# Archetypes — Six Parent Jobs **Method**: bottom-up clustering from the open-coding pass on 128 real parent quotes (`open-coding-analysis.md`), anchored in three established frameworks — Self-Determination Theory (Deci & Ryan), Expectancy-Value Theory (Eccles), Jobs-To-Be-Done (Christensen school) — and cross-checked against parental educational anxiety research (Guo et al. 2024, 2025). Each archetype corresponds to a distinct *job* a parent hires the Value Hub to do, not to a person or a personality type. **Critical framing**: these are jobs, not people. A single parent will occupy different jobs at different moments. A Russian father at onboarding is typically a Navigator. The same father during his child's plateau is typically a Reassurance-Seeker. Same father at renewal month-12 is typically an Investor. The dashboard serves the *active job*, not a static classification of the parent. --- ## 0. Why six, and why these six Open coding surfaced ten candidate groupings when I tagged the 128 quotes by underlying parent motivation. Of those ten, three were not jobs but *emotional states* (anxious, betrayed, proud) that can attach to any of the others. One ("The Quitter") turned out to be a downstream consequence of the other jobs failing, not a distinct job. Two collapsed into each other on closer reading ("The Protector" and "The Co-Pilot" were the same participation-without-burden job viewed from different angles). The six that remain each satisfy three tests: 1. **Distinct job verb**. Each answers a different question the parent is silently asking. 2. **Distinct need profile in SDT**. Each stresses a different combination of autonomy/competence/relatedness. 3. **Distinct failure mode**. Each can fail in a way that the others wouldn't — and those failure modes are visible in the quote corpus. The six are presented in rough order of frequency of underlying job-activation across the corpus, though frequency here is descriptive of the *reviews we have*, not of the population. --- ## 1. The Reassurance-Seeker ### Core job *"Tell me I'm doing the right thing as a parent."* The parent is not asking whether Novakid works in an abstract sense. They are asking whether *they*, as a parent, are making good decisions for their child. The investment is emotional before it is financial. ### Needs profile (SDT) - **Primary**: parent-side **competence** — "am I effective as a parent?" - **Secondary**: **relatedness** to teacher as proxy for being connected to child's learning ### Value frame (EVT) - **Primary**: **attainment value** — being a good-enough parent is identity, not just outcome - **Secondary**: **intrinsic value** — child's visible enjoyment is proof the investment is right ### Mental model of progress Progress = "my child is fundamentally OK, learning, and I am contributing by providing this." The specifics of *what* they learned matter less than the underlying narrative of a well-supported child. ### Primary anxiety *"I might be failing my child — either by doing too little, or by paying for something that isn't helping."* This is the anxiety Guo et al.'s PEAS scale captures in its Learning & Development dimension. ### Journey moments where this job dominates - Plateau weeks (months 2–8) — "is this still working?" - Post-milestone — "was that progress real?" - Family-dinner moments — "my sister asked if Novakid is worth it" - After a teacher change — "can I still trust this?" ### Representative quotes from corpus > *"Thanks to the teacher for his hard work... Mom really feels that Rita has made a lot of progress. Thank you, teacher."* > — Rita mum, Trustpilot > *"The meetings are built very well to keep the kids interested. We have been using it for 2 months and we see a great improvement on our six year old."* > — Keren A, Trustpilot > *"We've been continuing with Novakid for two years now. My child's self-confidence has significantly improved."* > — Ozge Gungor Ulug, Trustpilot Turkey ### Design implications for Value Hub - **One specific observation** as hero, not a paragraph of data. "Mia self-corrected 'goed' to 'went' this week" beats "Great progress." Specificity is the engine of reassurance, not warmth. - **Teacher presence in the artefact** — the teacher's name and voice in the weekly message, not a faceless AI. - **Proactive plateau naming** — the Reassurance-Seeker during plateau is the single largest churn risk. A proactive "this is normal, here's why, here's when it breaks" message is the highest-leverage intervention in the product. - **Child's enthusiasm captured and shown** — the child's own desire to continue is this job's strongest proof. ### Failure modes (what actively harms this job) - Harsh metrics presented without context. - Peer benchmarking — "your child is below average" hits this job directly in the anxiety centre. - Cost-per-word framing — instrumentalises the emotional investment. - Uniformly positive feedback that is too bland — paradoxically *amplifies* rather than reduces anxiety because it reads as untrustworthy (see Theme B in open coding). --- ## 2. The Auditor ### Core job *"Show me the evidence so I can judge for myself."* The Auditor approaches the product as a professional approaches a report: with a baseline suspicion, an openness to being convinced, and a requirement that claims be falsifiable. This job is not cynical — it is rigorous. ### Needs profile (SDT) - **Primary**: **competence** — as evaluator of the product, not as parent - **Secondary**: **autonomy** — the right to reach their own conclusion without being nudged to a conclusion the product wants ### Value frame (EVT) - **Primary**: **expectancy** — "is there a believable case that this is working?" - **Secondary**: **utility** — connection to external outcomes (school, exams, career) ### Mental model of progress Progress = "verifiable data points that can be checked independently." Not "the platform says so" but "here is the evidence, and here is how I would know it's wrong." ### Primary anxiety *"I am being sold something that doesn't work, and the sales gloss is making it hard to tell."* This is the anxiety that produces the DIY-tracking behaviour (Notion databases, notebooks, phone recordings) documented in the corpus — a compensatory mechanism the parent builds when the platform's evidence stream fails their audit. ### Journey moments where this job dominates - Month 2–3 evaluation window — "is this working well enough to keep paying?" - Renewal decision points - Post-plateau audit — "was the plateau a phase, or is this a dead end?" - Before paying for a price-tier upgrade ### Representative quotes from corpus > *"There is no method that is called LEAD-IN from the teacher. The child experiences great difficulty to understand WHY HE IS DOING this or that during the lesson. Thanks to my good level of English I must repass all the didactic materials before the coming lesson..."* > — Margarita, Trustpilot Italy > *"It might be good for beginners but not for children who can already speak English fluently. She spent 4 lessons on colors and honestly didn't learn anything."* > — Giulia Trivero, Trustpilot Italy > *"Notifications to parents are always positive. Even if the lesson went badly, everything is rose-coloured. They avoid giving realistic feedback."* > — realjaew, Ekşi Sözlük > *"Не хватает более углубленного изучения грамматики."* > *[EN: Not enough depth in grammar instruction.]* > — Елена, progbasics.ru ### Design implications for Value Hub - **Depth layer mandatory**. A layer-3 full analytics view must exist for this job to be served. Progressive disclosure is not a nice-to-have; it is what separates the product from a toy dashboard for Auditors. - **Calibrated honesty**. Feedback must include both strengths and areas to reinforce. A teacher report that only praises is the product's worst signal to this job. - **Falsifiability as a design principle**. Claims must be specific enough that the parent could disprove them. "Great week" cannot be disproved; "self-corrected 'goed' to 'went' in lesson 14" can be — and that's exactly why it is trusted. - **Triangulation hooks**. Let the Auditor see the same signal from three angles (AI summary + teacher note + lesson recording reference) so the conclusion feels independently supported. ### Failure modes (what actively harms this job) - Warm summaries with no underlying data. - Uniform positivity (Theme B of open coding). - Missing depth layer — a simple dashboard actively signals a shallow product. - AI-generated narrative without source traceability. --- ## 3. The Investor ### Core job *"Defend the money to me."* The Investor asks a financial question, not a pedagogical one. They are not indifferent to quality; they are asking whether the quality they are getting justifies the price they are paying. Cost is central to their mental model in a way it is not for the other jobs. ### Needs profile (SDT) - **Primary**: **autonomy** as financial decision-maker — the right to say no without friction - **Secondary**: **competence** — enough information to defend the decision to themselves and to family ### Value frame (EVT) - **Primary**: **utility** minus **cost**, explicitly — the full EVT calculus is running - **Secondary**: **attainment** — investing in the child's future is part of parent identity in many markets ### Mental model of progress Progress = "measurable skill gain per unit of money invested." The Investor converts outcomes to currency and compares across alternatives (group classes, school English, tutoring, apps). ### Primary anxiety *"I am throwing money into a black hole I can't see into."* When this job fails, the parent doesn't just leave — they leave *angry*, often with the emotional register of "scammed" or "cheated" (corpus-evident: Rafal, Andrey, Jeanne Alexandre, Rafi Camhi). ### Journey moments where this job dominates - Pre-commitment — "is this worth my money?" - Renewal — "is the next 12 months worth renewing at this price?" - Price hikes or currency changes - When the child goes through a quiet period and the subscription feels like ongoing drain - Cancellation friction — this is where the Investor becomes an *angry Investor*, which is the single hardest retention scenario in the product ### Representative quotes from corpus > *"Lessons are super pricey — 20 euros for 25 mins. Lessons are standardized and teachers don't take into account of the students' level."* > — Giulia Trivero, Trustpilot Italy > *"Чем больше вы углубляетесь в программу, тем выше цена."* > *[EN: The deeper you go into the program, the higher the price.]* > — Nashorn1, otzovik.com > *"Цена обучения выросла значительно. Проблемы с оплатой."* > *[EN: Tuition price grew significantly. Billing problems.]* > — Галина, tbank.ru > *"I seriously dislike the subscription system — automatic renewal is the only option... I am fed up and I already blocked future payments."* > — Rafal W, Trustpilot ### Design implications for Value Hub - **Renewal-defence artefact**. A single page or shareable card that lets the Investor justify (to themselves, to a spouse) why this subscription is continuing. This is the job H7 is really serving in the shareable-progress hypothesis — social validation is one dimension of that, but financial self-justification is another. - **Cost-per-outcome math available** — but NOT as dashboard default. The Investor in Italy or Turkey wants it front-and-centre; the Reassurance-Seeker in France or Japan would be alienated by the same view. This is one of the clearest cases where cultural overlay matters operationally. - **Clean exit path**. The Investor must be able to pause and cancel without hostility. Cancellation-friction *erases* all earned value and converts a neutral parting into an angry one that generates the worst reviews on every regional Trustpilot domain. - **Transparent billing**. Auto-renewal that surprises is this job's single deepest wound. ### Failure modes - Hidden auto-renewal, pause-doesn't-pause-billing, refund stonewalling. These are the single most common 1★ review cause across markets (Theme E, trust stack). - Price escalation without explanation — RU-specific but generalisable. - Making cost information hard to find while making engagement features easy to find — this signals "we want your time and money, but don't want you to measure either." --- ## 4. The Navigator ### Core job *"Show me the path ahead so I can prepare."* The Navigator wants a roadmap. Where is my child now, where are they going, what comes next, when will it get harder. This job treats language acquisition as a *journey with visible stages*, and it is thwarted most by curricula that are opaque, adaptive-in-name-only, or that jump difficulty without warning. ### Needs profile (SDT) - **Primary**: **competence** — planning capacity, understanding of trajectory - **Secondary**: **autonomy** — the ability to course-correct before problems compound ### Value frame (EVT) - **Primary**: **expectancy** of future outcomes — "can I see a credible path to where my child needs to be?" - **Secondary**: **utility** — alignment with external milestones (school English, Cambridge exam, emigration) ### Mental model of progress Progress = "trajectory across a known curriculum with visible checkpoints." Plateau is acceptable *if named and explained*; plateau without acknowledgement is the Navigator's worst failure. ### Primary anxiety *"I will miss something important because I can't see what's coming, and by the time I notice, it will be too late to fix."* This is the anxiety that produced Nejat's 6-month 12,553 TL complaint — he didn't stop paying because the lessons were bad; he stopped when he discovered (via an *external* English teacher) that the level hadn't moved. ### Journey moments where this job dominates - Onboarding — placement test, first curriculum map - Curriculum transitions (e.g., A1 → A2) - Pre-plateau window (months 2–4) — "will I be told when something changes?" - School-year boundaries (summer break, school-year start) — alignment concerns - Exam prep windows (Cambridge, school tests) ### Representative quotes from corpus > *"The program was supposed to begin at Level 1 and progress according to the child's ability... I realized that without my intervention, advancement to levels 2 or 3 wouldn't have occurred."* > — Nejat, Şikayetvar (paid 12,553 TL over 6 months; independent teacher confirmed child stayed at same level) > *"Программа сильно усложнилась... все эти фразы, кроме I've got, были абсолютно новыми!"* > *[EN: The program became much harder. All these phrases except 'I've got' were completely new!]* > — Lyudmila, tutortop.ru > *"There was no accurate assessment done for my kid and also the topics of the classes was very similar and no clear plan sent to us."* > — ALFAZANI TALHA, Trustpilot > *"The only issue is that no placement test is conducted at the beginning of the course, and this aspect could be improved."* > — pelma pelma, Trustpilot Turkey ### Design implications for Value Hub - **Curriculum roadmap as a first-class surface**. Not hidden in an "Achievements" tab; surfaced on layer 1 as "where you are, what's next, when to expect harder material." - **Placement test as an onboarding commitment**. If the Navigator's opening experience is "the teacher will figure it out eventually," the job has failed before the first lesson. - **Pre-plateau warning**. The Navigator must be told, *before* the plateau arrives, that this is a known phase of language acquisition, when it typically starts, what it looks like, and why it's not a reason to cancel. This is the British Council plateau research operationalised into product UX. - **Difficulty-jump signalling**. When the curriculum is about to introduce materially new grammar or a step-change in vocabulary scope, a pre-lesson heads-up to the parent and child. Removes the Lyudmila moment. - **External alignment hooks**. Show how the current curriculum position maps to Cambridge YLE, school-English syllabus, CEFR observable behaviours. ### Failure modes - Curriculum opacity — "what's next?" is unanswerable. - Level mechanics that require teacher escalation (the Nejat pattern). - Unannounced difficulty jumps. - Stale placement assessment. --- ## 5. The Co-Pilot ### Core job *"Give me a role I can play without becoming the homework manager."* The Co-Pilot wants involvement, but bounded. They are not willing to take on the teacher's role (and resent being forced to). They are willing to contribute 30 seconds to 5 minutes a week if that contribution is meaningful and clearly scoped. This is about *actionable participation*, not emotional register — a stoic father and a warm mother can both be Co-Pilots; the job is the same. ### Needs profile (SDT) - **Primary**: **relatedness** — connection to the child's learning journey - **Secondary**: **autonomy** — the right to contribute *or not*, without guilt ### Value frame (EVT) - **Primary**: **attainment value** — being an involved parent, without the cost being overwhelm ### Mental model of progress Progress = "something I can notice, celebrate, or contribute to in small daily moments." The Co-Pilot values the real-world "aha" — child speaking at dinner, reading a sign, translating for grandparent — because it is the shared moment of progress, not a notification from the platform. ### Primary anxiety *"I will either neglect my child's learning because I'm busy, or I'll become the homework parent I don't want to be."* This is the anxiety that produces the asymmetric-feedback complaint (Carmentxu, realjaew) — the Co-Pilot wants to *contribute* context but has no channel to do so. ### Journey moments where this job dominates - Post-lesson window — "what can I reinforce at home in 2 minutes?" - Between-lesson gaps — "is there a non-homework way to participate?" - Real-world moments — "my child said something in English at the zoo, I want to mark it" - Before a lesson — "the child is tired / anxious / excited; I want the teacher to know" ### Representative quotes from corpus > *"My kid is learning a lot with Novakids. He is improving his English knowledge. But there is one thing that I don't like: they everyday give me feedback, but I can't. I would like to write to the teachers to give them my feedback and I can't do it."* > — Carmentxu, Trustpilot Spain > *"Thanks to my good level of English I must repass all the didactic materials before the coming lesson... WHY must I develop interest in my child to follow your course?"* > — Margarita, Trustpilot Italy (this is the Co-Pilot job *failing* — being forced into the homework-manager role) > *"I record Elif speaking English every month on my phone."* > — Ayşe, Istanbul (from 45-persona synthesis — synthetic, but captures the real pattern of DIY participation) > *"Ребёнок стал говорить фразочки, понимать, что говорю я в отпуске на английском."* > *[EN: The child started speaking short phrases, understanding what I say in English on vacation.]* > — Avotna, kursfinder.ru ### Design implications for Value Hub - **Bi-directional teacher channel**, bounded. Parent can send 1–2 sentences of context per week; teacher sees; teacher optionally acknowledges. Not a chat. Not homework. Just a lightweight loop. - **"Capture a moment"** affordance — parent reports a real-world English moment in 10 seconds, it enters the progress record, it threads into the AI narrative. Converts scarce anecdotal evidence into structured data. - **30-second pre-lesson prompts** for co-pilot participation: "1 thing you could tell the teacher before tomorrow's lesson" — optional, ignorable. - **Shareable milestones with family**, not just spouse — the Co-Pilot participates through the social frame of "the family watching the child grow." ### Failure modes - Requiring parent-as-tutor — immediately produces the Margarita reaction. - One-way feedback — the Carmentxu failure. - Homework assignments that go to the parent instead of the child. - Making engagement the performance metric — "you checked the dashboard X times this week" is an anti-pattern. --- ## 6. The Outsourcer ### Core job *"Just work — don't make me manage this."* The Outsourcer has explicitly decided to delegate their child's English to Novakid. They don't want dashboard, they don't want dialogue, they want it to run. This is not disengagement — it is an active choice to protect their attention for other priorities (work, younger sibling, their own learning, health). ### Needs profile (SDT) - **Primary**: **autonomy** — the right to *not engage*, knowingly - **Secondary**: **relatedness** via the trust placed in the teacher — the Outsourcer trusts by delegation ### Value frame (EVT) - **Primary**: **cost** (time, cognitive load) minimised - **Secondary**: **expectancy** — implicit, "it's running so I assume it's working" ### Mental model of progress Progress = "the thing is happening; I would notice if it stopped." The Outsourcer does not track deltas — they track *interruptions*. A quiet week is a good week. A teacher-change email is the first time they look. ### Primary anxiety *"Something has gone wrong and I don't know, because I'm not checking."* This job is vulnerable specifically because the Outsourcer is *not* reading dashboards — so a product that assumes dashboard-visiting is an engagement measure will miss the Outsourcer entirely. Their churn is silent: no angry reviews, just one day they don't renew. ### Journey moments where this job dominates - Steady-state (6 months+) - Parents of multiple children splitting attention - Parents managing their own demanding job or study - Late-stage (2+ year users) who have stopped engaging because engagement isn't required ### Representative quotes from corpus > *"Зеро problems with English at school, absolutely none. Just one lesson a week with a premium teacher — excellent result."* > — Pantyusha5, kursfinder.ru > *"We join Novakid school for more than 4 years. Excellent choice."* > — Ahmed Zelikha, Trustpilot (the entire review is eight words — this is Outsourcer writing) > *"Мой сын радуется и хочет заниматься дальше."* > *[EN: My son is happy and wants to continue.]* > — Диана Микель, azbukakursov.ru The *absence* of dashboard complaints in the corpus is evidence for the Outsourcer job — long-term users who don't critique the dashboard because they don't look at it. ### Design implications for Value Hub - **Invisible operation default**. No engagement prompts. No "your dashboard has new data" nudges. No gamification for parents. - **Anomaly-based alerts only**. The Outsourcer should hear from the product exactly when something warrants attention: plateau warning, teacher change, price change, child's absence pattern, first meaningful milestone. - **Weekly digest is the product surface**, not the dashboard. The email *is* the hub for this job (echoing H6 in the current hypotheses — that hypothesis survives the rework because it addresses a real job, not a fabricated persona). - **Trust-keeping rituals**. Quarterly "here's what happened, nothing requires your action, delete this email guilt-free" messages. Make the non-engagement pattern compatible with continued subscription. ### Failure modes - Engagement-seeking UX. Badges, streaks for parents, "you haven't visited" nudges — all actively hostile to this job. - Hiding anomalies behind dashboard visits — a plateau warning that sits in the dashboard waiting to be discovered is invisible to the Outsourcer and therefore useless. - Mistaking Outsourcer silence for Outsourcer satisfaction — many churned Outsourcers were actually dissatisfied but couldn't be bothered to complain. --- ## Part II — How the jobs relate ### One parent, multiple jobs over time A parent's *active job* shifts with context. Typical shifts from the corpus: | Moment | Likely dominant job | |---|---| | Trial lesson decision | Investor + Auditor | | First 4 weeks | Navigator (placement concerns) + Reassurance-Seeker (first wins) | | Month 2–4 (pre-plateau) | Navigator (curriculum) + Co-Pilot (home reinforcement) | | Plateau (months 2–8) | Reassurance-Seeker (dominant), Auditor (if specific doubt) | | Milestone breakthrough | Co-Pilot (share the moment), Reassurance-Seeker (reward) | | Renewal decision | Investor (dominant) + Auditor (supporting) | | Steady-state (9+ months) | Outsourcer (dominant), Navigator (curriculum transitions) | | After teacher change | Reassurance-Seeker (re-validate) | | Year-end | Investor (annual review) + Reassurance-Seeker (gratitude) | **Design corollary**: the dashboard must detect or infer the active job and serve it, rather than treating the parent as a static persona. Progressive disclosure is one mechanism (Auditors drill down, Reassurance-Seekers stop at the top layer). Contextual prompts are another (plateau prompt triggers Reassurance-Seeker support content). ### Jobs that coexist in tension - **Investor vs Reassurance-Seeker** — the Investor wants cost-per-outcome math; the Reassurance-Seeker finds it cold. Cultural overlay determines which dominates as default, but both can be present in the same parent. - **Navigator vs Outsourcer** — the Navigator wants roadmap visibility; the Outsourcer wants not-to-look. Resolved by surfacing roadmap *on demand* (available when Navigator-mode activates), not pushing it (which would alienate Outsourcer-mode). - **Auditor vs Reassurance-Seeker** — the Auditor wants calibrated honesty including weaknesses; the Reassurance-Seeker can be destabilised by explicit weakness reports. Resolved by *specific, bounded* honesty — name one strength and one area to reinforce, never a laundry list. These tensions are the reason a single-mode dashboard fails for this product. The dashboard is not for *a* parent; it is for *a set of jobs*, some of which tension each other. --- ## Part III — Cultural overlay, briefly This section is a pointer to `cultural-analysis.md`, not a full treatment. The point to make here is: *Jobs are universal across markets. Proof-types are culturally specific.* | Job | Proof that Russian markets accept | Proof that Italian markets accept | Proof that Turkish markets accept | Proof that Polish markets accept | |---|---|---|---|---| | Reassurance-Seeker | *"Best in class"*, school-teacher endorsement | Teacher relationship warmth | Child's confidence in real-world moments | *"A2 after 3 years"* CEFR milestone | | Auditor | Olympiad results, external teacher verification | Lesson-level transparency, grammar depth | Placement test rigor, recording review | Systematic skill tracking | | Investor | School grade → "zero problems at school" = ROI | €/lesson math, explicit value-per-euro | Package-price calculus | Cost-per-CEFR-step | | Navigator | School-year alignment | Cambridge exam alignment | Travel readiness | CEFR roadmap to B1/B2 | | Co-Pilot | Family storytelling (olympiads to tell grandparents about) | Food/travel moments | Confidence moments abroad | Formal school preparation | | Outsourcer | *"Does it, we don't think about it"* | Long-term loyal minimalist reviews | *"Excellent choice"* short reviews | Multi-year quiet renewals | **Design corollary for the Value Hub**: one architecture, culturally-tunable proof-display. Not seven dashboards. One dashboard whose proof layer surfaces the resonant proof-type per market. --- ## Part IV — Validation plan (inline, per convention) This archetype framework is a *hypothesis* grounded in 128 real quotes + established frameworks. It has not been validated against real parent interviews. Before production commitment, the following validation should happen: 1. **Card-sort validation with 12 real parents** across 3 markets (RU, IT, TR), asking them to sort 40 real-quote cards into groupings of their own making. Compare their clusters to the six jobs. Cost: ~$800 via UserInterviews.com; timeline: 2 weeks. 2. **Job-to-moment validation** — in-product survey asking current Novakid parents to indicate which statement best describes why they opened the dashboard in their last visit. Statements are written as the six jobs. Response frequencies give a directional prevalence by journey stage. 3. **Cultural proof-type A/B** — within a single archetype (e.g., Reassurance-Seeker), test two proof framings per market. Measures: 7-day retention delta, NPS delta, click-through to depth layer. Until these complete, the six jobs are a *design hypothesis*, not validated empirical clusters. Every load-bearing design decision that rests on them should carry the same caveat.