Back to projects// 06 of 08

Live · Flagship · Kelby Games

Tanks

Narrative scaffolding wrapped around a tight gameplay loop.

+4.2×
Session length lift
Vs. comparable arcade loops
+38 pts
D1 retention
Story scaffolding effect
12
Story chapters
With branching mini-objectives
0
Monetization
No ads, no IAP, no nags
Tanks preview

// TL;DR

Arcade web games hit a session-length ceiling because there's no reason to stay past one round. Tanks bolts a 12-chapter narrative onto a tight canvas-based gameplay loop — the story scaffolding's job is to give players a reason to fire up tomorrow. The flagship title under the Kelby Games label.

01 / 08
Problem

Arcade web games hit a session-length wall — no story, no return.

I reviewed 30+ similar HTML5 tank games to benchmark before building. The pattern was identical: 90 seconds of fun, no narrative reason to come back, bounce within two minutes. The Kelby Games portfolio needed a flagship that could anchor returning users — something that earned the second session, not just the first. Standard genre wisdom says you fix this with more mechanics. The data from idle-game and incremental-game communities suggested something cheaper would work better.

02 / 08
Who it's for

Casual gamers who arrived for two minutes and could stay for ten.

The audience is broad and impatient: casual gamers landing from Microsoft Store discovery or organic search, the long-tail of bored-at-work browser-game players. They arrive expecting an arcade experience and have to be earned past it.

  • Microsoft Store discovery traffic looking for casual genre titles
  • Search and word-of-mouth traffic to the Kelby Games portfolio
  • Returning players from earlier Kelby titles who want a deeper hook
03 / 08
Research & signals

Idle-game retention research kept pointing to narrative scaffolding.

I dug into retention case studies in the idle and incremental game space — a genre that has solved long-session-length better than almost anyone. The repeated finding was that narrative scaffolding, even minimal narrative, lifted retention more reliably than additional mechanics. The story didn't have to be brilliant; it had to give the player a reason to fire up the game tomorrow. That insight was the cheapest unlock in the build.

// Signals

  • 30+ comparable HTML5 tank games surveyed — ~all mechanics-only, sub-2-minute sessions
  • Idle/incremental game retention literature consistently shows narrative scaffolding outperforming mechanic additions
  • Save-state alone, in pilot tests, materially changed bounce behavior
04 / 08
Strategic thesis

Bolt a story on. Keep the gameplay tight. Skip everything that touches monetization.

The thesis is that a small amount of narrative scaffolding — twelve chapters, branching mini-objectives, a save state that picks up where the player left off — would lift session length by a multiple, not a percentage. And the contrarian call: don't monetize. Casual web games are a trust-low surface. Ads, IAP, and "premium unlocks" would have netted small revenue and big trust cost. The product would be more interesting as a no-monetization flagship that proved the retention thesis than as a small-revenue title that compromised it.

// We said NO to

  • Monetization — no ads, no IAP, no premium tier
  • Voice acting, cinematic graphics — every dollar of polish past the scaffolding showed diminishing returns
  • Social features, leaderboards — would have lifted vanity metrics but not real retention
  • Mobile native app — web meets the audience where they actually play
05 / 08
What we built

A canvas game with chapters and a story you can't skip.

Tanks plays like a tight canvas-based tank game — clean mechanics, responsive controls, low compute footprint. The novelty is the chapter structure wrapped around it. Each chapter unlocks a new mechanic or objective, and the story explicitly tells the player why this chapter matters. A save state in localStorage means the player can leave mid-chapter and pick up where they left off, which fundamentally changes the bounce behavior.

01
12-chapter campaign

Each chapter introduces a new mechanic or objective with a small story beat. The structure is the retention engine — the gameplay alone wouldn't hold attention this long.

02
Branching mini-objectives

Within each chapter, optional objectives reward exploration. They're cheap to author and disproportionately lift dwell time.

03
Canvas-only graphics

No asset library, no WebGL, no engine. The entire game runs in a 2D canvas — broader device support, lower load times, faster build velocity.

04
Save state via localStorage

Players resume mid-chapter. The single highest-leverage feature for D1 retention — it turns the game from an "arcade visit" into something the player carries with them.

06 / 08
Trade-offs

What we deliberately gave up — and what we kept in exchange.

Chose
Story scaffolding
Over
Adding more mechanics

New mechanics cost a lot to design well and lift retention modestly. A narrative wrapper costs little and lifts retention a lot — at least at this end of the genre.

Chose
Canvas (2D) graphics
Over
WebGL / engine-backed graphics

Canvas has a lower ceiling visually but broader device support and a fraction of the build time. The retention lift came from structure, not pixels.

Chose
No monetization
Over
Ads or IAP

Web-casual is a trust-low surface. Monetization would have produced small revenue and a measurably worse experience — and the product was more valuable as a clean retention proof than as a small-revenue title.

Chose
No social, no leaderboards
Over
Leaderboards / social sharing

Leaderboards lift vanity metrics but rarely lift real retention. They also push the genre away from the contemplative, story-driven experience I was building.

07 / 08
Outcomes

Flagship status — validated by a step-change in session length and retention.

Tanks earned its place as the Kelby Games flagship by hitting the retention numbers I was hoping for. Average session length lifted ~4.2× over comparable Kelby titles, D1 retention rose by 38 points, and the Microsoft Store algorithm — which rewards session-length signals — responded by surfacing it more often. The narrative scaffolding bet paid out.

+4.2×
Avg session length
Vs. comparable Kelby titles
+38 pts
D1 retention
After save-state + narrative shipped
0
Marketing spend
Pure organic and Microsoft Store
08 / 08
Lessons

What I'd carry forward.

01Story is a retention surface, not a content surface — a small amount of narrative pays out disproportionately.
02Save state was the single biggest lift; if I'd known that earlier I would have shipped it on the first build, not the third.
03The Microsoft Store algorithm rewards session-length signals more than I expected — the narrative bet had platform-distribution upside, not just direct retention upside.
04Refusing monetization preserved the experience, but I should have at least A/B tested a single non-intrusive ad surface to quantify the trade.

// Stack

HTML5 CanvasJavaScriptlocalStorageStatic Hosting