windflow.ai windflow.ai
India · one decision, then fifteen years of its consequences · mobile and television · advertising only

The Decision, and Fifteen Years of It

One business model survives the brief; the rest is arithmetic.

—
Year 15 cash result
—
Peak cash needed
—
Breakeven
—
Year 15 unique reach
—
Mobile / TV revenue split
—
Year 15 net revenue
—
of India’s digital video ad market
01

The problem, the business, and every number we assumed

This page is the whole model written out in words. The figures in bold are live: change anything on the tabs that follow and these sentences change with them.

How to use this simulatorsix tabs · every figure is live · nothing is hard-coded

This is a calculator, not a document. Every figure on every tab is worked out from the numbers in the boxes. Change a box and the whole model moves with it — the seven figures in the bar at the top, the tables, the charts, and the sentences.

  1. Start on this tab. It says what the business is, and lists every assumption the model makes with a tag saying whether that number was measured, worked out, or invented.
  2. The bar at the top is the answer. Those seven figures stay on screen whichever tab you are on, and update as you type.
  3. Tabs 02 to 05 are the business in order — get the films, buy the audience, deliver the video, sell the advertising. Each one holds the numbers for that step and shows what it costs.
  4. Tab 06 is the result. It opens with a written summary of what this run says, then the overheads, the fifteen-year profit and loss, and three charts.
  5. To change something, type in a box. Click in, type a number or use the arrow keys. Two fields are greyed out because they are calculated rather than typed.
  6. Every input in one place: the Assumptions panel at the foot of this tab holds all of them together. It is the quickest way to drive the model, and a number changed there changes on its own tab too.
  7. Headings with a triangle open and close, and hovering the small mark beside a figure says where that figure came from — blue a published measurement, brass our own inference, red no published figure anywhere.
  8. Nothing is saved and nothing is sent anywhere. There is no undo: reload the page to put every number back where it started.
The problem we were given

Build an application for mobile and television, carrying a free film and serial library, entirely on demand, funded entirely by advertising — and own the customer relationship directly.

Four things in that sentence are fixed, and each one closes off an option.

  • Both screens, one service. Nothing that works on only a phone, or only a television.
  • Your app, your account, your player. Not a supplier sitting on somebody else's shelf.
  • On demand only. No scheduled channels — not ours, not anyone's.
  • Advertising is the only money. No subscriptions, no rentals, no selling the films, no commerce.
How the business would work

Studios hand over their films and serials and are paid nothing for them. Instead they take a fixed share of whatever advertising those films go on to earn. So the library costs nothing to get, and everything it earns is split.

Money goes out in four places. Once, to get the films ready to stream. Every year, to buy viewers — every year, because most of them leave and have to be replaced. Every hour anyone watches, to send the video to them. And all the while, the cost of running a company.

Money comes in one way only. Advertisers pay for every thousand ads shown. The same hour of watching is worth more on a television than on a phone, because advertisers pay more to reach a big screen.

At the current settings, here is year —. Revenue is —. Buying viewers costs —, delivering the video —, and running the company —. The library cost — once, at the start. What is left is —.

What we assumed — every claim, and what it rests on —
What backs each line Source a published measurement Derived inferred, or a planning choice we made No source no published figure exists anywhere Result computed, not assumed
The library
— hours and — titles in the library at launch.
DerivedA planning judgement. No catalogue of this shape has been counted, and everything downstream scales off the hours total.
Nothing is paid to acquire any of it.
DerivedThe deal, not a market rate. It is the premise of the whole plan rather than a finding.
— an hour to make it streamable — — in all.
DerivedA planning figure, not a quote. Assumes automated ingest and light QC on clean masters, and that subtitles arrive with them. Human subtitling in India is quoted in the low thousands per hour, which would make this line ₹30 cr+ and the largest cost in content.
—% of the launch catalogue added every year after that.
DerivedA planning figure. Nothing measures how fast a free catalogue goes stale.
The deal with the studios
Studios take —% of net advertising revenue, across both screens, for the life of the deal.
No sourceNo Indian FAST content deal has published terms. This is a placeholder for a negotiation, not a benchmark — and it is the only input that moves mobile and television together.
Nothing is paid up front, and almost nothing in a bad year.
DerivedStructural to a revenue-share deal. The downside is capped; the upside is shared forever.
Rights revert when the term ends — the catalogue is never ours.
DerivedStructural. Makes term length, exclusivity and the renewal ratchet worth more attention than the rate.
In year — the studios are paid —; over the full period, —.
ResultFollows from the rate and the revenue above it.
Buying the audience
— of marketing in year one.
DerivedA decision, not a forecast. The model shows the level of spend is linearly value-destroying at every allocation.
Growing —% a year until year —, then flat — a plateau near —.
DerivedA planning choice. The taper is there so nothing compounds for fifteen years unchecked.
—% of it goes to mobile, the rest to television home screens.
DerivedA decision. Moving it across its whole range changes peak funding by about ₹42 cr — this was never where the answer lived.
A mobile user still watching costs —.
DerivedMidpoint of a 14× band built from one agency's portfolio. Meta real-spend data puts the India median install at ₹112, above the ₹25–70 OTT band usually quoted.
That cost rises —% a year on both screens.
DerivedAn FX forecast, not auction inflation — USD/INR 83 → 95.4. Auction evidence is two-sided: APAC +9.4%, India Meta −35% in USD.
A television household still watching costs —.
ResultNot typed in anywhere. It falls out of the three numbers below.
The way onto a television
A home-screen advertisement costs — per thousand views.
SourceSamsung India's published card is ₹510; trade press reports ₹300–500.
— people in every thousand who see it install the app.
No sourceNobody anywhere publishes this. The entire range of the television plan lives in this one number.
—% of those installs are still watching later.
SourceSamsung: 70%+ of TV app users last three months or less.
An install therefore costs —, and a household that stays costs —.
ResultAt one install per thousand it is —; at eighteen, —. An eighteen-fold band from the unpublished number above.
We pay for no preloads, no ad swaps, and no button on the remote.
DerivedA decision. Every one of those is priced on request per OEM; none publishes a figure.
Being listed in an app store is free and brings almost nobody.
SourceThe average smart television uses 3.6 applications a month, and half the 43,500 in the Roku store have never been rated.
What a viewer does once they arrive
On mobile, —% watch — hours a month, —% watch —, —% watch —, and —% watch —.
No sourceNo Indian source publishes viewing concentration — Ormax measures reach only, BARC's panel is 0.025% of TV homes, platforms disclose nothing.
They come back at —%, —%, —% and —% a year — blended, —.
No sourceUnmeasured. For scale: AppsFlyer puts Android D30 retention in India at 3.8%, the lowest of any country measured, and subcontinent 30-day uninstall at 64–66%.
On television, —% watch — hours a month, —% watch —, —% watch —, and —% watch —.
No sourceUnmeasured, same reason. Blended it is — hours a month against mobile's —.
They come back at —%, —%, —% and —% a year — blended, —.
No sourceUnmeasured. Set below mobile at every tier — Samsung's finding that most TV app users last three months or less is the only support, and it is the conservative direction in any case.
The splits reconcile to Amazon MX Player's 50 minutes a day on mobile and 80 on connected television.
DerivedThe only single-service India comparison that exists, and it holds only at a DAU/MAU ratio of about 0.20 — which no Indian service publishes either.
The two screens together
—% of mobile accounts sit in a household that also has the app on its television.
No sourceNo Indian data exists. A pure assumption.
Each television household bought brings — mobile accounts along free.
No sourceUnmeasured. The only adjacent figure anywhere is 2.5 viewers per CTV impression.
Each mobile account brings — television households.
No sourceUnmeasured, and weaker by construction.
Delivering the video
Mobile streams at 640 × 360 — — GB an hour.
SourceApple HLS Authoring Specification. The 480p cap would be 0.55.
Television streams at 960 × 540 HEVC — — GB an hour.
SourceSame ladder. 720p HEVC would be 1.14, which is where television delivery stops paying.
Bandwidth costs — a GB.
DerivedAirtel is the only Indian telco publishing a price: ₹6 for the first 10 TB a month, down to ₹3 above 150 TB. This assumes a negotiated rate well under the card. A telco cache deal would be ₹0.20–0.50.
That is — an hour on mobile, — on television, — in year —.
ResultA fixed rupee cost per hour — it does not shrink when the ad price falls.
Selling the advertising
— minutes of —-second spots an hour on mobile — — slots.
DerivedA planning choice. TRAI caps broadcast at 12 minutes; streaming is not bound, and MIB removed television's cap in August 2026.
— minutes of —-second spots on television — — slots.
DerivedThe same slot count as mobile, so the two screens differ by price alone.
— of the slots we run are sold.
DerivedA planning figure describing the sales team, not the screen.
The ad-tech chain takes — of what is sold.
DerivedBlended across four channels, from 5% on direct-sold to 35% on open exchange.
One sales team, one rate card, one channel mix across both screens.
DerivedAn advertiser buys the audience, not the device. But holding the mix identical is the model's weakest structure — Indian mobile video clears mostly through open exchange while connected TV can be sold direct, so the screen difference is partly mix, not only price.
A television impression clears — per thousand after fees; a mobile one — — mobile at —% of the television card.
No sourceTwo Indian agencies disagree threefold on connected-TV CPM — ₹150–350 against ₹450–1,200. Both screens scale off this one card, so an error here moves everything together.
Running the company
— a year, flat, for all of finance, legal, compliance and office.
DerivedA planning simplification, not a build. Flat says the company does not grow as the service does.
The horizon
— years.
DerivedLong enough for retention and re-buying to show, which five years hides.
Growth rates on budget and acquisition cost stop compounding after year —.
DerivedA planning choice — nothing compounds for fifteen years in reality.
Nothing is capitalised, so what the model calls profit is also cash.
DerivedStructural. It is why the year-15 figure and the cash figure are the same number.
The five numbers that rest on nothing

Every figure above was either measured, worked out, or invented. These five were invented, and between them they decide the answer.

  1. The revenue share. No Indian content deal of this kind has published terms. The rate is a placeholder for a negotiation, not a benchmark.
  2. The install rate on television. How many households install the app after seeing it. Nobody publishes this, and on its own it moves the cost of winning a household eighteen-fold.
  3. The cross-screen numbers. How many people use both screens, and how many phones arrive free with a television. Neither has an Indian source at all.
  4. The viewer splits. How the audience divides into heavy, medium and light viewers. Indian sources publish how many people watch, never how much each one watches.
  5. The television rate card. What an advertiser actually pays for a big-screen ad. Two Indian agencies quote prices three times apart, and the only published card is less than half what we assume.

The tabs that follow mark every figure with where it came from, and say plainly where nothing came from. Hover any mark to read it.

Assumptions — every input in the model, in one place29 inputs · live

Every number the simulation runs on. These are the same fields that sit on their own tabs — change one here and it moves there, and the whole model with it. The television cost per household is derived, not typed. Hover any marker for what the figure actually measures.

02Content library creation
03Demand side creation — budget and allocation
03Demand side creation — cost per user
03Demand side creation — the television chain
03Demand side creation — cross-screen effect
04Content delivery to demand
05Monetisation
06Simulation
02

Content library creation

Evidence key Source published measurement — click through Derived inferred, or a planning choice No source no published figure exists anywhere Hover any marker for what it actually measures.

—

Acquisition universe—
CategoryDealUnits Hrs/unitHours% of library
Total— 100%
—
Unique hours
—
Titles in the library
—
Onboarding & QC at launch
—
Studios’ share of net revenue
—
Paid to studios in Year 15
—
Paid to studios, 15-year total
Deal terms—
03

Demand side creation

Evidence key Source published measurement — click through Derived inferred, or a planning choice No source no published figure exists anywhere Hover any marker for what it actually measures.
3.1

How much each user demands — mobile

Tiers—
Tier% of new accountsHrs/month = min/dayAnnual retentionSteady base% of hours
Blended—— ——— 100%
Calibration
—
Where every number in this table comes fromcell by cell
CellStatusWhat anchors it
Sourced — a real published figure stands behind it
Blended hours per monthly activeSourced The strongest cell in the table. Bracketed by Indian per-app mobile hours — MX Player 7.8, Hotstar 4.5 — and by global free services: Pluto TV 5.0–5.6, Tubi 8.6–10 per MAU. Both of the latter are CTV-weighted and so run above a mobile-first Indian service.
Casual and Marginal retentionSourced The only measurement anywhere of retention split by engagement: once-a-month openers retained at 9%, users at 2–6 opens a month at 12%. This table's 6% and 15% straddle that pair in the same order. US CTV, vendor data, and the source never states its retention window while these rates are annual — a real weakness.
Bounded — no figure, but measurement constrains it from both sides
How the blended row is computedArithmetic It is not the share-weighted average of the column above it. Weighting is by the steady-state base: a tier holding s% of intake at annual retention r accumulates to s/(1−r) of the standing audience, so tiers that retain better are over-represented in the base relative to their intake share. Power is 6% of new accounts but 9% of the steady base; Marginal is 45% of intake and 38% of the base. Blended hours are then Σ(base weight × hours). That is why the row reads 6.64 and not the 5.36 an intake-weighted average would give — a 24% difference, and the engine has always used the former.
Concentration — the heaviest fifth of the baseBounded Floor: the heaviest 20% of US television viewers take about 50% of viewing. Ceiling: the heaviest 25% of UK YouTube viewers take 87–90%. This model sits near the midpoint of the only two independent panels that publish the shape at all.
Power tier hoursBounded Cut from 30 to 24 hours a month on this evidence. The average Indian internet user spends about 90 minutes a day online in total; an hour a day on one free app was two-thirds of that. 24 hrs/month is 47 minutes a day — still assertive for a top-6% cohort, and still the number a careful reader should attack first.
Marginal tier hoursBounded Raised from 0.7 to 1.5 hours a month. The lightest 75% of UK YouTube viewers average about 6 minutes a day — four times the old figure. No published light-viewer segment anywhere runs as low as 1.4 min/day. At 3 min/day this is still half the only comparable measurement.
Power and Heavy retentionBounded above only Capped by all-app streaming retention of about 50%, and supported in direction — long-tenure users watch twice the hours of new ones. But nobody measures the top of this ladder, so the level is unanchored and the 6-point gap between Power and Heavy is invented.
Assumption — nothing anywhere anchors these
The four account shares — 6 / 16 / 33 / 45Assumption No source anywhere publishes a four-way split of a streaming audience by viewing intensity. Barb cuts at quartiles, Nielsen at quintiles, Samsung Ads by months active rather than hours. India's nearest equivalent segments by demographic archetype and shows only a 2× spread end to end — a caution against assuming too long a tail. The tier count and every boundary are a modelling choice, not a finding.
Heavy and Casual hoursNot independent Interior points on a curve whose endpoints are bounded and whose blend is anchored. Given the shares and the blend, these are consequences of the other assumptions rather than estimates in their own right — they should not be presented as separate judgements.
3.2

How much each user demands — television

Tiers—
Tier% of new householdsHrs/month = min/dayAnnual retentionSteady base% of hours
Blended—— ——— 100%
Calibration
—
3.3

Buying the users

Budget and allocation—
—
Mobile budget, Y1
—
TV budget, Y1
—
15-yr acquisition spend
Cost per retained user—
Where those two numbers come from — 25 routes, four cost regimesevidence behind M.cac and T.cac
RouteScreen₹ per retained userEvidenceVerdict
Regime 1 — priced per user · a real CAC exists in principle
Paid UA — Google UAC, Meta, programmaticMobile83–1,167CPI ₹25–112 ÷ 6–30% survival · one agency blog + Meta spend dataAbove ceiling at every point
Glance lock-screen install campaignsMobileunknowncorrected — no Glance rate card is published; InMobi quotes on requestUnpriceable
Phone OEM dynamic preload (AVOW-type)MobileunknownAVOW confirms CPI-then-CPA pricing, publishes no figuresUnpriceable
Regime 2 — negotiated · computable only after a quote
Smart-TV OEM preloadTV450 assumedno published benchmark anywhereGet a quote
TV featured row / launcher railTVunknown, below preloadnoneGet a quote
Set-top-box preload — Xstream, Binge+, JioFiberTVunknownnoneGet a quote
Telco / ISP / DTH bundlingBothunknownusually per subscriberGet a quote
Regime 3 — revenue share · CAC ₹0, cost follows revenue
YouTube · syndication · white-label · embedded playerBoth0you pay a % of what the user earnsMarketing only
Regime 4 — free or engineering-only · marginal CAC ≈ ₹0
Store listings · editorial · referral · retail demo · short-form · discovery integrations · Search & Discover · mobile web/PWA · WhatsApp · castingBoth0salary ÷ whatever volume it producesThe base layer
3.3b

Reconciling the tiers to published India figures

Amazon MX Player reports 50 minutes a day per user on mobile and 80 on connected TV — but that is per active day, not per monthly active, so it reconciles with the tiers above only through a DAU/MAU ratio. The survey figures are worse: the 2.9 hours a day of big-screen streaming is self-reported across roughly six services a viewer uses, not one. The tiers here are hours per monthly active on this one service — neither of those things. And no Indian streaming service publishes DAU, so the ratio that joins them is an assumption buried inside the tier numbers. This block drags it into the open.

—
Mobile — blended hrs/month per MAU
—
= min per calendar day
—
TV — blended hrs/month per household
—
= min per calendar day
—
TV ÷ mobile engagement
—
Top 22% of the base, share of hours
If DAU/MAU is…Implied mobile min / active dayImplied TV min / active dayAgainst Amazon MX Player's 50 and 80
What this block is telling you
—
3.3c

What a television household actually costs — the chain, not a number

The three links—
—
Cost per install, ₹
—
→ ₹ per retained household
—
× the mobile figure
—
vs ₹152 TV lifetime value
If the install rate is…Cost per install₹ per retained householdAgainst a ₹152 household
The one number that decides this
—
3.4

Cross-screen — overlap and spillover

How the screens feed each other—
—
Year 15 unique reach
—
Mobile accounts arriving free
—
Mobile effective cost
—
TV effective cost
What spillover is actually worth
—
3.5

Does buying a user pay for itself?

—
Mobile — lifetime hours
—
Mobile — lifetime value
—
Mobile — LTV ÷ cost
—
TV — lifetime hours
—
TV — lifetime value
—
TV — LTV ÷ cost
Mobile
—
Television
—
3.6

The demand that results

Mobile — are you running out of India?
—
Television — are you running out of India?
—
—
Total demand cost, Year 15
Sources
Sources for this tab — re-verified 5 September 2026

Universe — television. WPP Media × The Trade Desk, fieldwork by Ormax, August 2026 — 207M CTV viewers across 62–65M households; 2.9 hrs/day of big-screen streaming across all services, 3.4 hrs among FAST viewers; 2.5 viewers per CTV impression; rural CTV up 110% year on year. The strongest number on this tab. Note the definitional trap: FICCI-EY 2026 reports 68M CTV-owning households but only 40M weekly active — device ownership and actual streaming differ by 1.7×, and this model wants the active figure. A separate ceiling worth knowing: India has roughly 60M broadband-enabled homes, almost the same number, so the two constrain each other.

Universe — mobile. Ormax OTT Audience Report 2025 — 601.2M, defined as anyone who watched at least one online video, free or paid, in the past month. Not subscribers. FICCI-EY's comparable figure is 572M, a 5% gap between a panel measure and an industry model. AVOD is 72% of that universe and took essentially all the growth; SVOD has been flat near 150M. The 2026 edition's fieldwork is complete and its headline is expected within days — this number is about to go stale.

Engagement. Amazon MX Player (Business Standard, Nov 2025) — 80 min/day on CTV, 50 min/day on mobile, both roughly doubled year on year. This is the platform's own claim with no denominator disclosed, almost certainly per active user per active day. It is the only single-service India comparison that exists, and it implies a TV/mobile ratio of 1.6×. The WPP 2.9 hrs figure is not comparable — it sums roughly six services per viewer.

Cost per install. Superads India, drawn from over $3bn of Meta spend, Sept 2025–Sept 2026 — median India CPI ₹112, range ₹30–616, and falling in dollar terms from ₹67 to ₹40 across the year. All-industries, Meta only. The narrower ₹25–70 OTT band is one Indian agency's self-reported figure across its own portfolio, not measurement. iOS runs 30–80% higher on 4–6% of Indian handsets, so an Android-only plan is the honest default.

Retention. AppsFlyer via Business of Apps — Android D1 20.2%, D30 3.8%; the same source notes India has the lowest app retention of any country measured. Airship puts Entertainment D30 activation at a median of 8%. AppsFlyer's uninstall report shows the Indian subcontinent at 64–66% uninstalled within 30 days, the worst region measured. For a paid-subscription upper bound, Antenna's US premium SVOD data gives 36% twelve-month cohort survival and 4.6% monthly churn — and a free Indian AVOD service should not out-retain paid American SVOD.

Ad prices, for context on what a household is worth. The only published Indian rate card found is ZEE5 at ₹113 video CPM. Two agencies disagree threefold on CTV CPM — one estimates ₹150–350 open programmatic, another quotes ₹450–1,200. Independently reported: IPL 2026 at ₹21 lakh per 10 seconds on CTV against ₹18 lakh on linear. CTV advertising was ₹9,900 cr in 2025, up 42% (FICCI-EY).

Removed as unsourced, September 2026 — three figures this tab previously carried as benchmarks: "OTT retention D1 50% / D7 30% / D30 18%" traces to no primary source in Adjust, AppsFlyer, Sensor Tower or Airship and appears to be SEO filler; "regional-language content retains 20–35% better at D7" and "sub-₹10,000 devices retain 20–40% worse" have no published basis either. Both are directionally plausible and both were fabricated precision. Also removed: a claimed Glance published rate card and a JioHotstar cost per install of ₹65–75 — see 3.3. Still not sourced, and it decides the plan: television cost per retained household. No smart-TV preload rate card exists anywhere, in India or elsewhere, and India has no CTV install-attribution standard. The default was raised from ₹450 to ₹900 on structural reasoning rather than evidence: television has no cost-per-install network tier, weak attribution and scarce inventory, so it cannot plausibly cost only 1.5× a mobile user. Treat 3× as a floor for the ratio, not a measurement. Also unmeasured: the four-tier splits and every retention rate — no Indian source publishes viewing concentration at all, because Ormax measures reach only, BARC's panel covers 0.025% of TV homes, and platforms do not disclose. Cross-screen overlap and spillover likewise have no Indian data. And "retention" here means still monthly-active twelve months later; every Indian churn figure you will be quoted is subscription churn, which a free service does not have — there is no cancellation event, only lapse.
04

Content delivery to demand

Evidence key Source published measurement — click through Derived inferred, or a planning choice No source no published figure exists anywhere Hover any marker for what it actually measures.
CDN—
4.1

Encoding ladder — pick a rung for each screen

Ladder—
Top rungCodeckbpsGB/hr Mobile keeps ₹/hrTV keeps ₹/hrLoad into
—
Mobile — delivery ₹/hr
—
Mobile — kept ₹/hr
—
TV — delivery ₹/hr
—
TV — kept ₹/hr
Mobile unit economics
—
Television unit economics
—
4.2

What delivery costs to run

—
Total delivery cost, Year 15
—
of which CDN
Sources
Sources for this tab

Apple HLS Authoring Specification — the ladder is Apple's published H.264 and HEVC bitrate table verbatim. GB/hour is kbps × 3600 ÷ 8 ÷ 106 plus 128 kbps stereo AAC. Apple's own default variant is the 2000 kbit/s rung, and the spec notes HEVC "should use bit rates reduced by about 20%".

Airtel Edge CDN rate card — ₹6.00/GB to 10 TB, ₹5.00 to 50 TB, ₹4.00 to 150 TB, ₹3.00 to 500 TB, then "contact for quote". The card is priced on total volume, which is why a single negotiated rate across both screens is the honest way to model it.

Airtel × Qwilt Open Edge — open caching embedded inside Airtel's network for 400M+ subscribers. The route to the ₹0.20–0.50 telco-cache case.

JioCDN — Jio's internal CDN opened to enterprises on "India-first economics", proven on IPL and the T20 World Cup. Publishes no rate card.

JioHotstar IPL 2026 infrastructure — encodes 4K down to 360p plus 2G-grade renditions and holds >90% cache hit at 65.2M concurrent. Precedent for shipping a low bottom rung in India.

05

Monetisation

Evidence key Source published measurement — click through Derived inferred, or a planning choice No source no published figure exists anywhere Hover any marker for what it actually measures.
5.1

Sales capability — shared

Fill and the rate card—
ChannelSlots of 100 filled TV reference CPM ₹Ad-tech take % Gross ₹/100Net ₹/100
Filled—— ———
Why this is one number and not two
—
5.2

Ad load and price, by screen

What each screen bears and clears—
—
Mobile — slots per hour
—
Mobile — effective net CPM
—
Mobile — net ₹ / hour
—
TV — slots per hour
—
TV — effective net CPM
—
TV — net ₹ / hour
Why mobile CPMs are a fraction of television
—
5.3

What monetisation costs to run

—
Cost to sell, Year 15
—
Year 15 net revenue
—
Cost to sell as % of revenue
Sources
Sources for this tab

JioStar WPL 2026 rate card — CTV ₹350 CPM against mobile ₹180–220, the same property, the same matches and the same advertisers split by device. This is the cleanest available measurement of the mobile-to-television ratio in India: 51–63%, and it is where the mobile CPM level default of 55% comes from.

India CTV CPM benchmarks 2025–26 — open programmatic ₹150–350, audience-targeted PMP and direct ₹350–700, premium platform and content ₹700–1,200, Samsung Ads and OEM inventory ₹200–500, live IPL ₹400–600+. India CTV remains far below the US at $20–40 (₹1,700–3,400).

JioHotstar advertising rates — video CPM ₹170–200 blended across devices, display ₹50–60, and CPI ₹65–75 on its own install campaigns. Targeting layers add 15–25%; IPL and Diwali add 50%+.

JioStar raises IPL 2026 CTV rates 25% — CTV pricing power is rising, not falling.

One CTV impression, 2.5 viewers — co-viewing is the main structural reason a television impression is worth multiples of a phone impression.

India app CPMs versus global — AdMob and Meta in-app rates of $1.05–1.88 (≈₹90–165). Treat with care: these are banner, interstitial and rewarded inventory, not premium in-stream video, and using them to price OTT mobile video understates it badly.

Not sourced: the fill rate, and the assumption that television clears 100% of the market card. A library of unfamiliar regional films with no live sport arguably clears below the card on both screens — try television at 70–80% to see what that costs you.
06

Simulation

Evidence key Source published measurement — click through Derived inferred, or a planning choice No source no published figure exists anywhere Hover any marker for what it actually measures.
6.0

What this run says

The result, in words—

—

The result—
The cheque—
Where it goes—
Where it comes from—
The studios' share—
The hour watched—
The size of the gap—

Every sentence here describes this run and nothing else. Change any number on tabs 02 to 05 and they change with it. The model raises no capital, spends nothing on marketing the brand itself, and assumes no money from anything but advertising.

6.1

Overhead and the taper

Overhead and the taper—
LineYear 1 ₹ crGrowth %/yr Year 15 ₹ cr
What the taper is doing
—
6.2

Profit & loss

₹ crore — cash basis—
Cumulative cash
The trough is what you have to raise
Net revenue by screen
Mobile against television, year by year
Where the money goes
Where the money goes along the chain
Sources
Sources for this tab

FICCI-EY India M&E report — the ₹28,000 cr digital video and ₹9,900 cr connected-TV advertising denominators used for the market-share readout in the header. India's M&E sector crossed ₹2.78 trillion with digital media past ₹1 trillion.

India tops consumption, lags monetisation (EY–FICCI) — 1.23 trillion hours on smartphones with 59% on media, yet India is absent from the top 20 monetisation markets. The structural reason revenue per user stays low in this model.

Not sourced: the overhead figure, which is now a single flat ₹10 cr rather than a headcount build — a planning simplification, not an estimate. It replaced eleven separately-grown lines totalling ₹49.5 cr in year one. The taper exists because a rate set for a five-year plan produces artefacts rather than forecasts when run for fifteen. The peak-funding figure in the header is simply the deepest point of cumulative cash; the model does not assume any capital arrives to cover it.