data.
Open the demoManagement deck
Findings in plain language, one message per slide, with a methods annex for the sceptical reader.
Field report
Sample, quotas, exclusions, weighting and data-quality flags, written down before anyone asks.
Re-runnable analysis
Scripts and logs, so every figure in the deck can be reproduced from the cleaned data.
Questionnaire files
Survey files that passed automated import checks, plus a codebook that makes waves and countries stackable.
Standards
Six rules we do not bend
A precise number on a thin base is a risk to your decision. These rules keep the precision honest.
01
An interval on every estimate
A point value alone invites false precision. Results carry their 95% interval and their base.
02
A minimum base for every claim
Brand results rest on at least 30 respondents. Dashboard cells under 12 are suppressed, never shown as a noisy percentage.
03
Quality exclusions are reported
Attention checks and speeder rules are set before fieldwork. We tell you how many interviews they removed, and why.
04
Weights come with their cost
Every weighted result states the design effect and the effective sample size it leaves.
05
Countries are compared, not pooled
Different populations and quota plans are laid side by side. We compare only what both fields actually asked.
06
Association is not effect
Cross-sectional data show relationships. We write “is associated with” unless the design supports a causal word, and we correct for multiple comparisons.
About
Research-led, built for decisions
alpha-metrics.ai is a market research practice based in Vienna. We combine careful survey craft with statistical discipline, and we deliver it at the pace of a commercial project.
Our focus is brand and consumer research for companies in Austria, Germany and Switzerland: who knows you, who considers you, who buys, and what would change that.
- Based inVienna, Austria
- MarketsAustria, Germany, Switzerland
- MethodsOnline surveys, tracking, driver models
- FieldworkEstablished online panels
- LanguagesGerman and English
- Data protectionHandled under the GDPR
Contact
Tell us the decision you need to make
A few lines are enough. We will tell you whether research can help, and what it would take.
Vienna, Austria · LinkedIn