SB StartupBasket
All ideas
76 /100 GO Medium complexity

IzinLacak — listing clearance for Indonesian sellers

Tells an Indonesian online seller which of their listings will get suspended, and which certificate saves each one.

Retail SaaS SEA WhatsApp-firstCompliance-drivenSMBMultilingualAI-agent
— views
Evaluation Scores
76/100

GO

Overall Score

16
Problem
12
Demand
11
Build
12
Distrib.
11
Revenue
8
Time
6
Defense

IzinLacak

1. One-liner

Tells an Indonesian online seller which of their listings will get suspended, and which certificate saves each one.

2. Trend signal — why now?

Three enforcement clocks landed on the same Indonesian seller inside five months of each other, and they compound.

Permendag 19/2026 took effect 8 June 2026. It replaced Permendag 31/2023 and did something the old rule never did: it made the marketplace the enforcer. Shopee, Tokopedia, Lazada and TikTok Shop are now legally required to reject seller registrations without a valid NIB, and to suspend access when a seller’s grace period expires. Platforms themselves face sanctions if they fail to block non-compliant sellers. Existing sellers registered before 8 June 2026 get 18 months; new sellers get 6 months from registration. In the interim, the platform must brand their storefront with a “Dalam Proses Legalisasi” label — a public scarlet letter on the shop profile while the clock runs.

The sanction ladder is specific: up to three written warnings at 14-day intervals, then priority-supervision listing, then blacklisting, then temporary platform blocking, then licence revocation.

Here’s the part the market is sleeping on. The regulation’s minimum licensing is two things, not one: (i) an NIB in the trade sector, and (ii) proof of compliance with mandatory standards and technical requirements for the goods or services being sold. Every consultancy in Indonesia is currently selling NIB filing. As one Indonesian licensing explainer puts it flatly: “NIB hanyalah legalitas minimal” — NIB is only the minimum legality. The second half — SPPT SNI for regulated electronics, BPOM notification for cosmetics, SPP-IRT for home-produced food, halal for consumables — is per-product, not per-business, and nobody is mapping it to the seller’s actual catalogue.

Meanwhile SNI enforcement hardened. Under Ministry of Industry Regulation No. 7 of 2025, the SNI regime was overhauled effective 24 July 2025, with full enforcement from 24 July 2026 — splitting product certification from market-access licensing and routing all applications through the SIINAS system. There are 200+ product types under mandatory SNI.

And halal goes live 17–18 October 2026. BPJPH confirmed mandatory halal certification takes effect October 2026 for food and beverage, having already slipped from October 2024 under Government Regulation 42/2024. Enforcement and active market supervision begin 18 October.

So a seller with 40 SKUs across food, cosmetics and a few electronics accessories is now facing a per-SKU certificate matrix, a platform that is contractually obliged to switch them off, and a public label on their shop in the meantime. The seller anger is already in the press — Indonesian trade coverage in June 2026 ran under the headline “Seller Shopee-TikTok Shop Cs Menjerit Toko Ditutup Sepihak” (sellers scream as stores are closed unilaterally), and Kemendag had to publicly respond to seller complaints about NIB processing on 25 June 2026. Note the sequencing: Permendag 19/2026 was itself issued after waves of merchant protest that peaked following new logistics fees from TikTok Shop and Shopee in May 2026. This population is loud, organised, and already paying attention.

Provenance:

3. The opportunity

The gap is the difference between a filing and a catalogue.

Indonesia has a large, mature cottage industry of biro jasa and licensing consultancies — POP Jasa, StaffAhli, izin.co.id, Komplace, dozens more — all of whom will file your NIB. Pricing starts around Rp500,000 and it’s a one-off transaction. Their entire product is: you tell them your KBLI code, they hand you a document. Some will also broker a BPOM or halal application if you ask and know to ask.

None of them start from your actual product listings. That’s the whole miss. The seller’s question is not “how do I get an NIB” — that’s free on oss.go.id and every consultancy in the country is shouting about it. The seller’s question is:

I have 60 SKUs live on three platforms. Which of them are going to get me suspended, when, and what exactly do I have to hold for each one?

Nobody answers that. Answering it requires joining three things no consultancy joins: (a) the seller’s live listing data, pulled from platform APIs, (b) the mandatory-standard matrix — 200+ SNI product types, BPOM categories, SPP-IRT thresholds, halal scope — and (c) the seller’s own document wallet, with expiry dates.

The incumbents’ structural weakness is that they are transaction businesses in a world that just became a surveillance problem. Permendag 19/2026 isn’t a one-time filing event. It’s a permanent condition: platforms now run continuous compliance monitoring, new SKUs added next month inherit the same duty, certificates expire, and the sanction ladder resets warnings at 14-day intervals. A consultancy that sold you a document in July has no relationship with your November catalogue.

The second weakness is that this is unaffordable to serve manually at the seller’s price point. A consultant reviewing 60 SKUs against a 200-row standards matrix is a day of billable work. At the wallet of a micro-seller, that math never closes. This is exactly the shape my persona hunts: a task that costs a professional two hours and a model two minutes, in a market where the professional’s price excludes the entire customer base.

I’m explicitly not building a certification broker. Getting the actual SPPT SNI or BPOM notification stays with the labs and LSPro bodies. I’m building the layer that tells you what you need, what you’re missing, what’s about to expire, and what gets switched off first — then hands off.

4. Target market

  • Primary customer: Indonesian online sellers with 15–300 active SKUs across Shopee, Tokopedia, TikTok Shop or Lazada, selling in a regulated category — packaged food, beverages, cosmetics, skincare, supplements, small electronics and accessories, kids’ products, textiles. Monthly GMV roughly Rp30 juta – Rp1 miliar. Typically 1–8 staff. The buyer is the owner or the single “admin toko” who handles listings and legality. Concentrated in Jabodetabek, Bandung, Surabaya, Semarang, Medan, Solo.

  • Why they buy: Because the suspension is not hypothetical and it is not gradual — it is the platform switching off their only revenue channel. Indonesian trade press in June 2026 was already reporting sellers “menjerit” over unilateral store closures and the difficulty of getting funds released after a suspension. The pain has a date attached to it, which is what converts anxiety into a purchase. A seller who added a store in July 2026 has a hard six-month wall in January 2027. A seller who was live before 8 June 2026 has until roughly December 2027 — and a “Dalam Proses Legalisasi” badge on their shop the whole way.

  • Rough TAM reasoning: OSS data through end-February 2026 shows 15.4 million NIBs issued, 96%+ held by micro businesses — that’s the registered base, not the seller base, but it sizes the population the government is corralling. Indonesian e-commerce is dominated by micro and small merchants. I don’t need a big share: my serviceable target is sellers with enough SKUs and enough regulated-category exposure to feel real risk. Call it low hundreds of thousands of shops nationally that are both regulated-category and large enough to pay. Capturing 2,000 of them at Rp399K/month is Rp9.6 miliar ARR ($590K). That’s a business.

  • Why now for them: Three deadlines inside five months (24 Jul 2026 SNI, 17–18 Oct 2026 halal, rolling 6/18-month Permendag windows), plus a platform that is now legally obligated to act against them rather than merely permitted to. Before June 2026, ignoring this was free. It isn’t anymore.

5. Product sketch (MVP)

  • Connect your shops. OAuth into Shopee/Tokopedia/TikTok Shop seller accounts, pull the live catalogue. For sellers who won’t connect an API, accept a CSV export or a screenshot of the product list.
  • Per-SKU verdict. Every listing gets one of four states: Aman (clear), Perlu Dokumen (needs a certificate you don’t have), Kedaluwarsa (you had it, it expired), Perlu Cek (ambiguous, needs a human eye). Each verdict names the specific instrument — SPPT SNI, BPOM notification/izin edar, SPP-IRT, halal — and cites the rule behind it.
  • Suspension clock. One number on the dashboard: days until your grace window closes, per platform, per shop, based on registration date. Plus the sanction stage you’re currently at if warnings have started.
  • Document wallet with expiry watch. Upload the certificates you hold. The system reads them, extracts holder, product scope and expiry date, and matches them to the SKUs they actually cover — the common failure is holding a valid certificate that doesn’t cover the variant you’re selling.
  • New-listing pre-check. Before you publish a new SKU, paste the title and category and get a go/no-go plus what you’d need. This is the habit-forming surface.
  • WhatsApp alerts in Bahasa. “3 produk kamu berisiko disuspend dalam 21 hari.” Weekly digest, plus event alerts when a certificate nears expiry or a new SKU lands dirty.
  • Referral handoff. When a SKU needs a real certificate, one tap sends the spec to a vetted LSPro/consultant partner. I don’t do the certification; I originate the lead.
  • Audit-ready export. A PDF of your catalogue’s legality status, for when the platform or a Kemendag supervisor asks.

6. AI angle — what’s load-bearing

Remove the AI and this product dies. That’s the test, and it passes.

The load-bearing task is classification under ambiguity at catalogue scale. Indonesian marketplace listing titles are not clean product taxonomies. A real listing reads like "Setrika Uap Portable Mini Travel Steamer 800W Garansi Resmi COD Murah". The mandatory-SNI matrix is organised by product type and HS code. Mapping that title — plus the category the seller picked, which is frequently wrong because sellers game categories for search placement — to “is this a regulated electrical appliance under mandatory SNI, and does 800W put it in scope” is exactly the judgment a compliance analyst makes and exactly what a model does well and cheaply. Times 60 SKUs, times every week, times a catalogue that changes.

The second AI job is document understanding: reading an uploaded SPPT SNI or BPOM notification — usually a phone photo of a printed Indonesian certificate — and pulling holder name, product scope, certificate number and expiry. Then the non-obvious bit: deciding whether the scope on that certificate actually covers the SKU the seller is selling. Sellers routinely believe one certificate covers a whole product family when it covers one variant.

Third: vernacular explanation. The output has to be readable by someone who runs a shop, not a lawyer. Turning “Permenperin 7/2025 mandatory SNI, Scheme 5, SIINAS” into “produk ini wajib SNI, kamu belum punya, ini yang harus diurus dan kira-kira makan waktu segini” is a generation task in Bahasa Indonesia.

What is not AI, and must not be: the rules matrix itself. The list of mandatory-SNI product types, BPOM categories and halal scope is curated, versioned, human-maintained reference data. Letting a model freelance on whether a product is regulated is how you get a seller suspended and your reputation destroyed. Model classifies into the matrix; the matrix is ground truth. Anything the model is unsure about routes to Perlu Cek rather than guessing — a compliance product that is confidently wrong is worse than no product.

7. Localization angle

This is the localization play — it’s Indonesia-specific by construction and can’t be a global product wearing a translation.

  • Language: Bahasa Indonesia first, and specifically seller-register Bahasa, not legal Bahasa. The regulatory source material is Indonesian-only.
  • Distribution rail: WhatsApp. Indonesian sellers live in WhatsApp groups and run their business from a phone. Alerts, weekly digests and the new-listing pre-check should all work over WhatsApp without opening a dashboard.
  • Payment rails: QRIS and bank transfer via Midtrans or Xendit. Card-only checkout would kill conversion outright. Monthly billing, not annual — micro-sellers don’t prepay years.
  • Price point: Rp299K–Rp799K/month ($18–$49). A $99/mo global compliance SaaS is simply not sellable here; the whole opportunity exists in the gap between what a consultant costs and what a seller can pay.
  • Regulatory quirk as moat: the SNI/BPOM/SPP-IRT/halal matrix is Indonesian domestic law with no analogue elsewhere. A foreign entrant has to build this from Indonesian-language primary sources and keep it current. That’s the barrier — and it’s the same reason this doesn’t generalise to a global product, which is fine. Adjacent expansion is Philippines and Vietnam, which have structurally similar marketplace-legality pushes, not the US or EU.

8. Business model — path to $1M–$5M ARR

  • Pricing (per shop group, monthly):

    • Cek — Rp149K/mo (~$9): up to 25 SKUs, one platform, weekly scan, WhatsApp alerts.
    • Toko — Rp399K/mo (~$25): up to 150 SKUs, three platforms, document wallet with expiry watch, new-listing pre-check, audit export. This is the anchor tier.
    • Bisnis — Rp999K/mo (~$61): up to 600 SKUs, multi-shop, team seats, priority human review of Perlu Cek items.
    • One-off catalogue audit — Rp750K, credited against the first two months of subscription. This is the wedge product: it converts panic into cash today and subscription later.

  • ACV: blended Rp4.8 juta/year ($295) assuming the mix lands on Toko.

  • Rough math to $1M ARR: ~2,700 paying shops at ~Rp4.8 juta/yr ≈ Rp13 miliar ≈ $800K, plus referral commission on certification handoffs. Realistically I get to $1M with ~3,000 shops and a partner-referral line. Against an Indonesian seller base in the hundreds of thousands of regulated-category shops, 3,000 is a small share — but it is 3,000 individual small merchants, and I should be honest that this is a volume-of-smalls business, not an enterprise one.

  • Rough math to $5M ARR: needs three things true. (1) Mix shifts up — agencies and brand aggregators managing 10–40 shops each buy Bisnis, lifting ACV toward Rp12–18 juta. (2) Certification referral revenue becomes material: at Rp15 juta+ for an SPPT SNI process and Rp350K per halal product, even modest origination commission on a few thousand handoffs is a real second line. (3) Geographic extension to Philippines/Vietnam marketplace-legality regimes. Without (1) and (2), this is a comfortable $1–2M business and I’d take that.

  • Expansion path: SKU-count tiers are the natural meter — a growing seller crosses tiers without a sales conversation. Then multi-shop, then team seats, then the referral line, then a “managed” tier where a human handles the whole remediation queue.

9. Go-to-market wedge — first 100 customers

The population is unusually reachable, which is the best thing about this idea.

  • Seller communities, directly. Indonesian marketplace sellers organise in large WhatsApp and Telegram groups, Facebook seller groups, and around seller-education creators. The June 2026 protest wave over TikTok Shop and Shopee logistics fees proves these channels are active and coordinated. Move: join 30–50 of these groups, and post one genuinely useful free asset — a “cek cepat wajib SNI/BPOM/halal” checker that takes a listing title and returns the verdict, free, no signup. Free tool first, paid audit second. Target: 15–25 paying shops from this in the first 6 weeks.

  • Free public checker as an SEO and sharing asset. Indonesian sellers are actively searching this right now — the search landscape is full of “NIB marketplace 2026”, “wajib SNI produk”, “cara mengurus NIB seller Shopee”. Every consultancy blog is competing on the NIB filing keyword. Almost nobody owns the “produk saya wajib SNI atau tidak” keyword, which is the higher-intent question. A free per-listing checker that outputs a shareable verdict card gets forwarded inside WhatsApp groups, which is the only distribution that matters here.

  • Ride the deadline calendar with paid audits. Two hard dates — 24 July 2026 (SNI) already passed and 17–18 October 2026 (halal) imminent — plus rolling per-seller Permendag windows. Run a countdown campaign into October: “audit katalog kamu sebelum 17 Oktober.” Deadline-driven compliance selling converts because the date does the persuading, not the copy. Target: 40+ one-off audits at Rp750K through September–October, most of which roll into subscription.

  • Consultant and biro jasa partnerships — as a channel, not a competitor. The licensing consultancies have the seller relationships and no recurring product. Offer them white-labelled catalogue audits: they run the scan under their brand, keep a margin, and I take the subscription. Sign 5 firms, each with a few hundred seller clients. This is also my certification-referral supply in reverse.

  • Direct outreach to visibly-exposed shops. Public marketplace storefronts show product listings. Scrape regulated-category shops with 20+ SKUs, identify SKUs that are obviously missing required marks, and send a WhatsApp message with a free three-SKU sample verdict — the specific products, the specific risk. Personalised risk beats a generic pitch. 500 shops, expect low-single-digit conversion to a paid audit.

10. Build complexity — justification

Medium. Two to three months to a credible v1 for a pair.

Off-the-shelf: platform seller APIs (Shopee Open Platform registration is free and issues per-shop OAuth tokens; Tokopedia and TikTok Shop have official seller APIs), standard web stack, off-the-shelf models for listing classification and certificate OCR, Midtrans/Xendit for QRIS billing, WhatsApp Business API for alerts.

The real work is not engineering, it’s the rules matrix: assembling and maintaining the mandatory-SNI product list (200+ types, HS-code keyed), BPOM category scope, SPP-IRT thresholds and halal scope from Indonesian primary sources, and versioning it as regulations shift. That’s a domain-research grind, not a technical one, and it’s also the moat — so it’s the right place to spend the effort. Budget a solid chunk of the first two months on it. Secondary risk is platform API access approval timelines, which is why CSV/screenshot import must exist in v1 as a fallback path rather than a nice-to-have.

Nothing here needs custom models, novel infrastructure, or regulatory approval to operate.

11. Gating checklist

GatePass?Note
Legal in target market✅An informational compliance-monitoring tool. Not brokering certification, not filing on the seller’s behalf, not a licensed activity. Must carry clear “not legal advice” positioning.
Ethical — no harm / dark patterns✅Helps small merchants comply with law that exists to protect consumers. The deadline urgency is real, not manufactured — but the marketing must not overstate suspension risk to drive fear sales.
Market exists (evidence above)✅Regulation in force since 8 June 2026, documented seller complaints, an existing paid consultancy market.
1–5 person team can build this✅Pair in 2–3 months; one of them must be able to read Indonesian regulation carefully.
Launchable with <$50K / ₹40L✅Well under. Inference, hosting, WhatsApp API, and the domain-research time.

All five pass.

12. Feasibility score

AxisWeightScoreNotes
Problem intensity2016/20Losing the only sales channel is severe and the platform is now obliged to act. Docked because for the 18-month cohort the wall is late 2027 — real, but not this week, and procrastination is the true competitor.
Demand evidence1512/15Regulation in force, sanction ladder specific, seller complaints in national trade press, an established paid market for adjacent filing services. Docked: I could not source direct seller quotes asking for this product specifically — the demand for the underlying compliance is proven, the demand for a monitoring SaaS is inferred.
Build feasibility1511/15Standard stack and off-the-shelf models, but the rules matrix is a genuine multi-week research grind and platform API approvals are outside my control.
Distribution clarity1512/15Named channels — seller WhatsApp/Telegram/FB groups, the free-checker viral asset, biro jasa partnerships, deadline campaigns. Conversion rates unproven, and these groups punish anything that smells like spam.
Revenue mechanics1511/15Pricing benchmarked against real consultancy fees and micro-seller wallets. Docked for volume-of-smalls economics: 3,000 individual micro-merchants for $1M means churn and support load are the whole ballgame.
Time to first revenue108/10The Rp750K one-off catalogue audit can sell before the subscription product is complete — semi-manual delivery for the first customers is legitimate. Revenue in ~6 weeks is realistic.
Defensibility106/10Soft moat: the curated rules matrix compounds, the document wallet creates switching cost, and Indonesian regulatory depth is a genuine barrier to foreign entrants. But a local consultancy with engineers could replicate this in months.
Total10076/100

13. Qualitative modifiers

Founder-fit tags

technical-heavy · content-heavy

Needs someone who can build API integrations and ship model-backed classification, paired with someone who can read Indonesian regulation precisely and produce the vernacular content that wins the seller-group channel. Indonesian language fluency is non-negotiable — this is not a market you serve through translation. Not domain-expertise-required in the licensed-professional sense, but you must be willing to become a genuine expert on the standards matrix.

Key assumptions to validate (3–5)

  1. Assumption: Sellers will pay for knowing rather than for doing — i.e. a diagnosis has standalone value even though the certification still has to be bought elsewhere. How to test: offer 40 sellers a Rp750K catalogue audit with the explicit caveat that it does not include obtaining any certificate. If they balk and demand end-to-end, the product is a brokerage, not a SaaS, and the economics change completely.

  2. Assumption: Listing-title classification against the mandatory-standards matrix is accurate enough to trust — say ≥90% on clear cases with honest abstention on the rest. How to test: hand-label 300 real listings across food, cosmetics and electronics, measure precision/recall, and specifically measure the false-clear rate. A false Aman is the fatal error mode.

  3. Assumption: Platform seller APIs are obtainable by a small third-party developer on a workable timeline. How to test: apply to Shopee Open Platform and TikTok Shop partner programs in week one, before writing integration code. If approvals stall, CSV/screenshot import has to carry v1.

  4. Assumption: Seller WhatsApp/Telegram/Facebook groups convert rather than eject. How to test: seed the free checker in 10 groups and measure link-shares and signups per group before scaling to 50.

  5. Assumption: The 18-month cohort acts before Q4 2027 rather than sleeping until the wall. How to test: compare conversion between new sellers on the 6-month clock and established sellers on the 18-month clock. If only the short-clock cohort buys, the addressable market this year is far smaller than the total.

Risk flags

  1. Platform dependency: the whole product assumes access to marketplace seller APIs. Shopee, Tokopedia or TikTok Shop could restrict third-party access, or — worse — build a native “your listing needs a certificate” checker directly into Seller Centre. Given Permendag 19/2026 makes them liable for non-compliant sellers, they have a real incentive to do exactly that. This is the single biggest threat to the idea, and it argues for moving fast and for being multi-platform so no single decision kills you.

  2. Regulatory churn: Indonesian deadlines move. Mandatory halal already slipped from October 2024 to October 2026. If the Permendag windows get extended under seller pressure — and sellers are already applying that pressure successfully — urgency deflates and the sales cycle stretches. The product survives; the campaign calendar doesn’t.

  3. Accuracy liability: telling a seller a SKU is clear when it isn’t could contribute to a suspension. Even with disclaimers, one loud failure in a WhatsApp group of 5,000 sellers is a reputational event in a market that runs on word of mouth. Bias hard toward Perlu Cek, and never let the model overrule the matrix.

  4. Volume-of-smalls economics: thousands of Rp399K subscriptions means support burden and churn dominate. Micro-merchants churn when they close shop, and many do. Self-serve onboarding and WhatsApp-native support aren’t features here — they’re survival requirements.

14. Structured verdict

Score:                  76/100
Verdict:                GO
Confidence:             Medium
Best-fit builder:       Indonesian-fluent technical pair — one shipping integrations and
                        model-backed classification, one owning the standards matrix and
                        the seller-community content channel
Time to revenue:        6-8 weeks via the one-off Rp750K catalogue audit, sold semi-manually
                        ahead of the full product
Capital to launch:      $8-15K (Rp130-250 juta)
Top 3 assumptions to validate first:
  1. Sellers pay for diagnosis without remediation — sell 40 paid catalogue audits with an
     explicit "certificate not included" caveat and measure refusal rate
  2. Classification accuracy clears 90% with a near-zero false-clear rate — hand-label 300
     real listings across food, cosmetics and electronics before writing marketing copy
  3. Marketplace seller API access is obtainable by a small third party — apply to Shopee
     Open Platform and TikTok Shop partner programs in week one
Kill criteria:
  - Abandon if fewer than 8 of 40 sellers offered the Rp750K audit convert — the diagnosis
    has no standalone value and this is a brokerage business, not a SaaS
  - Abandon if any major platform ships a native per-listing certificate checker in Seller
    Centre before v1 launches
  - Abandon if false-clear rate exceeds 3% on the 300-listing labelled set and cannot be
    driven down by routing ambiguity to Perlu Cek
  - Abandon if the Permendag grace windows are extended by 12+ months and the short-clock
    new-seller cohort proves too small to sustain acquisition

15. Next step — 1-week validation sprint

  • Day 1–2 — Build the matrix slice and the labelled set. Pick three categories: packaged food, cosmetics, small electrical appliances. Assemble the mandatory-standards rules for just those three from Indonesian primary sources. Pull 300 real listings from public marketplace storefronts and hand-label each as regulated/not-regulated and with which instrument. This doubles as the accuracy test set and as proof the matrix is buildable.

  • Day 3 — Run the classifier against the labelled set. Off-the-shelf model, listing title plus seller-chosen category as input, matrix as ground truth. Measure precision, recall, and above all the false-clear rate. This is a falsifiable technical gate before any customer conversation.

  • Day 4–5 — Sell the audit, not the software. Identify 40 regulated-category shops with 20+ SKUs. Send each a WhatsApp message containing a free three-SKU verdict on their actual listings — naming the products and the specific risk — and an offer of a full catalogue audit at Rp750K, stating plainly that obtaining certificates is not included. Simultaneously file the Shopee Open Platform and TikTok Shop partner applications so the clock starts.

  • Decide go / no-go on two numbers: ≥8 of 40 sellers pay Rp750K for a diagnosis-only audit, and false-clear rate ≤3% on the 300-listing set. Both must clear. Fail the first and this is a brokerage, not a SaaS — a different, worse business. Fail the second and the product is a liability rather than a service, and no amount of demand fixes that.

Interested in a detailed proposal?

Get a deep-dive with market research, competitive analysis, and implementation roadmap.

Contact us

info@startupbasket.ai