How creative affects advertising effectiveness.

Contents

1. About this review
2. Where the “half of effectiveness” claim comes from
3. What academic meta-analyses establish
4. Reading the evidence visually
5. Field studies and industry evidence
6. Digital formats and agency research
7. Why there is no pooled percentage
8. Practical interpretation and limitations
9. References

About this review

This is an extended English summary of the Russian review. The complete article, methodological appendix and downloadable PDF and Word documents are in Russian. The review compares peer-reviewed research, published meta-analyses and international industry reports; it does not calculate a new pooled statistical effect.

Creative execution matters substantially, but the evidence does not establish a universal rule that creative determines 50% of advertising effectiveness. Consumer attitudes, brand memory, purchase intentions, incremental sales and ROI are different outcomes. Similar-looking numbers cannot be combined without examining what each measures.

Where the “half of effectiveness” claim comes from

The clearest verified attribution is Google → Nielsen/NCS 2017. Rodrigo Carone’s 2022 Google article refers to creative delivering nearly 50% of ROI and cites Nielsen Catalina Solutions, US, 2017 [1–2]. The original Nielsen decomposition assigns 47% to creative, 22% to reach, 15% to brand factors, 9% to targeting, 5% to recency and 2% to context. The database covers almost 500 FMCG campaigns from 2016 and the first quarter of 2017. This concerns advertising’s sales contribution, not all company sales; sales contribution and ROI are not interchangeable.

The 2023 NCS update uses almost 450 TV and digital CPG campaigns and 18 model features. It reports 49% for a creative block that includes message and context; it also gives 49% for 2017 under a comparable grouping [3–4]. The change from 47 to 49 is therefore not evidence of an increase in creative’s importance. Public information is insufficient to reproduce the full estimation procedure and its statistical uncertainty.

Ipsos/ARF’s widely quoted 75% relates to ad/brand recall, while Kantar’s 49% concerns brand salience [5–7]. The underlying Ipsos/Admap 2014 article was not independently checked in the review. These figures neither replicate the NCS sales estimate nor establish shares of profit.

What academic meta-analyses establish

Rosengren, Eisend, Koslow and Dahlen pooled 878 effects from 93 datasets in 67 publications [8]. No sales data were included. Their reliability-corrected correlations show the clearest relationships with attitudes towards the ad and the brand. Creativity is understood as originality together with appropriateness. Most outcomes display heterogeneity; a multilevel model addresses dependent effects, while the publication-bias analysis uses fail-safe N.

The last two correlations are not statistically significant in the original table. A correlation of 0.491 does not mean “49.1% of effectiveness”; squaring it would not establish a causal share of sales either. Liking an ad does not guarantee identifying its brand, and intending to buy is not the same as making a purchase.

Darley and Lim analyse 48 publications and 298 data points [9]. Their publisher abstract supports positive relationships with cognitive, emotional and behavioural intentions, with differences by media modality. The paid full text was not assessed. The two meta-analyses overlap in their underlying literature and must not be counted as 115 independent studies.

Reading the evidence visually

Ad attitude0.491

Brand attitude0.317

Purchase intention0.306

Ad recall0.311

0.1.2.3.4.5.6

Correlation r and 95% confidence interval. Scale 0–0.6. Rosengren et al. [8]. Not a share of sales.

Dots show published mean correlations; lines show 95% confidence intervals. Four estimates are reproduced from Rosengren et al. [8], Table 2 and the results text. This is not a new meta-analysis, and different outcomes are not pooled.

47%Advertising sales contributionNielsen/NCS [2]

75%Ad / brand recallIpsos / ARF [6]

49%Brand salienceKantar [7]

These estimates cannot be added, averaged or compared as shares of the same outcome.

These figures describe different outcomes [2, 6–7]. They cannot be ranked as estimates of a single effectiveness measure, added or averaged.

Field studies and industry evidence

Hartnett and colleagues examine 312 TV ads across categories and countries, relating more than 150 execution elements to short-term sales effectiveness [10]. Individual devices offer limited explanatory power, with no universal recipe. Mars funded the research and supplied the data. Williams and colleagues reanalyse the same ads using 158 variables and several analytical methods [11]. The timing of visual branding is among the more robust findings; this is a sensitivity check on the same sample, not an independent replication.

For online display, Braun and Moe model multiple creatives and individual exposure histories [12]. In their simulation, changing the creative sequence increases expected visits by 12.7% and conversions by 13.8%. These are modelled improvements on specific data, not an average randomised treatment effect. Chen and colleagues’ 2 × 2 × 3 experiment examines originality, relevance and repetition [13]. Originality combined with relevance better resists wear-out under the studied conditions, but does not establish a universal refresh schedule.

Thinkbox/accelero describe a 12-fold ROI range based on 7,000 UK campaigns [14]. This represents potential variation, not the expected benefit of redesigning an ad. IPA’s review of almost 600 cases from 1996–2018 finds declining advantages for creatively awarded campaigns, including no advantage in the later part of the data [15]. Awards are therefore not a guaranteed commercial premium.

DMA reports four times as many business effects for highly creative versus less creative campaigns; the full database exceeds 1,500 campaigns and the quartile comparison covers 2021–2023 [16]. This counts reported effects, not four times the sales. Ipsos tests ten award-winning ads and reports average short-term and long-term potential indices of 129 and 111, with 30% higher brand attention [17]. These are pre-test measures, not an observed 29% sales increase.

Large industry databases broaden the commercial evidence, but sample size does not eliminate selection bias, differences in units of analysis or limited causal identification. Research suppliers, agencies and media companies have commercial interests; this does not invalidate their findings, but increases the importance of transparent samples, outcome definitions and methods.

Digital formats and agency research

Dentsu’s 2024 Japanese study covers over 8,000 mobile users, more than 45 creatives and seven formats [18]. Recall is 21% after less than one second of viewing and 41% after more than ten seconds. This associates attention duration with memory; it does not show that forcing longer viewing doubles sales. Dentsu MENA also stresses early brand signals [19]. Neither source establishes Dentsu as the origin of a universal 50% claim.

WPP Media, System1 and TikTok report 1,217 creator ads, eight markets, 182,550 users and comparisons with more than 350 brand-lift studies [20]. Creative Quality, combining emotion and early brand recognition, is attributed 54.6% of Brand Memory Lift. The outcome combines awareness and recall with 2:1 weights; it is not sales. The full gated report was not obtained.

Gain Theory’s 2025 report attributes 50–70% of sales lift to creative [21], but does not disclose enough information alongside that estimate to reproduce this decomposition, including campaign counts and standard errors. Its 2026 press release repeats the estimate; the CMO survey is not an independent replication of creative’s sales effect [22].

Why there is no pooled percentage

Modelled contribution shares, correlations, relative gains, pre-test indices and ratios of extreme ROI values do not share one estimand. Averaging 47%, 75% and a 12-fold multiplier would be meaningless; 47–75% is not a confidence interval. Pooling also requires variances and independent samples, which are not available consistently. Respondents are nested within ads, campaigns and brands; impressions cannot simply be treated as independent observations.

Sources are dependent: Google cites NCS; the 2023 NCS report and release concern one analysis; Williams reuses Hartnett’s data; the academic meta-analyses overlap. Causal interpretation is further limited by differences in brand strength, offers, production resources, media pressure and algorithmic audience selection.

These are qualitative judgements made in the review, not formal GRADE ratings. A new meta-analysis would require primary effects, variances, shared-sample identifiers, bias assessment and models that account for clustering and dependence.

Practical interpretation and limitations

The most consistent interpretation is that creative changes the productivity of an advertising contact. Media determine the opportunity, cost, frequency and context; the message shapes what the consumer takes from it. This is a conceptual interpretation, not a newly estimated model. It does not justify neglecting reach or assuming that more impressions can repair a weak message.

Define the business outcome first: incremental revenue, contribution profit or a prespecified brand-lift measure. Compare meaningfully different concepts under comparable offers, audiences, placements and frequencies, preferably with random allocation. An A/B comparison determines which execution performs better; estimating advertising’s total incremental effect requires a no-ad control or a suitable geographic experiment.

Specify the primary metric, meaningful effect size, duration and stopping rules before testing. Account for multiple comparisons and validate a winner on new data. Consider the full cost of production, rights, adaptations, testing and media. A factor’s share of explained performance is not its optimal budget share. Numerical estimates require local validation before being applied to Russian campaigns.

Limitations include English-language web search without comprehensive bibliographic database coverage, no independent second coder, partial access to paid papers and proprietary data, and no complete map of overlapping primary samples. The review has not undergone independent peer review. It supports creative’s substantial role, especially for consumer attitudes, but does not establish a universal commercial percentage or production-budget rule.

References

The 22 source records below are preserved from the original review. Access notes are in Russian. The review was written on 30 October 2026; the full Russian article includes the search-query appendix and source-dependency register.

1. Carone, R. (2022). YouTube advertisers, focus on these 4 key drivers of ROI. Think with Google. Июнь; раздел Creative, сноска 1.Полный веб-текст; цепочка цитирования.Open source ↗
2. Nielsen (2017). When it comes to advertising effectiveness, what is key? Октябрь. Nielsen / Nielsen Catalina Solutions.Полный веб-текст и диаграмма; публичное описание исследования.Open source ↗
3. NCSolutions (2023). The five keys to advertising effectiveness are creative, brand, targeting, reach and recency. Публикация сохранена на сайте Circana.Публичное описание; раздел About the study.Open source ↗
4. NCSolutions (2023). Five Keys to Advertising Effectiveness. Updated Insights for the Modern CPG Marketer. С. 3–4.Полный PDF, 11 страниц; копия отчёта NCS на MediaPost.Open source ↗
5. Ipsos (2008). Lessons Learned: Advertising Creative and Your Brand.Полный веб-текст; исходная методика не раскрыта.Open source ↗
6. Snyder, J. (2016). How Advertising Works, Today. Advertising Research Foundation, RE:THINK. Слайд 25 со ссылкой на Ipsos / Admap 2014.Полная презентация; вторичный след оценки Ipsos, не исходное исследование.Open source ↗
7. Morley, N., & Morrell, S. (б. д.). How to align your innovation and advertising around meaningful difference. Kantar. Раздел Support, support, support.Полный веб-текст; результат Crossmedia, без полного протокола.Open source ↗
8. Rosengren, S., Eisend, M., Koslow, S., & Dahlen, M. (2020). A Meta-Analysis of When and How Advertising Creativity Works. Journal of Marketing, 84(6), 39–56. DOI 10.1177/0022242920929288.Полный HTML; таблица 2 и текст результатов проверены.Open source ↗
9. Darley, W. K., & Lim, J.-S. (2023; online 2022). Advertising creativity and its effects: a meta-analysis of the moderating role of modality. Marketing Letters, 34, 99–111. DOI 10.1007/s11002-022-09627-6.Издательская аннотация и библиография; полный текст платный.Open source ↗
10. Hartnett, N., Kennedy, R., Sharp, B., & Greenacre, L. (2016; online 2015). Creative That Sells: How Advertising Execution Affects Sales. Journal of Advertising, 45(1), 102–112. DOI 10.1080/00913367.2015.1077491.Издательская аннотация и сведения о финансировании.Open source ↗
11. Williams, J., Hartnett, N., & Trinh, G. (2023). Finding creative drivers of advertising effectiveness with modern data analysis. International Journal of Market Research, 65(4), 423–447. DOI 10.1177/14707853221134258.Университетская запись с аннотацией; общий массив с [10].Open source ↗
12. Braun, M., & Moe, W. W. (2013). Online Display Advertising: Modeling the Effects of Multiple Creatives and Individual Impression Histories. Marketing Science, 32(5), 753–767. DOI 10.1287/mksc.2013.0802.Издательская аннотация; численные результаты симуляции.Open source ↗
13. Chen, J., Yang, X., & Smith, R. E. (2016; online 2014). The effects of creativity on advertising wear-in and wear-out. Journal of the Academy of Marketing Science, 44, 334–349. DOI 10.1007/s11747-014-0414-5.Издательская аннотация; полный текст платный.Open source ↗
14. Thinkbox / accelero (2023). The drivers of profitability. 18 июля; страница обновлена 23 октября 2026. Исследование Paul Dyson.Полный веб-разбор; раздел Creativity.Open source ↗
15. Field, P. (2019). The Crisis in Creative Effectiveness. Institute of Practitioners in Advertising.Официальное резюме; полный отчёт не анализировался.Open source ↗
16. Data & Marketing Association (2024). The Value of Creativity. С. 14; см. также официальное описание Value of Creativity Report.PDF и официальная страница dma.org.uk; сравнение квартилей, не денежный lift.Open source ↗
17. Ipsos (2023). Can Award Winning Ads be Effective? 22 сентября.Полный веб-текст; Creative|Spark, 10 объявлений.Open source ↗
18. Dentsu Japan International Brands (2024). Introducing the Attention Economy, an Attention Research Report Based on Eye Tracking Data of Ad Viewing. Пресс-релиз, 28 октября.Официальный веб-релиз с методологией и результатами.Open source ↗
19. Dentsu (2024). Attention Economy 2024 Report. MENA.Официальное открытое резюме.Open source ↗
20. WPP Media, System1, & TikTok (2026). The Creator Effectiveness Playbook.Открытые Methodology и FAQ; полный gated report не получен.Open source ↗
21. Gain Theory (2025). Creative Effectiveness Decoded: A Guide to Data-Informed Creative Impact. С. 7.Полный PDF, 27 страниц; оценка 50–70% без достаточных данных для репликации.Open source ↗
22. Gain Theory (2026). Gain Theory Launches Creative Effectiveness CMO Study: 62% Struggle to Prove the Value of Their Creative. 30 июня.Полный веб-релиз; повторение [21], не независимая оценка sales lift.Open source ↗

← All posts · Talks and publications