Setting the future of digital and social media marketing research: Perspectives and research propositions

This is **not an experiment**. The paper is a conceptual review and agenda-setting piece. The "intervention" is a structured elicitation of expert perspectives from 15 leading researchers in digital and social media marketing. The "outcome" is a set of research propositions, challenges, and opportunities organised across eight thematic areas:

Artificial intelligence (AI) in marketing

Augmented reality (AR) marketing

Digital content management

Mobile marketing and advertising

Business-to-business (B2B) marketing

Electronic word-of-mouth (eWOM)

Ethical issues in digital marketing

General digital and social media marketing strategy

There are no comparators, no control groups, no measured outcomes. The paper does not test any hypothesis — it proposes what should be tested.

**No human subjects were studied.** The "participants" are 15 invited experts (academics and practitioners) who contributed short perspective pieces. The paper does not provide demographic details, years of experience, or selection criteria for these experts. The authors themselves are the editorial team who synthesised these contributions.

**Setting:** Academic journal (International Journal of Information Management), 2020.

**No measurement instruments were used.** The paper does not collect or analyse data. Instead, the authors:

1. Invited experts to write short perspective articles on specific topics.

2. The editorial team reviewed, categorised, and synthesised these contributions.

3. They identified recurring themes, gaps, and propositions.

There are no scales, no surveys, no experimental protocols, no statistical analyses. The "method" is qualitative synthesis of expert opinion — essentially a structured literature review combined with invited commentary.

**Study design:** Conceptual review with invited expert perspectives. This is not a systematic review, meta-analysis, or empirical study. It is a **position paper** that aggregates expert opinion to set a research agenda.

**Key design features:**

**No randomisation:** Not applicable — no subjects were assigned to conditions.

**No blinding:** Not applicable — no treatment was administered.

**No control group:** Not applicable.

**No duration:** The paper is a snapshot of expert opinion at one point in time (2020). There is no longitudinal component.

**No statistical approach:** The paper does not report p-values, effect sizes, confidence intervals, or any quantitative analysis.

**What this design can prove:**

It can identify what leading researchers believe are the most important unanswered questions in digital marketing.

It can highlight consensus and disagreement among experts.

It can provide a structured framework for future research.

**What this design cannot prove:**

It cannot establish cause-and-effect relationships.

It cannot tell you whether any specific marketing tactic works.

It cannot provide effect sizes, success rates, or ROI estimates.

It cannot generalise beyond the opinions of the 15 invited experts.

**Major methodological weaknesses:**

**Selection bias:** The experts were chosen by the editorial team — their views may not represent the broader field.

**No systematic search:** Unlike a systematic review, the authors did not follow a reproducible search strategy for literature.

**No quality assessment:** Expert opinions are not weighted by evidence quality.

**No quantitative synthesis:** The paper cannot tell you how many studies support any given proposition.

**No preregistration:** The analysis plan was not specified in advance.

Since this is not an empirical study, there are no numerical findings. Instead, the paper presents **research propositions** — statements about what should be studied. Here are the key propositions organised by theme:

**1. Artificial Intelligence in Marketing**

Proposition: AI will fundamentally change how marketers personalise content, but research is needed on consumer trust and privacy concerns.

Proposition: The effectiveness of AI-driven chatbots versus human customer service representatives is unknown.

Proposition: Ethical frameworks for AI in marketing are underdeveloped.

**2. Augmented Reality Marketing**

Proposition: AR can enhance product trial experiences (e.g., virtual try-ons), but the effect on purchase intention versus traditional media is unquantified.

Proposition: AR may reduce return rates for online purchases, but no controlled studies exist.

**3. Digital Content Management**

Proposition: Content personalisation increases engagement, but the optimal degree of personalisation (and when it becomes "creepy") is unknown.

Proposition: User-generated content may be more persuasive than brand-generated content, but the effect size is unclear.

**4. Mobile Marketing and Advertising**

Proposition: Location-based mobile ads may increase conversion rates, but privacy concerns may offset benefits.

Proposition: Mobile ad format (video vs. static, interstitial vs. native) effects are poorly understood.

**5. B2B Marketing**

Proposition: Social media's role in B2B lead generation is under-researched compared to B2C.

Proposition: The ROI of B2B social media marketing is not established.

**6. Electronic Word-of-Mouth (eWOM)**

Proposition: Negative eWOM spreads faster and has a larger impact than positive eWOM, but the magnitude of this asymmetry is unknown.

Proposition: Fake reviews and astroturfing undermine trust, but detection methods are inadequate.

**7. Ethical Issues**

Proposition: Data privacy regulations (e.g., GDPR) affect marketing effectiveness, but the trade-offs are not quantified.

Proposition: Vulnerable populations (children, elderly) are disproportionately affected by targeted advertising.

**8. General Digital Marketing Strategy**

Proposition: Omnichannel marketing (consistent experience across platforms) outperforms single-channel approaches, but the incremental benefit is unknown.

Proposition: The optimal frequency of social media posting varies by platform and audience.

**No p-values, effect sizes, or confidence intervals are reported anywhere in the paper.**

**Not applicable.** The paper does not report any effect sizes. It is a qualitative agenda-setting piece, not an empirical study. There are no numbers to translate into plain English.

**As acknowledged by the authors:**

The paper is based on invited expert perspectives, not a systematic review.

The experts may not represent all viewpoints in the field.

The propositions are speculative and require empirical testing.

The paper does not prioritise which propositions are most important or urgent.

**As a critical reader would note:**

**No empirical data:** The paper contains zero original data. It cannot answer "what works" — only "what should we study next?"

**Selection bias:** The 15 experts were chosen by the editorial team. Their views may reflect the editors' own biases or networks.

**No conflict of interest disclosure:** Several experts may have commercial interests in the technologies they discuss (e.g., AI, AR), but this is not addressed.

**No reproducibility:** The method of synthesising expert perspectives is not described in enough detail to replicate.

**Temporal limitation:** Published in 2020, the paper predates major developments in generative AI (e.g., ChatGPT, 2022), which has dramatically changed the digital marketing landscape.

**No quantitative prioritisation:** The paper lists many propositions but does not rank them by importance, feasibility, or potential impact.

**No practical guidance for individuals:** The paper is written for academic researchers and marketing practitioners, not for individuals running personal experiments.

**For someone running their own n=1 experiment:**

This paper is **not a good source for designing a personal experiment** because it contains no empirical findings. However, the research propositions can inspire self-experiments. Here is how to translate the paper's ideas into actionable tests:

### What to test (specific intervention and dose)

Choose one proposition and test it on yourself or your small business. Examples:

**AI vs. human content:** For 2 weeks, use an AI writing tool (e.g., ChatGPT) to write social media posts. For the next 2 weeks, write posts yourself. Compare engagement (likes, shares, comments).

**Posting frequency:** Post once daily for 2 weeks, then 3 times daily for 2 weeks. Track follower growth and engagement rate.

**AR try-on:** If you sell products online, add an AR try-on feature (e.g., for glasses, furniture) for one product category. Compare conversion rate and return rate for that category vs. a control category without AR.

**Personalisation level:** Send a generic email newsletter to half your list and a personalised version (using name, past purchases) to the other half. Compare open rates and click-through rates.

### Minimum meaningful duration

**For engagement metrics:** 2 weeks per condition (4 weeks total for an A/B test).

**For conversion/sales:** At least 4 weeks per condition (8 weeks total) to account for weekly cycles.

**For return rates:** 8–12 weeks per condition, since returns happen after purchase.

### What to measure (specific metrics)

**Primary metric:** Choose one — engagement rate (likes+comments / impressions), click-through rate (clicks / impressions), conversion rate (purchases / visitors), or return rate (returns / purchases).

**Secondary metrics:** Time spent on page, bounce rate, cost per acquisition, customer satisfaction (1–5 scale).

**Confound check:** Track day of week, time of day, holidays, and major events that might affect engagement.

### Key confounds to control for

**Seasonality:** Run your experiment during a stable period (avoid holidays, product launches, or crises).

**Audience growth:** If your follower count changes during the experiment, normalise metrics per follower.

**Platform algorithm changes:** Social media algorithms change frequently. Note any platform updates during your experiment.

**Content quality:** If testing frequency, keep content quality constant (same images, same writing style).

**Your own behaviour:** Don't change other marketing activities (e.g., email, ads) during the experiment.

**Time of day:** Post at the same time each day, or randomise posting times across conditions.

### What a positive result would look like

**For AI vs. human content:** AI-generated posts achieve ≥10% higher engagement rate (e.g., from 2% to 2.2%) with no increase in negative comments.

**For posting frequency:** 3x daily posting yields ≥15% higher total engagement per week than 1x daily, without increasing unfollow rate.

**For AR try-on:** Products with AR have ≥20% higher conversion rate and ≥15% lower return rate than products without AR.

**For personalisation:** Personalised emails have ≥25% higher open rate and ≥15% higher click-through rate than generic emails.

**Important caveat:** These thresholds are guesses based on general marketing benchmarks, not from this paper. The paper provides no effect size estimates. Your results may differ dramatically based on your audience, product, and platform.

**Bottom line:** Use this paper to generate hypotheses for your own experiments, but do not rely on it for effect size estimates, best practices, or evidence-based recommendations. For that, you need empirical studies — ideally randomised controlled trials or meta-analyses — not expert opinion pieces.