3. Feature-by-Feature Scrutiny (The Stated Features)
Categories:
- Table stakes (TS): customers expect it. It earns nothing, but its absence costs you.
- Differentiator (D): a real reason to choose you over the alternatives.
- Distraction (X): built but unlikely to move revenue or retention.
- Liability (L): creates support, compliance, legal or trust risk.
The primary bucket comes first.
3.1 Classification table
| # | Stated feature | Category | Reasoning |
|---|---|---|---|
| 1 | Google reviews widget on the business's website | TS | Free from Trustindex (1M+ installs), Elfsight, EmbedSocial, Featurable and 48 Wix apps. EmbedSocial says it can be built "in an afternoon" with AI. It earns nothing on its own. But how you build it decides whether it's legal: see 3.2. |
| 2 | Facebook reviews (recommendations) in the same widget | X at launch | It needs pages_read_user_content and a Page token with manage rights, probably with Meta app review (unverified). Only some local businesses have active Facebook recommendations. Add it once 30% of paying agencies ask. |
| 3 | Low cost ($5–15/month) | L as the strategy | You can't undercut free. The proof case left $5–15 for $49–149. At $5, a Places-based widget loses money (3.2), and support per dollar is the worst in SaaS. Low per-client wholesale cost for agencies is fine. Low retail price as the pitch isn't. |
| 4 | Review requests ("added over time") | D, and it should come first, not later | This is what people pay for: LocalImpact $49+, Birdeye $299+, Podium $399+. Fresh reviews matter to 74% of consumers. It has to be the same ask to every customer: no "happy? → Google, unhappy? → private form" gating. |
| 5 | Monitoring (new-review alerts, replies) | TS / D | Alerts are table stakes. AI-drafted replies, approved by a human and posted through the Business Profile API with the client's authorisation, are a modest differentiator. LocalImpact sells them. Auto-posting without approval is a Liability. |
| 6 | White-label resale through agencies | D only with the report | White-label alone is $3 per client at Repuso, and bundled in GoHighLevel. It becomes a reason to buy only with an agency-branded monthly client report and the agency's own backlink in the widget. |
| 7 | "Powered by" links in the widget (implied by the proof case) | D for the free direct plan / L on agency clients | It's the loop that built ReviewsOnMyWebsite. On agency client sites it has to be off, or be the agency's link, or agencies won't use it. |
| 8 | SEO as the main channel (implied) | X as the lead channel | "Google reviews widget" results are held by vendors with free plans and years of backlinks. Use SEO for a niche you can own (UK review rules), not the head term. |
3.2 The Google API reality (this decides the build)
| Route | Reviews you get | Storage | Cost | Verdict |
|---|---|---|---|---|
| Places API (New), Place Details | Max 5, sorted by relevance, no paging | No caching except the place ID | $25 per 1,000 after 1,000 free a month (Enterprise + Atmosphere) | Don't use it for display. One paid call per page view. 5 reviews can't be chosen. A notice on ordering and the Google logo are required. |
| Business Profile API (owner OAuth) | All reviews, paged | ≤30 days, secure, no manipulation or aggregation | No per-call fee. Quota 0 until approved, then 300 QPM | Build on this. Needs your own profile verified 60+ days, plus Google approval. Unverified: whether its "outside your project" storage rule allows showing reviews on the owner's site through your widget. Get that confirmed in writing. |
| Scraping or third-party scraper APIs | All | Whatever you like | Cheap | Liability. It breaks Google's terms. One crackdown breaks every client's widget at once, and the agency blames you in front of its clients. |
Rules that follow from this:
- Refresh reviews at most daily, and store them for ≤30 days.
- Show Google's own rating and review count. Never compute your own blended score across sources.
- Credit every review's author (avatar at minimum).
- Show the "how these reviews are chosen" note on every widget.
3.3 Scoreboard
| Category | Count | Features |
|---|---|---|
| Table stakes | 2 | Google widget, review alerts |
| Differentiator | 3 | Review requests (moved to v1), AI-drafted replies with approval, white-label with the client report and agency backlink |
| Distraction | 2 | Facebook at launch, SEO on head terms as the lead channel |
| Liability | 2 | Low retail price as the strategy, "Powered by" on agency clients' sites |
About a third of what was described is a reason to buy, and it's the part that was planned for "later".
3.4 "Building to avoid selling": explicit call-outs
- Widget layouts. Carousels, masonry, badges, dark mode and animations are fun to build and demo well. Buyers don't pay for them, since Elfsight has dozens free.
- More review sources. Trustindex claims 130+ and Repuso 100+. Adding Yelp, Tripadvisor and Trustpilot feels like progress. Each one is a new API or scraper to maintain.
- A price war. Building a cheaper Elfsight is building toward a $0 market.
3.5 What a disciplined v1 looks like
Keep:
- Google via the Business Profile API only;
- an honest-by-default widget with the notice;
- a request link, QR card, and email + SMS sequence (the same ask for everyone);
- new-review alerts;
- AI-drafted replies with approval;
- a monthly white-label client report;
- an agency dashboard (add a client in <10 minutes).
Defer: Facebook, CSV/Zapier triggers beyond the basics, AI review summaries, and a second market. Cut: star filters without disclosure, gating, incentives, scraping, Places for display, blended scores, and listings sync.