The Science of Online Reviews and Sales

Online reviews are no longer a supplement to a business's reputation. For most local service providers, they are the primary mechanism by which that reputation is formed. But beyond intuition and marketing benchmarks, a body of rigorous academic research measures the causal effect of reviews on revenue and purchasing decisions with statistical precision.

This article examines four real, verified studies, explains what each found and where its limits lie, and draws the practical implications for home-service and construction businesses operating in local US markets.

---

Why Academic Research Matters Here

Any marketing agency can claim "reviews increase sales." What separates that claim from a useful conclusion is methodology: how were causal effects isolated, what biases were controlled for, and how generalizable are the findings?

The four studies presented here share one important methodological characteristic: none of them relies on simple correlation. Each uses a design that supports stronger causal claims than an opinion survey or a cross-sectional dataset alone.

---

Study 1: The Revenue Effect of One Extra Star

Luca, M. (2011, revised 2016). "Reviews, Reputation, and Revenue: The Case of Yelp.com." Harvard Business School Working Paper, No. 12-016.

HBS Faculty Page | Full paper on SSRN

What He Did

Michael Luca at Harvard Business School merged Yelp review data with restaurant revenue records from the Washington State Department of Revenue. To solve the causality problem — good restaurants attract both more reviews and more customers, contaminating a simple correlation — he used a regression discontinuity design.

The mechanism: Yelp rounds each restaurant's average rating to the nearest half-star. A restaurant with a true average of 3.24 displays 3 stars; one with 3.26 displays 3.5 stars. That 0.02-point difference in the underlying rating reflects no actual change in quality, but it creates a visible discontinuity in the public-facing score. Luca exploits this threshold to identify the causal effect of the displayed rating on revenue.

Key Finding

A one-star increase in a Yelp rating produces a 5-9% increase in revenue for independent restaurants.

The effect is driven entirely by independent businesses. Chain restaurants show no statistically significant effect, which suggests that reviews substitute for brand reputation when no established brand exists.

Limits to Keep in Mind

The study focuses on restaurants in the United States. Transfer to other sectors and service types is plausible but not directly demonstrated. The data also covers the early years of Yelp, when platform penetration was lower; effects in today's higher-saturation environment may differ in magnitude.

---

Study 2: Negative Reviews Hurt More Than Positive Ones Help

Chevalier, J. and Mayzlin, D. (2006). "The Effect of Word of Mouth on Sales: Online Book Reviews." Journal of Marketing Research, 43(3), 345-354.

Journal of Marketing Research | NBER Working Paper 10148

What They Did

Chevalier and Mayzlin collected review and relative sales data for books on Amazon.com and BarnesandNoble.com at three points in time. They used a differences-in-differences approach: if a book receives a one-star review on bn.com but not on Amazon, its relative sales at bn.com should fall. This design controls for factors that affect both platforms equally — actual book quality, national advertising — and isolates the specific effect of reviews on each site.

Key Findings

• An improvement in a book's average rating on one platform increases its relative sales share at that platform.
• The sales impact of one-star reviews is greater in magnitude than the impact of five-star reviews on relative sales.
• Consumers read review text, not just the aggregate score: review length has a statistically significant effect on Amazon sales, suggesting people engage with the content itself.

Limits to Keep in Mind

Books are an experience good with high homogeneity across platforms. In home services, where the "product" differs across providers, effects may be larger because uncertainty about service quality before hiring is substantially higher than before purchasing a book.

---

Study 3: Half a Star More Fills Tables at Peak Hours

Anderson, M. and Magruder, J. (2012). "Learning from the Crowd: Regression Discontinuity Estimates of the Effects of an Online Review Database." The Economic Journal, 122(563), 957-989.

Full PDF (UC Berkeley) | EconPapers record

What They Did

Anderson and Magruder analyzed 148,000 Yelp reviews for 328 restaurants in the San Francisco area and matched that data to real table reservation availability during peak dining hours (Thursday, Friday, and Saturday evenings), queried 36 hours in advance. They applied the same regression discontinuity design as Luca, exploiting Yelp's half-star rounding threshold to create quasi-random variation in the displayed rating.

Key Findings

An extra half-star on Yelp reduces peak-hour table availability by 19 percentage points — meaning restaurants sell out more frequently during high-demand slots.

The effect amplifies significantly when alternative information is scarce:

• Restaurants without external accreditation (press rankings, Michelin recognition): 27 percentage point increase in peak-hour sell-out rate.
• Restaurants with Michelin recognition or San Francisco Chronicle ranking: effect is statistically insignificant.

The jump from 3 to 3.5 stars raises the probability of selling out during peak hours from 13% to 34%. The jump from 3.5 to 4 stars adds another 19 percentage points.

Limits to Keep in Mind

The study focuses on San Francisco, a market with unusually high Yelp user density and restaurant competition. In markets with lower platform penetration, effects may be smaller in magnitude but are unlikely to reverse in direction.

---

Study 4: A Meta-Analysis Across 26 Studies and 443 Elasticities

Floyd, K., Freling, R., Alhoqail, S., Cho, H.Y., and Freling, T. (2014). "How Online Product Reviews Affect Retail Sales: A Meta-analysis." Journal of Retailing, 90(2), 217-232.

ScienceDirect | DOI: 10.1016/j.jretai.2014.04.004

What They Did

Floyd et al. synthesized 26 prior empirical studies that together produced 443 sales elasticity measurements, using established meta-analytic methods to identify patterns that hold across different markets, products, and review platforms. This is the most generalizable source of the four because it does not depend on a single context.

Key Findings

Dimension	Mean Elasticity	Interpretation
Review valence (star rating quality)	Es = 0.69	High: the average score meaningfully moves sales
Review volume (number of reviews)	Es = 0.35	Moderate: more reviews correlates with more sales
Third-party platforms vs. own website	Higher for third-party	External platforms are perceived as more credible
High-involvement services	Higher than low-involvement	Home services fall firmly in this category

Review valence has roughly double the effect on sales compared to review volume, but both are statistically significant and practically relevant across the studies surveyed.

Sales elasticities are higher for independent platforms (Google, Yelp, Trustpilot) than for reviews hosted on a business's own website — likely because consumers perceive less opportunity for editorial manipulation on a third-party platform.

Limits to Keep in Mind

The 26 base studies were published primarily between 2005 and 2013 and focus mainly on e-commerce of tangible goods. Generalization to high-trust local services like remodeling or installation requires methodological caution. The direction of effects is well-supported; the exact magnitude should not be transferred without adjusting for sector and local context.

---

Comparative Summary: The Four Studies

Study	Authors	Year	Key Finding	Practical Implication
HBS Working Paper 12-016	Luca	2016	+1 star = +5-9% revenue (independents)	Every rating point has direct economic value
Journal of Marketing Research	Chevalier & Mayzlin	2006	1-star reviews outweigh 5-star reviews in sales impact	Managing negatives is as urgent as collecting positives
The Economic Journal	Anderson & Magruder	2012	+0.5 stars = +19 pp peak-hour sell-out rate	Rating thresholds matter as much as the average
Journal of Retailing	Floyd et al.	2014	Valence Es = 0.69; Volume Es = 0.35	Review quality has double the effect of review count

---

What Research Says About Biases in the System

Understanding that reviews affect sales is not enough. The research also documents systematic biases that distort the information available to consumers and to business owners interpreting their own profiles.

Positivity Bias in Review Distribution

Chevalier and Mayzlin found that reviews tend to be positive on average. Consumers who leave reviews are disproportionately those who had an extreme experience — very good or very bad — not those who had an average one. This produces a bimodal distribution, not a normal bell curve.

Implication: a business with many five-star reviews and few three- or four-star reviews may appear more polarizing than it actually is. Consumers who understand this adjust; those who do not take the raw average at face value.

Consumers Do Not Process All Available Information

Luca found that consumers are more sensitive to rating changes that are more visible and respond more when a profile has more reviews, but they do not use all available information — for example, they do not adjust for the reviewer's Yelp social network size, even though that information is displayed. This means the average rating and total review count are the two signals that carry the most decision weight in practice.

The Effect Is Asymmetric Across Business Types

Both Luca and Anderson and Magruder document that the review effect is larger for businesses without pre-established brand reputation. For home-service businesses in the US, this is directly relevant: an independent electrician or local remodeling company is structurally dependent on its reviews in a way that a national franchise is not. Reviews fill the trust gap that brand recognition would otherwise occupy.

---

Practical Applications for Home-Service Businesses

The research does not speak directly to contractors in US local markets, but the underlying mechanisms apply with few adaptations.

1. Rating Thresholds Matter More Than Marginal Incremental Gains

Anderson and Magruder demonstrate that the jumps from 3 to 3.5 stars and from 3.5 to 4 stars produce large, discrete effects on customer flow. Moving from 4.1 to 4.3 may matter less than moving from 3.7 to 4.0. Focus energy on crossing the next visible threshold.

See how this plays into your overall local visibility strategy.

2. Reviews Do Not Just Attract Leads — They Convert Them

Lead-to-customer conversion is strongly correlated with perceived trust before the first call is made. Reviews are the primary trust signal for a service the consumer cannot evaluate until after the job is done. A strong review profile reduces decision friction at exactly the moment when a potential customer is comparing providers.

3. A One-Star Review Without a Response Is the Worst-Case Scenario

Chevalier and Mayzlin document that the negative sales impact of a one-star review exceeds the positive impact of a five-star review. A public business response moderates that effect by demonstrating professionalism and willingness to resolve problems. Silence amplifies the damage.

For a step-by-step guide to building and managing your review profile, see Google Reviews for Home Service Businesses.

4. Automate Post-Service Review Requests

If the revenue effect of reviews is economically significant, it makes sense to systematize how you generate them. A consistent post-job follow-up sequence — SMS, email, or WhatsApp — removes the friction of asking manually after every completed appointment and ensures the request reaches the customer while the experience is still fresh.

For plumbing businesses specifically, see the complete approach in Marketing for Plumbers: USA Complete Guide.

5. Prioritize Third-Party Platforms Over On-Site Testimonials

Floyd et al. document that third-party platforms carry higher sales elasticities than reviews on a business's own website. For US home-service businesses, Google Business Profile is the primary channel because it also determines visibility in local search results and Google Maps — the starting point for most local service searches.

Explore industry-specific strategies across all industries we serve, or browse the marketing glossary to build fluency with the underlying concepts. The blog covers additional guides across visibility and conversion.

---

What This Research Does Not Say

Honesty about limits matters here:

• None of the four studies covers home-service or construction businesses directly. The transfer relies on shared mechanisms — information asymmetry, trust signaling, social proof — not direct measurement of contractors or remodelers.
• The Yelp-based studies are US-specific and from a period when Yelp was more dominant in local search. Google dominates most local service categories today, and the distribution of effects across platforms may differ.
• All four studies predate 2016. The review ecosystem has evolved substantially: higher volume of reviews overall, more AI-mediated discovery (Google AI Overviews, Perplexity), and a continued shift toward Google as the dominant local review platform.
• The exact magnitude of effects should not be applied to a specific business without accounting for local competitive context, service category, and current platform penetration in the relevant market.

What is generalizable is the direction of effects and the logic of the mechanisms: reviews reduce consumer uncertainty, substitute for brand reputation when none exists, and negative signals consistently outweigh positive ones in their influence on decisions under uncertainty.

---

Conclusion

The academic literature on online reviews is not large compared to other marketing topics, but what exists is methodologically strong. The four studies examined here converge on a coherent conclusion: reviews have measurable causal effects on revenue and demand, with greater intensity for businesses that lack established brand recognition.

For an independent plumber, electrician, roofer, or contractor in the US, that translates into a clear strategic priority: active review management is not a community-relations task — it is a revenue lever with quantitative evidence behind it.

Explore how to act on this evidence through the visibility pillar, the conversion pillar, and the full blog.