This document proposes how docus.ao should be priced. But a price is an answer to a question — "versus what?" — so before any number, here is the market it lives in: who buys it, who we're really up against, and why that makes the pricing logic what it is. Five minutes here makes the Pricing tab obvious.
Every Angolan business runs on documents with expiry dates: alvarás, licences, certidões, insurance, vehicle and personnel papers. When one lapses unnoticed, the cost is not abstract:
One lost tender is worth millions of kwanzas — and a single expired document can disqualify the bid.
An expired licence triggers penalties, and sometimes a halt to operations until it's resolved.
When the bank, regulator, or auditor asks, the team is hunting through folders and inboxes instead of answering.
A junior admin tracking this in a spreadsheet costs more per month than our mid-tier plan — and still misses things.
If any of these land, the value is already established — docus.ao is the system that ensures it never happens again. The price discussion is then trivial: $25/month against the loss they just described.
Lives the spreadsheet treadmill. Wants end-of-quarter calm and to stop being the single point of failure. Feels the pain daily and champions the tool.
Signs the cheque. The case is trivial: one missed Alvará costs more than a year of docus.ao. At $25/month, approval needs no procurement cycle.
An internal competitive scan found no comparably positioned product in the Portuguese-first, Angola-regulator-aware compliance space. The nearest adjacents are all foreign, generic, and English-first — none built for Angolan documents or regulators:
| Nearest adjacent | What it is | Model | Typical price |
|---|---|---|---|
| FileFlo | US · PDF compliance folders | Per-user + per-GB | ~$299/mo |
| Expiration Reminder | US · generic expiry tracker | Per-record tiers | $49–349/mo |
| Remindax | SaaS expiry reminders | Per-user flat | $29–49/mo |
| MyPass | AU · workforce compliance | Per-worker / year | ~$93/worker·yr |
Source: internal competitive scan (DEC-034). Figures are published list rates, not Angola-localised — shown to bound the category, not as direct comparators.
The honest competitor isn't a SaaS product. It's an informal, unowned tracking habit: one or two people watching expiry in a spreadsheet, an Outlook or Google calendar, or from memory — with no formal ownership. Nobody is paid for this job today. (Despachantes and external firms get involved only to file or renew a specific document once it's already due — not to track what's coming.)
It works — until it doesn't. The cracks are predictable:
When the bank asks for the actual file, you're still hunting through folders.
A spreadsheet can't tell you what's missing — the most dangerous gap of all.
The knowledge lives with one person. When they're away, tracking stops silently.
Pulling the right documents for a bid is 45 minutes of manual stitching, every time.
This is the bridge to the Pricing tab. Because the real alternative is an informal habit — not a product anyone is paying for — two things follow that shape the entire model:
There's no benchmark SKU to sit just under. So the entry point is set at the cost of a business lunch (~$25/mo, ~22,000 AOA) — small enough to approve without a procurement conversation.
The cost of a miss grows with how many documents you must keep valid — so the meter is the tracked entity (person, vehicle, equipment, property), and the rate falls as volume grows.
→ Now open the Pricing Model & Economics tab for the structure, the calculator, and the five-year picture.
Following extensive research and analysis, this is the proposed pricing structure for docus.ao, the compliance document platform for Angolan businesses — from small SMEs to large enterprises (the model meters by tracked entities, not headcount, so there is no upper bound). Angola is the first market; the ambition is worldwide as the model proves out. The recommendation — progressive, per-tracked-entity pricing — emerged from evaluating five alternative models against the realities of the Angolan market (see §2 / Rationale). Everything here is open for challenge: rate points, tier boundaries, the $25 floor, and the 5-year assumptions.
Five-year base case, Angola only. Shared-team portfolio allocation assumed (the single biggest assumption in the model).
Tracked entities — the people, vehicles, equipment, and properties whose compliance documents we manage. One driver's licence, one truck inspection, one excavator certificate = one billable entity.
Not users. Not documents. Not headcount. We price the thing that creates the compliance surface.
Progressive tax-bracket rates: $2.50/entity/month at the top, sliding down to $0.70 at volume. Nine tiers. $25/month floor protects the entry price.
All features included at every level. No plan names. No feature gates. One calculator on the website.
This is the part worth challenging. Five pricing models were analysed against the realities of the Angolan market; two were seriously prototyped before landing on the third-generation design. The evolution is the rationale — each rejection taught us what the right driver looks like.
| Model evaluated | Mechanism | Verdict |
|---|---|---|
| Employee-band tiers | Price by headcount | Rejected — wrong driver; platform usage doesn’t scale with company size |
| Document-volume tiers | Price by document count | Rejected — gameable and abstract to the buyer |
| Base + flat per-entity | $39 base + $2/entity | Rejected — variable bills are hard for enterprise finance to approve |
| Named plans + entity quotas | Essencial / Profissional / Empresa | Rejected — cliff edges (1% growth → 150% jump) + needless plan-name friction |
| Progressive per-entity | Tax-bracket rates · no plan names · $25 floor | Selected ✓ — no cliffs, fair at every scale, self-serve-able, value-aligned |
Inspiration: managed-services licensing (Kaspersky, CodeTwo, Dropsuite) and income-tax brackets. Both handle the same problem — pricing fairly across a wide volume range without arbitrary thresholds.
Each tier applies only to entities in that band. So every customer — from a 5-person consultancy to a 3,000-person bank — pays $2.50 on their first 10 entities, then $2.20 on the next 15, and so on.
The rates above are marginal — each one applies only to the entities inside its band. So the $0.70 tier never applies to a whole account; it only touches the entities beyond the 2,500th. The number that actually lands on the invoice is the blended average across all of a customer's bands — and that average falls steeply with size, but levels off well above the marginal floor.
Every additional entity is billed at the next tier's marginal rate. No renewal conversations. No "plan upgrade" moments.
A 3,000-entity customer still pays the top rate on their first 10 — same entry as a 5-person shop. Price grows with value received.
Three-field calculator on the website (employees + vehicles + locations) replaces a 10-page comparison grid. Angolan SME buyers expect "one price, one concierge call" — not a feature matrix.
The bigger the customer gets, the harder the switch back to Excel. Every entity added is another reason to stay.
The first question every prospect asks: "do we pay per user?" No. The rate chart in §3 applies only to entities — tracked people, vehicles, equipment, properties. Seats, legal entities, documents, and every feature scale automatically and sit in the box, not on the invoice. This is the most important commercial distinction in the model.
One licence = one compliance surface. The tracked thing whose documents we manage.
Entry and top-tier values per the pricing brief (15 and 175 seats respectively). The intermediate schedule scales proportionally across the 8 tiers — the exact step points are partner-reviewable.
Recall Stage 1 from §2 — the rejected employee-band model. A per-user meter was rejected because user count doesn't track with compliance surface. A 500-person bank has 3–5 compliance officers; a 50-person logistics firm has 1. Billing per seat would punish the bank for its size and undercharge the logistics firm for its risk. So we inverted it: seats are bundled, entities are the meter. The bank pays for the 3,000 regulated items it tracks; the 5 people who manage them are free.
Same math as §3, rendered the way a prospect would experience it at docus.ao/precos. Four numbers in, one price out. The tier mechanics happen invisibly — the customer doesn't need to know there are 8 brackets, just what their price is. Try it: change any number and the price updates.
A deliberate product decision: the tier breakdown (Tier 1 at $2.50, Tier 2 at $2.20, etc.) stays hidden from the customer. That mechanism is how we explain fairness internally — and the commercial defence of "no cliff edges" for partners and investors. But on the pricing page, the customer just needs to know their price. Four inputs, one number. No procurement conversation, no plan comparison, no feature matrix.
Progressive pricing only works if the commercial motion works. Here's the journey, month by month, for a realistic logistics/services SME.
The commercial point: every price increase is the customer's choice (they added entities) and every increase delivers more value. No upgrade conversation ever happens.
Five archetypes spanning the full spectrum. Same pricing rules apply to all.
First, the framing: these margins are a consequence of value-based pricing meeting a near-zero marginal cost — not the reason for the price (see the philosophy note in §2). The $2.50 rate is justified by the value at risk, not by a markup on cost; it would be defensible even if our costs were several times higher. With that said — the per-entity margin is extreme because the product is small files, multi-tenant, and runs on shared infrastructure. The real watch-point isn't per entity — it's per customer, where support cost kicks in.
Marginal cost per tracked entity: ~$0.016/month (storage + 1 SMS + email + compute). Against a $2.50 top rate, that's 99.4%. Even at the $0.70 floor of the discount tiers, it's still 97.7%. The SMS bulk-pack pricing means margins improve as we scale, not compress.
These ratios look aggressive because they assume Exponencial portfolio allocation, not dedicated sales/CS. The CAC figures are the item most worth challenging — they benefit heavily from founder-led sales and brand halo.
Key insight: floor customers ($25/mo) are only ~40% contribution-positive. They are land-and-expand seeds, not the margin engine. The margin engine lives above ~50 entities, where support cost amortises. The floor is a commercial hook, not a profit centre.
Base case, Angola only. Customer mix shifts up-market over time as land-and-expand takes effect.
Our cost base runs 20–40 percentage points below the SaaS Capital 2025 industry median. The portfolio advantage is the single biggest reason.
Y1 Support/CS shows 25% because the denominator (ARR) is tiny. Normalises by Y3. Industry median: SaaS Capital 2025 survey, 1,000+ bootstrapped B2B SaaS companies.
None of these are hidden. All are open. An experienced commercial review adds the most value here.
Nothing above is locked except the product definition and the architectural choice of progressive per-entity pricing. Everything else is up for challenge — and the earlier, the better. What we need most from this review:
A subscription holds a number of billable units, each covering one tracked thing — a person, a vehicle, a piece of equipment, a location, a card. Naming that unit is harder than it looks: entidade didn't land, and the obvious alternatives each collide with something in Portuguese. Eight candidates were tested — here is the analysis, the winner, and how the customer actually experiences it.
Most of the difficulty was one word being asked to do three jobs. Separate them and the problem shrinks.
The commercial container — the term-based wrapper the customer signs up to. Settled; no issue.
A fungible slot, N per subscription, freely allocated. This is the word we're choosing.
The real-world subject a slot points at — colaborador, viatura, equipamento, localização, cartão. Named by its concrete category, not an umbrella word.
The partner's discomfort was really with using one abstract umbrella noun (entidade) for both the billable unit and a person or a card. Splitting the three roles is most of the fix.
The constraint isn't effort, it's pt-AO vocabulary: one word has to fit a person, a truck, a generator, a warehouse and a bank card, carry a compliance tone, and not collide with an existing term inside or outside the product.
| Candidate | Verdict | Why |
|---|---|---|
| Entidade | Rejected | Collides with entidade legal (the company / NIF) inside our own data model — and reads as "organisation," which jars for a car or a card. |
| Registo | Rejected | Collides with Angola's state registries (Registo Comercial, Criminal, Predial, Automóvel). For a compliance product, "10 registos" could read as ten official government registrations. |
| Ficha | Rejected | Breaks on locations — ficha de armazém reads as an inventory stock-card. |
| Ativo | Rejected | An accounting term for bens, direitos e recursos; employees are recursos humanos, never ativos on a balance sheet. |
| Unidade / Item / Objecto | Rejected | Too generic to feel concrete — and objecto quietly objectifies the person case. |
| Perfil | Rejected | Stretches awkwardly for equipment and locations; also collides with the user profile in the app. |
| Posto | Rejected | Collides with posto de trabalho (workstation / job position). |
| Crédito | Fallback | Mechanically works (OpenAI, Zapier precedent) but carries a consumable / refillable connotation, softens the compliance tone, and needs sales explanation. |
| Licença | Selected | Market-standard in Angola, fits all five categories uniformly, needs zero buyer education. See below. |
Menshen, Lda. (Microsoft Country Partner of the Year, Angola), SAP Angola and Adobe all use licenças / licenciamento in their Angola-facing materials. The buyer has seen it for years.
Angolan SME buyers already grasp a "licence pool" from Microsoft. No explanation needed — unlike crédito or ficha.
One licença reads naturally over a colaborador, viatura, equipamento, localização or cartão alike — no register-shifting.
The Microsoft Customer Agreement uses licença (unit of access) + subscrição (term wrapper) side by side — exactly our model.
The instinct to drop licença was to avoid Angola's royalty withholding. But that tax is triggered by the invoice wording, CAE code and contract — not the marketing noun. A rename buys nothing on tax while discarding the strongest, most familiar word. The tax is solved at the invoice (bifurcation) — see the Tax & Invoicing tab.
In practice the customer rarely meets an abstract noun. The calculator asks in concrete terms, the meter counts licenças, and entidade never leaves the back-end. Proposed Angola strings:
| Surface | Português (AO) | English |
|---|---|---|
| Pricing line | A sua subscrição inclui 100 licenças. | Your subscription includes 100 licences. |
| Calculator prompt | Quantos colaboradores, viaturas, equipamentos e localizações pretende acompanhar? | How many employees, vehicles, equipment and locations do you want to track? |
| Usage meter | 85 / 100 licenças em uso | 85 / 100 licences in use |
| Add action | Adicionar colaborador · viatura · equipamento… | Add employee · vehicle · equipment… |
| Empty state | Ainda não está a acompanhar nada. Comece pelo primeiro colaborador, viatura ou equipamento — cada um usa uma licença. | You're not tracking anything yet. Start with your first employee, vehicle or equipment — each uses one licence. |
The UI leads with the concrete category for the action ("Adicionar colaborador") and reserves licença for the count ("85 / 100 licenças"). That is what makes the "entidade" question quietly disappear.
The unit noun is not hard-coded. It's a terminology token resolved per market — so going international is a config change, not a rewrite of the app, invoices and copy. Three reasons this matters beyond ordinary translation:
pt-AO licença, pt-BR assinatura / registro, pt-PT licença — locale files alone don't capture market-specific word choice.
The royalty-withholding quirk is Angolan. The invoice-line token especially must vary by jurisdiction.
One token referenced everywhere — UI, invoice templates, copy — instead of the word baked into dozens of files.
| Market | Unit token (illustrative) | Invoice-line token |
|---|---|---|
| Angola (pt-AO) | licença / licenças | Serviço de subscrição · CAE 63110 |
| Portugal (pt-PT) | licença / licenças | Serviço de subscrição |
| Brazil (pt-BR) | assinatura / registros | Prestação de serviço (SaaS) |
| US / UK (en) | licence / seat / record | Subscription service |
Non-Angola rows are illustrative — each market's word and invoice wording get their own check before launch. The point is the architecture: terms are data, not constants.
→ The naming decision and the terminology-token design are recorded in DEC-035. The tax mechanics and exact invoice wording are in the Tax & Invoicing tab.
The pricing unit is a licença. Angola has a royalty-tax trap for anything that reads as a software licence — but it is triggered by the invoice, not the pricing word, and SaaS is genuinely a service. Below: the decision, the substance, the exact invoice wording, and where to corroborate every claim.
The full naming analysis (eight candidates, why licença won) is in the Naming the Unit tab. The short version: the instinct was to rename the unit away from licença to dodge Angola’s royalty tax. Independent verification says the opposite: the word is safe and market-standard; the tax is solved at the invoice.
Keep licença on the pricing page, contract and product UI (Microsoft/Menshen, SAP, Adobe all use it in Angola). Renaming does not avoid the royalty tax — it’s triggered by the invoice, not the marketing word. The real fix: word the invoice as a serviço de subscrição, code it CAE 63110, and reserve all IP in the contract.
A fungible billing slot, N per subscription. Called licença — keep it; it’s market-standard and needs no explanation.
A person, vehicle, equipment, location or card. Name the category directly (colaborador, viatura…). “entidade” stays only in legal text — it reads as “organisation” and collides with entidade legal (the company/NIF) in our own data model.
In the product UI you rarely need an umbrella word at all: count licenças and label the concrete categories. The “entidade” discomfort largely disappears.
Angola taxes software money in one of two buckets. The customer’s finance team decides which bucket by reading the invoice line — not the pricing page.
IAC royalty withholding, kept back by the customer at source.
Industrial-tax provisional withholding on services.
The fork is at the invoice — the document the tax team actually inspects. The name on the website never reaches that desk, so renaming the unit changes nothing here.
The service side isn’t a label we picked — it’s what docus.ao actually is. Two standard tax tests both put us there.
A payment is a royalty only when it buys the right to use a copyright (reproduce, distribute, modify, exploit). SaaS sits further from royalty than boxed software: the customer never receives a copy and gets none of those rights.
When the predominant value is a hosted service, the whole transaction is a service — and service payments fall outside the royalty definition in both the OECD and UN models.
| Boxed software reads as a licence | docus.ao reads as a service |
|---|---|
| Customer receives a copy, runs it on their PC | No copy delivered — runs only on our servers |
| One-time right to use the artifact | Continuous access to a live, hosted platform |
| Vendor’s job ends at delivery | Ongoing duty: hosting, databases, uptime |
| No further involvement | Updates, maintenance, support + compliance layer |
| Keeps working without the vendor | Stops the moment the subscription ends |
The only unfavourable reading is “royalty” (10%, non-creditable). The two realistic alternatives — ordinary service (6.5%) or “technical services” — both land in the favourable, creditable bucket.
Name hosting, maintenance, updates and support in the contract and on the invoice; never write “concedemos o direito de usar o nosso software.” Our case is clean because the product isn’t customer-customisable code — customisation is what blurs the line in cloud-tax disputes.
The protection lives in two layers. The friendly word and the tax-safe word coexist — exactly the playbook of every major software vendor in Angola.
| Document | Say “licença”? | If not, say… |
|---|---|---|
| Pricing page / calculator / contract body / UI | Yes ✓ | — |
| Invoice line item | No ✕ | Serviço de subscrição à plataforma docus.ao |
| Accounting / AGT code | No ✕ | CAE 63110 (data processing / hosting) |
| Credit notes · SAF-T · bank narrative | No ✕ | prestação de serviços |
You sell licenças but invoice and word them as a service. Amount = licenças × unit price, per period; the count still shows — as unidades de serviço, not “licenças de software.”
| Descrição | Qtd | Preço unit. | Valor |
|---|---|---|---|
| Serviço de subscrição à plataforma docus.ao Acesso SaaS para gestão documental de conformidade — Junho 2026. Capacidade contratada: 100 unidades de serviço (Cláusula 4). |
100 | 2 500,00 | 250 000,00 |
| ✅ Use on the invoice line | ✕ Never on the invoice line |
|---|---|
| Serviço de subscrição à plataforma docus.ao | Licença de software / de utilização |
| Acesso à plataforma SaaS · prestação de serviços informáticos | Aquisição de licenças |
| Unidades de capacidade de serviço (the count) | Direito de uso / de exploração do software |
Numbers illustrative. Some large/public customers also withhold part of the IVA (cativação, Lei 14/23) — handled the same way. The exact IVA-cativação % is a detail for the accountant/law firm to confirm.
Three independent AI research passes (Apr 2026) converged; a fresh June re-check then verified each claim against PwC, EY and the statutes. Every load-bearing claim with a source:
| Claim | Corroborate at | Verdict |
|---|---|---|
| Royalties → 10% IAC, withheld by the paying customer if local | PwC — WHT | Confirmed |
| Services (resident) → 6.5%, creditable (Lei 26/20) | PwC flash · Lei 26/20 | Confirmed |
| IVA standard rate 14% | PwC — Other taxes | Confirmed |
| AGT-certified e-invoicing (DP 71/25) | EY Angola | Confirmed* |
| SaaS taxed by substance (service vs royalty), not the label | Vistra · CIAT | Confirmed |
| Circular 01/GACA/GJ/AGT/2026 excludes IP-licence payments from CEOC | EY — Circular 1/2026 | Partial |
* DP 71/25 confirmed real; only a date detail differs (effective 1 Oct 2025). “Three reports” = three AI runs, not a lawyer — which is why this leans on PwC/EY and why a law-firm sign-off is the final step.
The invoice fix is strong protection, not a 100% guarantee: there is no binding Angolan ruling on SaaS-as-service, and a very cautious customer could still withhold 10%. We treat it as risk-reduction and keep a contractual tax-gross-up. One short opinion from an Angolan law firm (ALC / PRIME) to bless the invoice wording is the recommended next step — worth far more than any rename.
① Keep subscrição + licença customer-facing. ② In the UI, count licenças and name categories — no customer-facing “entidade”. ③ Implement the invoice bifurcation (service wording + CAE 63110 + IP reservation + gross-up). ④ Get one law-firm opinion, then lock the wording.