HelloPeter Scraper

Scrape HelloPeter, South Africa's largest review platform, into two datasets: business profiles with TrustIndex, star ratings, industry rankings, NPS and response metrics, plus individual consumer reviews with text, ratings, verified-purchase flags and business replies.

Built for local-market research, competitor review monitoring, brand reputation tracking, customer-experience analytics and South African market entry.

Two datasets: BUSINESSES and REVIEWS, linked by businessSlug.

What you get

TrustIndex — the metric South African buyers actually shop on

trustIndex, trustIndexText, starRating, starDistribution, totalReviewCount, nps. TrustIndex is HelloPeter's proprietary composite score and it is the number South African consumers use to choose between otherwise similar local businesses. It is not recomputable from star rating and review count, so you cannot substitute one for the other.

starDistribution gives you the full 5→1 breakdown rather than just the average, and reviewHistory carries the score's movement over time — that pair is what turns "this business is well rated" into "this business is improving or declining".

Consumer-experience metrics, not just stars

responseTime, negativeReplyTime, negativeResponseRate. These are the operational half of a review platform, and they are the differentiator: how fast a business answers, and — critically — how fast it answers its negative reviews. A business with a 3.2 TrustIndex that replies to criticism in 4 hours is operationally very different from one that ignores it for three weeks, and only these fields separate them.

Industry leaderboards as a competitive set

kind: "industryListing" rows carry industry, industrySlug, industryRank, subCategory, subCategoryRank and rankings[]. Instead of guessing who your competitors are, you walk HelloPeter's own category leaderboard and get a ranked, verified competitive set per industry — including where a business sits in its subcategory.

AI-generated business summaries

aiSummary, aiThemes, keywords, popularMentions. HelloPeter generates its own AI summary and theme extraction for each profile, and aiThemes arrives as structured objects. Useful as a ready-made thematic digest of hundreds of reviews — but read Legal & Compliance before republishing platform-generated summaries.

Full contact blocks from profiles

contact (address, phone, email, website), website, logo. For a local business directory or an outreach list this is the payload, and it comes from the profile page rather than from guesswork.

Reviews with the metadata that makes them analysable

authorName, authorId, authorMemberSince, verifiedPurchase, rating, title, text, date, dateIso, helpfulCount, replyCount, businessReply, businessReplyDate. verifiedPurchase separates genuine customer experience from unattributed noise, authorMemberSince helps you discount sockpuppet patterns, and businessReply lets you measure whether a business actually responds to complaints.

How it works

HelloPeter is a JavaScript application, so the Actor runs a CheerioCrawler that fetches HTML and parses the embedded payload — no JavaScript execution, no headless browser, no captcha solving.

Only sitemap-declared and public paths are touched

Every incoming URL is classified before anything is fetched, and disallowed categories are skipped by design:

Path	Treatment
/sitemaps/**	Allowed — sitemap discovery, explicitly declared in robots.txt
/{business-slug}	Allowed — public business profile page
/{business-slug}/reviews/{id}	Allowed — public review page
/industries/**	Allowed — industry leaderboard pages
/profile/*, /sign*, /admin*, /private*, /terms*, /about*, /search*, /	Skipped — never fetched

That skip list is enforced in code, not just documented. In particular the Actor never requests the robots-disallowed /profile/ path — business data always comes from public root URLs.

The one exception you should know about

searchQueries is the only input that touches /api/search, which HelloPeter's robots.txt disallows under /api/. There are compliant alternatives that do the same job:

• startUrls with direct business URLs,
• businessSlugs for a known list of businesses,
• industrySlugs to enumerate businesses by category,
• useSitemapDiscovery: true for bulk discovery from the published sitemap index.

Prefer those. Use searchQueries only if you accept the extra constraint, and see Legal & Compliance.

Where reviews come from

HelloPeter publishes weekly review sitemaps under /sitemaps/reviews/. One weekly sitemap contains exactly the reviews posted that week. The Actor reads up to sitemap-index.xml → filters review sitemaps → scans the 6 most recent, groups review permalinks by business slug and queues them newest first.

So reviews arrive recency-first, which is exactly what you want for monitoring — and there is no pagination to walk, because the weekly sitemap already tells you what is new.

Budget note: each extra review sitemap scanned costs one request, and maxReviewsPerBusiness (default 28) caps the total per business across profile page plus sitemap.

Discovery requests don't consume the crawl budget

Sitemap and search fetches run through a separate lightweight got-scraping client rather than the crawler, so maxRequestsPerCrawl (default 200) is spent on profile and review pages — the pages that produce data — not on XML.

truncatedFields records any field the page shortened server-side, so you can tell "HelloPeter doesn't publish this" apart from "our fetch cut it off".

Input

Field	Type	Default	Notes
startUrls	string[]	[]	Any mix of business, industry, review, sitemap or search URLs. Auto-classified.
searchQueries	string[]	[]	Hits /api/search, which robots.txt disallows. Compliant alternatives below.
businessSlugs	string[]	[]	Known business slugs, e.g. capitec-banking. Direct profile fetch.
industrySlugs	string[]	[]	Industry slugs to enumerate as leaderboard pages.
useSitemapDiscovery	boolean	false	Bulk discovery from /sitemaps/sitemap-index.xml. The compliant way to go broad.
sitemapMaxSitemaps	integer	1	How many business sitemaps to follow per run.
maxBusinesses	integer	20	Cap on business records.
includeReviews	boolean	true	Pull reviews as well as profiles.
maxReviewsPerBusiness	integer	28	Total reviews per business, profile page + sitemap combined.
maxIndustryPages	integer	1	Cap on industry leaderboard pages.
maxRequestsPerCrawl	integer	200	Crawl budget for profile and review pages.
maxConcurrency	integer	8	Parallel requests.
requestHandlerTimeoutSecs	integer	60	Per-request handler timeout.
proxy	object	—	Standard Apify proxy picker.

Compliant discovery, instead of searchQueries:

{
  "useSitemapDiscovery": true,
  "sitemapMaxSitemaps": 2,
  "maxBusinesses": 100
}

Specific businesses:

{
  "businessSlugs": ["capitec-banking", "nedbank"],
  "includeReviews": true,
  "maxReviewsPerBusiness": 50
}

An industry competitive set with reviews:

{
  "industrySlugs": ["banks", "insurance"],
  "maxIndustryPages": 2,
  "maxBusinesses": 60,
  "maxReviewsPerBusiness": 30
}

Output

Two datasets, cleanly separated.

BUSINESSES — 36 fields

Discriminate on kind: business (full profile), industryListing (leaderboard row), searchResult (API search hit), error (diagnostic row with no work done).

Group	Fields
Discriminator	kind, message
Identity	businessName, businessSlug, businessId, profileUrl, website, logo
Content	about, aiSummary, aiThemes, keywords, popularMentions, faq
Contact	contact (address, phone, email, website)
Ratings	trustIndex, trustIndexText, starRating, starDistribution, totalReviewCount, reviewCountEmbedded, reviewCountLast12Months, reviewHistory
Service	nps, status, responseTime, negativeReplyTime, negativeResponseRate
Classification	industry, industrySlug, industryRank, subCategory, subCategoryRank, rankings
Hygiene	truncatedFields, scrapedAt

Views: Business overview.

REVIEWS — 22 fields

kind, id, businessName, businessSlug, businessUrl, industry, authorName, authorId, authorMemberSince, verifiedPurchase, rating, title, text, date, dateIso, helpfulCount, replyCount, businessReply, businessReplyDate, url, source, scrapedAt

Views: Review overview, Review text.

Field notes that change how you read the data:

• Join the datasets on businessSlug, not businessName — names repeat and change.
• kind: "error" rows exist so a failed fetch is visible instead of silently missing. Filter them out before analysis; message says why.
• reviewCountEmbedded vs totalReviewCount are two sources of the same thing. The embedded count comes from the profile page; the total may come from a richer path. When they disagree, the profile page's own number is usually the one shown to users.
• reviewCountLast12Months isolates recent activity — the field that distinguishes a business with 3,000 lifetime reviews from one actually being reviewed now.
• businessReply is deliberately untyped (object, string, or null) because its shape varies. Handle all three cases.
• date is the human-readable date; dateIso is the one to sort and filter on.
• aiSummary/aiThemes are HelloPeter's own generated content, not verbatim review data.
• truncatedFields lists fields the server shortened — check it before trusting a suspiciously short about.

Example input

Compliant broad discovery:

{
  "useSitemapDiscovery": true,
  "sitemapMaxSitemaps": 2,
  "maxBusinesses": 100,
  "maxRequestsPerCrawl": 400
}

Track specific businesses and their latest reviews:

{
  "businessSlugs": ["capitec-banking", "standard-bank", "sasol"],
  "includeReviews": true,
  "maxReviewsPerBusiness": 50,
  "maxRequestsPerCrawl": 300
}

Competitive set from an industry leaderboard:

{
  "industrySlugs": ["banks", "airlines"],
  "maxIndustryPages": 3,
  "maxBusinesses": 80,
  "maxReviewsPerBusiness": 25
}

Direct URLs, mixed with reviews off:

{
  "startUrls": [
    "https://www.hellopeter.com/capitec-banking",
    "https://www.hellopeter.com/industries/banks"
  ],
  "includeReviews": false,
  "maxBusinesses": 50
}

Example output

{
  "kind": "business",
  "message": null,
  "businessName": "Capitec Business",
  "businessSlug": "capitec-business",
  "businessId": "biz_8f21c4a90",
  "profileUrl": "https://www.hellopeter.com/capitec-business",
  "website": "https://www.capitec.co.za",
  "logo": "https://www.hellopeter.com/assets/logos/capitec-business.png",
  "about": "Capitec Business provides business banking solutions to South African SMEs, including business accounts, credit facilities, vehicle finance and merchant services.",
  "aiSummary": "Reviewers frequently praise fast account opening and helpful relationship managers, with recurring complaints about long hold times on the phone line and proactive fee changes.",
  "aiThemes": [
    { "theme": "Service speed", "sentiment": "positive", "mentions": 184 },
    { "theme": "Fees", "sentiment": "negative", "mentions": 97 }
  ],
  "keywords": ["business banking", "small business", "credit"],
  "popularMentions": ["relationship manager", "account opening", "vehicle finance"],
  "faq": [
    { "question": "Does Capitec Business offer a business account?", "answer": "Yes — a free business account with no monthly fee." }
  ],
  "contact": {
    "address": "15 Simmonds Street, Cape Town, 8001",
    "phone": "021 555 1234",
    "email": "business@capitec.co.za",
    "website": "https://www.capitec.co.za"
  },
  "trustIndex": 4.2,
  "trustIndexText": "Excellent",
  "starRating": 4.2,
  "starDistribution": { "5": 1284, "4": 402, "3": 118, "2": 47, "1": 203 },
  "totalReviewCount": 2054,
  "reviewCountEmbedded": 2054,
  "reviewCountLast12Months": 611,
  "reviewHistory": [
    { "month": "2026-09", "trustIndex": 4.1, "reviewCount": 58 },
    { "month": "2026-10", "trustIndex": 4.2, "reviewCount": 63 }
  ],
  "nps": 62,
  "status": "Claimed",
  "responseTime": "2 hours",
  "negativeReplyTime": "4 hours",
  "negativeResponseRate": 0.94,
  "industry": "Banks",
  "industrySlug": "banks",
  "industryRank": 3,
  "subCategory": "Business Banking",
  "subCategoryRank": 1,
  "rankings": [
    { "name": "Capitec Business", "rank": 1, "trustIndex": 4.2, "reviewCount": 2054 }
  ],
  "truncatedFields": [],
  "scrapedAt": "2026-10-04T06:12:41.220Z"
}

Integrations & automation

Call it over the Apify API (POST /v2/acts/zBUcP0d6qle75s6Qo/runs), or from Python, JavaScript, the CLI or MCP. Both datasets export as JSON, CSV, Excel, XML or RSS.

For reputation monitoring, schedule a daily run over a fixed businessSlugs list, upsert BUSINESSES on businessSlug, insert REVIEWS on id, and alert when trustIndex moves, when reviewCountLast12Months drops sharply, or when a new 1- or 2-star verifiedPurchase review appears. The weekly review sitemaps are newest-first, so a daily run captures new reviews without any pagination walking.

For South African market entry, use industrySlugs to get a ranked competitive set, then join reviews to compute category-level sentiment norms — so you can tell whether a 4.2 TrustIndex is good for that industry or just good in absolute terms. For customer-experience research, filter REVIEWS on verifiedPurchase: true and analyse businessReply against rating to measure whether replying to a complaint actually correlates with better subsequent scores.

Typical pipeline: run on schedule → drop kind: "error" rows → join on businessSlug → upsert businesses → insert reviews → alert on trustIndex deltas and new negative verifiedPurchase reviews.

What a run costs

Configuration	Billable events	Cost
20 businesses, no reviews	20 × Profile	$0.02
20 businesses × 28 reviews	20 Profile + 560 Review	$0.55
100 businesses, no reviews	100 × Profile	$0.10

Two separate event types: $0.00095 per business profile and $0.00095 per review. Reviews dominate cost by volume, so includeReviews: false or a low maxReviewsPerBusiness is the fastest way to cut a bill. Free accounts get a trial quota; paying accounts get full caps and tiered rates.

Related Actors

Actor	Why
google-maps-scraper	The other major local-reputation source — pair Google ratings with HelloPeter TrustIndex for the same business.
trustpilot-scraper	International review data, useful for South African businesses trading internationally.
indeed-scraper	South African employer reviews alongside job data.
glassdoor-scraper	Employee-side reputation to complement customer-side review data.
yellowpages-ca-scraper	Business directory data for enriching a HelloPeter competitive set with contact and category data.

FAQ

Which input should I use instead of searchQueries?

searchQueries is the only input that touches /api/search, which HelloPeter's robots.txt disallows. The compliant alternatives do the same work: businessSlugs for a known list, industrySlugs to enumerate by category, useSitemapDiscovery: true for bulk discovery from the published sitemap index, or startUrls with direct business URLs. Prefer those unless you have a specific reason not to.

Do I need a browser?

No. The Actor uses CheerioCrawler, which fetches HTML and parses the embedded payload without executing JavaScript. No headless browser, no captcha solving.

How do I join the two datasets?

On businessSlug. Do not join on businessName — names are not unique and businesses rename. businessSlug is stable, and REVIEWS also carries businessUrl if you prefer URL-based joins.

Why are there rows with kind: "error"?

So failures are visible. Rather than silently dropping a fetch that returned nothing usable, the Actor pushes a diagnostic row with kind: "error" and a human-readable message. Filter them out before analysis — but don't delete them blindly, since a run full of error rows usually means a URL problem, not an empty market.

Why is totalReviewCount different from reviewCountEmbedded?

They come from different sources: reviewCountEmbedded is read from the profile page payload, while totalReviewCount may come from a richer path with additional context. When they disagree, prefer reviewCountEmbedded as the number a user would see, and use totalReviewCount as the broader figure.

How do I get only the newest reviews?

Reviews are queued newest-first from the weekly review sitemaps, and maxReviewsPerBusiness caps the total. So a daily run with a low maxReviewsPerBusiness — say 10 — gives you the 10 most recent reviews per business and nothing older. Keep id as your dedupe key across runs.

What does truncatedFields mean?

It lists fields the server itself shortened when rendering the page. If about looks suspiciously short, check this array before assuming a parse bug — the page may simply have truncated it.

Can I compare businesses across industries on TrustIndex?

You can, but be careful. TrustIndex is computed against a category-relative baseline in HelloPeter's model, so a 4.2 in banking is not the same achievement as a 4.2 in restaurants. Use industrySlug and industryRank to normalise before drawing conclusions, rather than comparing raw scores across unrelated categories.

Do reviews contain real customer names?

Yes — authorName and authorId are real individuals, and authorMemberSince reveals account age. That makes this personal data with real obligations. Read Legal & Compliance before you store, analyse or redistribute it.

SEO Keywords

hellopeter scraper, hellopeter.com, south africa reviews, trustindex, south africa business reviews, local business reputation, sa customer reviews, review monitoring, competitor review analysis, business reputation tracking, star rating data, south africa market research, industry leaderboard, nps data, customer experience analytics, verified purchase reviews, business response time, sa brand reputation, review sentiment analysis, local seo data, south africa business directory, reputation management, review data api

For AI Agents & LLM Apps

Returns HelloPeter data as two datasets — BUSINESSES (36 fields, discriminate on kind) and REVIEWS (22 fields) — linked by businessSlug.

Minimal input — smallest valid run:

{ "businessSlugs": ["capitec-business"] }

Compliant broad discovery (use this rather than searchQueries):

{ "useSitemapDiscovery": true, "sitemapMaxSitemaps": 2, "maxBusinesses": 100 }

Profiles without reviews (cheapest):

{ "businessSlugs": ["capitec-business", "nedbank"], "includeReviews": false }

Full competitive set with reviews:

{ "industrySlugs": ["banks"], "maxIndustryPages": 2, "maxBusinesses": 60, "maxReviewsPerBusiness": 30 }

Output fields — BUSINESSES (36): kind, message, businessName, businessSlug, businessId, profileUrl, website, logo, about, aiSummary, aiThemes, keywords, popularMentions, faq, contact, trustIndex, trustIndexText, starRating, starDistribution, totalReviewCount, reviewCountEmbedded, reviewCountLast12Months, reviewHistory, nps, status, responseTime, negativeReplyTime, negativeResponseRate, industry, industrySlug, industryRank, subCategory, subCategoryRank, rankings, truncatedFields, scrapedAt

Output fields — REVIEWS (22): kind, id, businessName, businessSlug, businessUrl, industry, authorName, authorId, authorMemberSince, verifiedPurchase, rating, title, text, date, dateIso, helpfulCount, replyCount, businessReply, businessReplyDate, url, source, scrapedAt

Behaviours an agent should know

• Two datasets. Run IDs and dataset IDs differ — you must fetch both. Link them on businessSlug, never businessName.
• Always filter BUSINESSES on kind (business, industryListing, searchResult, error). error rows are diagnostics, not data — filter them out or your analysis silently includes failures.
• searchQueries hits a robots-disallowed endpoint. For broad work use useSitemapDiscovery, businessSlugs, industrySlugs or startUrls instead.
• Two billing event types: $0.00095 per business profile and $0.00095 per review. Reviews dominate by volume — includeReviews: false or a low maxReviewsPerBusiness cuts cost fastest.
• Reviews arrive newest-first from weekly review sitemaps, and only the 6 most recent weekly sitemaps are scanned. Anything older than ~6 weeks is not discoverable this way; use businessSlugs for historical depth.
• dateIso is the field to sort and filter on. date is display-formatted.
• businessReply is untyped — it can be an object, a string, or null. Handle all three.
• totalReviewCount and reviewCountEmbedded are two counts of the same thing from different sources; prefer reviewCountEmbedded.
• aiSummary, aiThemes, keywords and popularMentions are HelloPeter's generated content, not verbatim source data.
• contact and truncatedFields are objects/arrays respectively — parse defensively.
• The Actor never requests /profile/, /sign, /admin or /private paths; those are skipped in code.
• Sitemap fetches bypass the crawl budget, so maxRequestsPerCrawl (default 200) is spent on data-producing pages only.
• No browser is used — CheerioCrawler only.

Legal & Compliance

This Actor is an independent third-party tool. It is not affiliated with, endorsed by, or sponsored by HelloPeter, or by any business whose profile or reviews appear in its output.

What it actually accesses: public HelloPeter business profile pages at their root URLs, public review pages at /{slug}/reviews/{id}, public industry leaderboard pages, and the XML sitemaps that HelloPeter's robots.txt explicitly declares. No login, no authentication, no captcha solving, no browser.

What it deliberately does not touch: HelloPeter's robots.txt disallows /api/ and /profile/. The Actor's URL classifier skips /profile, /sign, /admin, /private, /terms, /about and /search paths by design, and never requests the disallow-listed /profile/ path. The single exception is the searchQueries input, which calls /api/search. Use businessSlugs, industrySlugs, useSitemapDiscovery or startUrls instead — they cover the same use cases on sitemap-declared, permitted paths.

Personal data — this is the serious part. The REVIEWS dataset contains identifiable consumer data: authorName, authorId, authorMemberSince and review text. In South Africa this is governed by the Protection of Personal Information Act (POPIA), with GDPR applying to any EU residents' reviews and CCPA/other state laws to California residents. Reviews also carry reputational significance, which means a wrong or outdated record can cause real harm to a real person.

If you process this data you are the responsible party and you must: give notice to affected individuals, honour access and erasure requests, restrict processing to a stated lawful purpose, apply appropriate security safeguards, and delete or anonymise once the purpose is met. Do not build profiles of individual reviewers, do not re-identify pseudonymous reviewers, and do not use review text to make automated decisions about people.

Platform-generated content. aiSummary, aiThemes, keywords and popularMentions are HelloPeter's own generated output. Treat them as HelloPeter's intellectual property, not as public-domain facts, and get permission before republishing them.

Review and profile content ownership. Review text belongs to its individual authors; business profiles, logos and contact details belong to the businesses. Review aggregation and display may also carry attribution obligations to the individual reviewer. Republishing review text at scale without the author's consent can breach both the authors' rights and HelloPeter's Terms of Use.

Commercial terms. HelloPeter's Terms of Use reserve rights over its database, ratings model and generated content. Building a competing review or reputation product from this data is likely to breach them. If you need South African review data commercially, a licence from HelloPeter is the defensible route.

You are solely responsible for compliance with HelloPeter's Terms of Use, robots.txt, POPIA, and applicable law. Keep runs modest — the default maxRequestsPerCrawl: 200 and maxConcurrency: 8 bound a run without effort — and prefer the sitemap and slug-based inputs over searchQueries.