Evidence-based marketing is marketing based on evidence, not just data

1How marketing knowledge develops

How can we tell whether a marketing rule really holds? Most accepted truths come from a single study on one sample and fall apart when repeated. The tradition of empirical generalisations works the other way round: it collects dozens of datasets across categories, countries and years, and calls something a law-like pattern only if it survives everywhere. Meta-analyses and replications apply the same filter. Psychology has shown that without them, barely half of published effects can be replicated.

This topic also includes the science of science: citation data and the publication system have their own biases. A new threat comes from synthetic respondents and AI-contaminated data. Until shown otherwise, a simulated respondent is not a substitute for a person, and data provenance is part of the evidence.

2Law-like patterns in buying behaviour

The oldest and most firmly established part of marketing science. The NBD-Dirichlet model describes how people actually buy: most customers buy infrequently and are loyal to a repertoire rather than a single brand. The Double Jeopardy law says that small brands suffer twice: they have fewer buyers and those buyers are slightly less loyal. Duplication of purchase shows that brands share customers with competitors in proportion to their size.

These patterns have survived fifty years of replication across categories from washing powder to football clubs and petrol, and more recently in B2B too. They help marketers calibrate expectations: when someone promises results that break the pattern, the Dirichlet benchmark shows what is normal.

3Growth of brands, categories and new products

Brands grow primarily through penetration: attracting new and occasional buyers, rather than increasing the loyalty of existing ones. Retention and acquisition are not slogans but empirical questions with measured ratios. The same behavioural mechanisms drive the growth of whole categories.

Growth also includes the spread of innovations: the Bass model describes the adoption curve, and meta-analyses show which new-product success factors replicate. For technology products, the leap from enthusiasts to pragmatists is critical.

4Segmentation, targeting and positioning: when they help and when they limit growth

Most marketing questions begin with ‘who is our target audience?’. Yet actual buying data show that brands in a category share the same buyers in proportion to size, rather than by segment. Buyers of competing brands barely differ in demographics or attitudes. Market-based assets theory argues that narrow targeting and differentiation restrict growth. Nor is there confirmation that targeting at-risk customers or psychological profiles delivers the returns attributed to it.

This does not mean segmentation is pointless. There is a difference between segmentation for understanding (who buys in the category, when and why, and which buying situations exist) and segmentation that restricts reach. Positioning also has two meanings: an internal decision about the category and situations in which people should recall the brand, and external communication. Good positioning shows up in stronger memory links to situations, rather than a sentence in a brand book, and must not narrow the market more than necessary. Work on B2B positioning shows that there too it is a long-term capability rather than a claim. Personas are a qualitative aid, not a scientific segment. This topic is contested among evidence-based marketers, and the assistant presents both sides.

5Brands: equity, salience and culture

What exactly is brand equity in a customer's mind? Keller's framework describes knowledge and associations; signalling theory sees the brand as a quality guarantee under uncertainty. The behavioural school adds salience: a brand wins when people recall it in a buying situation. Distinctive assets (colours, shapes, characters) can be measured and benchmarked.

Brands also live within culture: they help build identities and communities, and can become targets of protest. Brand equity can also be connected to finance: it measurably affects acquisition, retention and company value.

6Mental and physical availability: buying situations and ease of purchase

A brand grows when people recall it in a buying situation and can easily buy it. Mental availability is neither awareness nor liking. It is a network of memory links between the brand and situations in which people enter the category (Category Entry Points). Salience can be measured, and brands mostly differ in the number of links rather than their quality. Physical availability is the other half: distribution is closely associated with market share, and the relationship is non-linear. Large brands gain more from each additional outlet.

The evidence base is uneven. Salience and the relationship between distribution and share have peer-reviewed support across markets. The specific Category Entry Points method is so far described mainly in work from the Ehrenberg-Bass cluster and reports from the LinkedIn B2B Institute that are not peer-reviewed, so the assistant presents it with a caveat. In B2B, physical availability translates into being easy to find, an available salesperson, a clear website and price, partners and easy onboarding. Direct B2B measurements of this are currently missing from the corpus; this is a transferred implication.

7Advertising: how it works and how to measure it

Advertising elasticities are small and surprisingly stable: meta-analyses of hundreds of campaigns show that advertising sells, but less and more slowly than promised. Much of the effect is long-term, through memory and brand availability in people's minds. Creative execution has a measurable influence on sales, and share of voice is associated with growth in market share.

The other half of the topic is measurement. Observational data systematically flatter advertising: clicks and attribution measure correlations rather than causes. Only experiments, such as split-cable tests and platform randomised trials, show the true additional effect, and their findings tend to be more restrained. The economics of measurement are unfavourable: reliable evidence requires enormous samples.

What an advertisement contains (creative) and how much of it is bought and how (media) are separate levers with their own literature: see Creative and attention, and Media and investment.

8Creative and attention: what makes advertising effective

Budget and reach determine how many people advertising reaches. Its content determines how much of that reach becomes memory and sales. Creative execution has a measurable influence on sales beyond media weight, and emotionally led campaigns have better business results in case databases than rational ones. Humour helps attention and recall, but only under certain conditions. Advertising mostly does not persuade: it refreshes memory and keeps the brand in the running. Brand linkage, distinctive assets and whether people notice the advertisement at all are therefore decisive.

Attention is scarce and not all reach is equal: people overlook advertising, video viewing is brief and dynamic, and packaging on a shelf is also advertising. Marketers are poor at estimating advertising effectiveness, and pre-tests focused only on attention can mislead.

Limits: the peer-reviewed core comes from consumer markets. Practical terms (fame, fluent devices, the cost of dullness, wear-out curves) come from case databases and proprietary tests (IPA, System1, LinkedIn), rather than experiments. Direct B2B measurement of creative is missing from the corpus, and the assistant says so.

9Media and investment: reach, frequency, continuity and share of voice

Knowing that advertising works through memory is one thing; knowing how to buy reach is another. Advertising elasticities are small, positive and stable: around 0.1 in the short term, two to three times greater in the long term, and smaller for established brands. Effective frequency is low and further repetition brings diminishing returns, so reaching light buyers and non-buyers is worth more than frequency among loyal buyers.

Excess share of voice predicts long-term changes in share, although the exact ratios are contested. Advertising effects persist, but briefly for most brands. When a brand stops advertising, sales fall slowly at first and then faster. Multi-platform campaigns outperform single-platform campaigns, and timing across the economic cycle matters.

Limits: elasticities, frequency and persistence are supported by meta-analyses of hundreds of brands. Rules about share of voice and the brand-versus-activation ratio rely on case databases and one research cluster. The direction is supported, but the specific ratios are not. Evidence about attention quality by channel currently comes from books.

10Customer value, pricing and discounts

Price elasticity has its own empirical generalisations, and reference prices explain why customers respond to a price change rather than its level. Discounts send sales up in the short term, but breaking down the effect shows that most of the increase is a shift over time and between brands. Price promotions generally do not produce long-term growth, and frequent discounts erode the reference price.

Zero is a special price with its own psychology. Pricing is also an organisational capability: who decides prices and how has a measurable influence on results.

Price begins with customer value: the difference between benefits and costs, value in use, willingness to pay and the brand price premium can be measured in both consumer and business markets. A value proposition should rest on a few differences that really matter to the customer and a quantified impact. Value-based pricing is rare in practice but more profitable than cost-based or competitor-based approaches, and it is subject to cognitive biases. Switching costs change the relationship between satisfaction and loyalty.

11Loyalty, satisfaction and CRM

Satisfaction is associated with loyalty only under certain conditions: switching costs moderate the relationship, and satisfied customers still leave. NPS as the ‘best predictor of growth’ did not survive independent replication. It holds up as one measure associated with revenue, but not as the only number. Behavioural loyalty tracks Dirichlet norms far more than it responds to loyalty programmes.

The CRM side: churn can be predicted, but proactive retention can also do harm. The value of the customer base (CLV) can be linked to company valuation. Customer metrics bridge marketing and finance.

12The psychology of decisions and influence

People do not make decisions like calculators. Heuristics and biases, such as anchoring, framing, the status quo and loss aversion, are systematic and predictable. Social norms and social proof change behaviour, and mere exposure builds liking without persuasion.

But be careful about effect sizes: a famous meta-analysis of nudging became contested. After correcting for publication bias, the average effect is uncertain, and critics also question the variation between effects. The assistant therefore always presents both sides on nudging. Specific nudges work under certain conditions, depending on the domain and design.

13Digital marketing, platforms and AI

Algorithms increasingly mediate the market: recommenders and playlists determine exposure, platforms have their own power, and demand both spreads into the long tail and concentrates around superstars. Online reviews and word of mouth have measured sales elasticities, and viral content has a psychological structure.

This topic also includes people's relationship with AI: they respond differently to algorithmic decisions than to human ones, resistance to machines depends on conditions, and chatbot acceptance is measurably different. Algorithmic pricing influences trust and search behaviour.

14Social media, influencers and online recommendations

Online recommendations work. Meta-analyses confirm that electronic word of mouth is associated with sales, but the effect's strength depends on the platform, product and whether it concerns review volume or valence. Content that evokes strong emotions spreads more. Content published by a company on social media can increase purchases, especially alongside advertising. Social media advertising can be measured through experiments, and ordinary correlational methods usually overestimate its effect.

Influencer marketing has a growing peer-reviewed literature: credibility and fit with the product matter, rather than necessarily youth or audience size. In B2B, social media is used for brand building, content marketing and social selling. Evidence here is mostly conceptual or survey-based, rather than sales measurement. The corpus lacks data on organic reach, paid formats beyond Facebook and the long-term effect of social media on brands.

15Marketing and company value

A language the finance director understands. A company's market orientation measurably translates into profitability, confirmed by meta-analyses across countries, with moderators. Market-based assets, such as brands, relationships and distribution, accelerate and stabilise cash flow, and marketing investment has generalised elasticities for company value.

This also includes the resource-based view (an advantage from scarce resources that are hard to imitate), marketing capabilities as a benchmark, and empirical findings that the influence of the marketing department is associated with company performance.

16B2B: market structure, competition and the economics of growth

Many poor B2B decisions start with defining the market: who we compete with, where the category ends and whether an internal solution or postponing a purchase also counts as competition. A market can be defined from the customer's perspective, by what substitutes for what in use. Brands then compete directly across the category, rather than in protected segments.

Business buying follows the same patterns as consumer buying: in aviation fuel contracts, foreign exchange services and industrial purchases, shares, penetration and repeat buying fit the Dirichlet pattern, and brands share customers according to size. Market share is associated with profit, but less strongly than assumed, and how it was gained matters. A small proportion of buyers accounts for a large proportion of revenue, but large brands do not attract light buyers disproportionately.

Limits: B2B evidence is peer-reviewed but comes from few categories and one research cluster. Customer concentration is measured only through its influence on supplier performance, and studies disagree: one finds a direct negative effect, another an inverted-U relationship. The corpus has no direct measurements of penetration versus depth or doing nothing as a competitor; the assistant presents these as transferred implications.

17B2B: buying situations, risk and the buying group

Companies do not buy more rationally than consumers. They buy in groups and above all fear making the wrong choice. A buying group with several roles and stages decides the purchase, rather than an individual with a spreadsheet. Classic research on organisational buying documented this in the 1970s, and modern buying-journey data show that most of the buying journey takes place without a salesperson.

Brands therefore work in B2B too. They provide insurance against risk. No one ever got fired for buying IBM. Empirically, the greater the risk and importance of a purchase, the more weight the brand carries. Trust develops in both the company and the individual salesperson, through different routes.

In practice, distinguish the buying situation: a new task, a modified rebuy and a straight rebuy involve different numbers of people, durations and risks. Procurement, the budget owner and the user have different interests within the buying group. Internal agreement and fear of a wrong choice are central: companies form shortlists and choose a safe option, rather than necessarily the best offer. Professionalised purchasing processes reinforce these patterns.

18B2B: brand growth and the 95:5 rule

Brand growth patterns apply beyond the supermarket. The Double Jeopardy law also replicates in B2B: smaller brands have fewer buyers, and those buyers are slightly less loyal. B2B brands share customers with competitors according to market share rather than segment. Growth here too is driven by penetration rather than loyalty.

The 95:5 rule says that the vast majority of companies are not buying right now. Advertising therefore mainly works for future buyers: it builds memory links to buying situations so that they recall the brand when the purchase comes. Much of this evidence comes from the Ehrenberg-Bass and LinkedIn B2B Institute ecosystem, and some consists of reports that are not peer-reviewed. The direction is increasingly supported; treat the specific numbers cautiously.

19B2B: relationships, trust and value-based selling

Once a company enters the buying process, relationships and proof of value win. Relationships rest on trust and commitment. Commitment-trust theory is a core framework for business relationships, and working partnerships between distributors and manufacturers follow the same logic. In knowledge-intensive services, the customer co-creates value, changing both parties' roles.

Value-based selling is a capability in its own right: the salesperson understands the customer's business model and can quantify the impact, selling differently and more effectively than a solution salesperson. This is the other half of the B2B story alongside brand growth: the brand opens doors; value wins deals. Value measurement and value-based pricing are covered under Customer value, pricing and discounts.

20B2B sales: sales teams, ABM and omnichannel

The sales force is the most expensive part of B2B marketing and can be managed with models: team size, territories and effort allocation have verified decision frameworks with a large impact on profit. Account-based marketing formalises selectivity: concentrating resources on named accounts.

The buying journey is also moving towards self-service: buying groups spend only a fraction of their time with salespeople and expect a mix of digital and personal channels. Evidence about this shift mostly comes from consulting surveys rather than peer-reviewed measurement. Treat it as a map of the terrain, rather than measured effects.