Creative Effectiveness & Why Intuition Fails

Copper Sun · 4 entries · last verified July 2026

Copper Sun tracks the empirical literature on advertising creativity — work that measures which executional choices track sales rather than which ones win admiration. The most useful finding for teams adopting AI is also the least comfortable: experienced marketers predicting which ads will sell perform at close to chance. That result sets a standard any production process has to be measured against, including one that produces more.

Contents — 4 entries
  1. 1.Modeling the Determinants and Effects of Creativity in Advertising
  2. 2.Creative That Sells: How Advertising Execution Affects Sales
  3. 3.Marketers' Intuitions about the Sales Effectiveness of Advertisements
  4. 4.Finding Creative Drivers of Advertising Effectiveness with Modern Data Analysis
  5. Frequently Asked Questions

Modeling the Determinants and Effects of Creativity in Advertising

Smith, MacKenzie, Yang, Buchholz & Darley, 2007. Marketing Science 26(6), 819–833.

Copper Sun draws on this for a workable definition of creative quality. Across a series of studies running from scale development through model testing, the authors show that perceived ad creativity is determined by the interaction of divergence and relevance, with overall creativity mediating their effects on consumer processing and response. Divergence without relevance is noise. Relevance without divergence is wallpaper. Treating creativity as two measurable dimensions rather than one aesthetic judgment gives a brief something to specify.

Examines:
What determines consumer perceptions of advertising creativity, and how those perceptions affect processing and response, tested across multiple studies.
Copper Sun draws on:
The divergence-by-relevance interaction — used when Copper Sun frames a creative brief, since both dimensions can be requested and reviewed separately.

Creative That Sells: How Advertising Execution Affects Sales

Hartnett, Kennedy, Sharp & Greenacre, 2016. Journal of Advertising 45(1), 102–112.

Copper Sun cites this as the large-sample link between execution choices and actual sales. The authors coded 158 creative variables across 312 television advertisements carrying commercially validated short-term sales outcomes, replicating and extending Stewart and Furse's earlier work on creative devices. The value is the method: executional features tested against in-market sales rather than against recall or liking, which are the measures most creative testing settles for.

Examines:
Whether specific advertising execution features predict short-term in-market sales, across 158 coded creative variables and 312 television advertisements.
Copper Sun draws on:
The sales-validated approach to execution — used when Copper Sun distinguishes creative choices with measured commercial support from those justified by taste.

Marketers' Intuitions about the Sales Effectiveness of Advertisements

Hartnett, Kennedy, Sharp & Greenacre, 2016. Journal of Marketing Behavior 2(2–3), 177–194.

Copper Sun treats this as the most important result in the set. A global sample of marketers predicted which television advertisements were more or less sales effective, and their predictions averaged 51% accuracy — no better than chance. Multivariate analysis found category experience and marketing or insights roles produced slightly better predictions, which is a modest qualification on a stark finding. Confidence in creative judgment is not supported by measured accuracy, and that applies before any question about AI arises.

Examines:
Whether experienced marketers can intuitively identify which advertisements are more sales effective, measured against commercially validated sales outcomes.
Copper Sun draws on:
The 51% accuracy result — the basis for Copper Sun's position that creative decisions should be argued from evidence and tested, not settled by seniority in the room.

Finding Creative Drivers of Advertising Effectiveness with Modern Data Analysis

Williams, Hartnett & Trinh, 2023. International Journal of Market Research.

Copper Sun monitors this as the methodological update to the creative-effectiveness question. The paper applies modern analytical techniques to the problem of identifying which creative drivers relate to advertising effectiveness, extending a line of research that has historically been limited by small samples and by measures that stand in for sales rather than measuring them. Copper Sun follows this work because the analysis method, not the opinion, is what moves the field.

Examines:
Application of modern data analysis techniques to identify creative drivers of advertising effectiveness.
Copper Sun draws on:
The analytic approach to isolating creative drivers — followed by Copper Sun as the standard of evidence a claim about 'what works creatively' should meet.

Frequently Asked Questions

Can experienced marketers tell which ads will sell?

Not reliably. Hartnett, Kennedy, Sharp and Greenacre (2016) put the question to a global sample of marketers and found predictions averaging 51% accuracy against validated sales outcomes, which is close to chance. Category experience and a marketing or insights role produced slightly better results. The finding does not say creative judgment is worthless; it says confidence in it is not calibrated, which is an argument for testing rather than for deferring to the most senior opinion.

What actually makes advertising creative, in measurable terms?

Smith, MacKenzie, Yang, Buchholz and Darley (2007) found perceived creativity is produced by the interaction of divergence — how much the work departs from the expected — and relevance to the audience and the category. Neither dimension alone produces the effect. That makes creativity specifiable in a brief: a piece can be reviewed for whether it is genuinely divergent and whether the divergence is relevant, as separate questions.

Does this research mean creative awards do not predict effectiveness?

These papers do not test awards directly. What they establish is that in-market sales outcomes are the measure worth using, and that recall and liking are weaker proxies. Hartnett and colleagues (2016) linked coded execution features to commercially validated sales for that reason. Any claim about which creative work performs should point to the outcome measure it was validated against.

What does this imply for AI-generated creative?

It cuts in two directions. The intuition result undermines the assumption that a human reviewer will catch weak creative, which is often the stated safeguard for AI production. It equally undermines any claim that an AI model knows what will sell, since the training signal available to it is largely the same admired work whose sales performance experts cannot predict. The reasonable conclusion is that volume raises the value of measurement, because neither party in the review is calibrated.