Most CTV companies run growth strategy meetings on top of infrastructure that can't actually answer which row placement, which FAST deal, or which ad exposure drove the household to convert. Winston Francois builds the attribution pipelines, event tracking, and experimentation infrastructure that make CTV growth decisions provable instead of guessed.
The Problem
Attribution is fractured across every OEM you distribute on
Roku exposes device and ACR data under one set of terms, Fire TV under another, and Samsung and LG license their ACR feeds separately with their own household graphs and refresh schedules. Nobody has built the pipeline that stitches device IDs, ACR signals, and household-level identity together into one view of a subscriber's journey. The result is a growth team that can tell you aggregate installs by platform but can't tell you which platform's content-discovery surface actually drove the subscription.
App store discovery is a black box, not an instrumented funnel
Getting listed and certified on the Roku Channel Store, Fire TV Appstore, and Samsung/LG smart hubs is treated as a one-time submission event rather than an ongoing feedback loop. There's no tracking on impression-to-click-to-install by row placement or category rank, so ASO decisions – which artwork, which category, which keyword set – are made on gut feel instead of data pulled back from the store's own discovery surfaces.
Experimentation infrastructure wasn't built for set-top box telemetry
Web and mobile experimentation tooling assumes real-time event streams and fast client updates. Living-room app runtimes on Roku sticks, older Fire TV hardware, and smart TV SDKs batch telemetry, cache aggressively, and update on release cycles measured in weeks. Standard A/B testing frameworks silently underperform or produce corrupted reads when pointed at this hardware, so most CTV teams either don't test browse-to-watch changes at all or trust results that are quietly wrong.
Server-side ad insertion severs the link between ad exposure and outcome
SSAI stitches ads into the video stream on the server before it reaches the device, which is good for broadcast-quality playback and bad for attribution – the growth analytics stack often can't see which specific ad pod a household was served, so AVOD and FAST reactivation campaigns can't be tied back to actual viewing behavior. Marketing pays for exposure it can't verify converted anyone.
How We Help
We start by mapping your current data topology: what ACR data you're licensed to receive from each OEM, what device and household identifiers you can legally join across platforms, what your SSAI vendor exposes via logs or callbacks, and where your event tracking simply doesn't exist yet. Most CTV companies are surprised by how much usable signal is sitting unstitched in vendor dashboards nobody has piped into a shared warehouse.
From there we design the attribution pipeline itself – the ETL and identity-resolution layer that takes Roku's device-level events, Fire TV's advertising ID, and Samsung/LG's ACR feeds and reconciles them into a household-level graph you actually own. This is engineering work, not strategy slideware: schema design, join logic that respects each platform's data-use restrictions, and a refresh cadence that matches what each OEM actually allows.
For app store discovery, we build the instrumentation loop back from the storefront: row placement and impression data pulled through each platform's partner APIs where available, correlated against install and first-session events, so your ASO decisions – artwork, category placement, search keyword targeting – are informed by what the store's own discovery surface is showing users, not assumptions carried over from mobile ASO playbooks that don't map to 10-foot UI behavior.
On playback and discovery, we instrument the full browse-to-watch funnel: row impressions, hover/select events, autoplay-driven starts versus deliberate selections, and abandonment points inside the player. This is the layer that tells you whether a content placement change actually moved viewing behavior or just moved where people clicked before bouncing.
For experimentation, we build test infrastructure that accounts for the real constraints of set-top box and smart TV SDKs – batched telemetry, delayed client updates, and limited real-time flags – so you can run valid tests on browse UI, row ordering, and autoplay behavior without the false positives that come from applying web-style A/B tooling to living-room hardware.
We also close the loop on SSAI: working with your ad server and SSAI vendor to get ad-pod-level logs into the same pipeline as your ACR and device data, so AVOD and FAST reactivation spend can be measured against actual household exposure and conversion instead of aggregate impression counts.
What makes this different from a typical agency engagement is that we're fractional operators, not a strategy deck team. We embed with your engineering and analytics leads, write the actual data contracts and instrumentation specs, and stay through implementation – we don't hand off a plan and disappear before it's built.
What we deliver
- Cross-platform attribution pipeline design: ACR, device ID, and household-graph reconciliation across Roku, Fire TV, Samsung, LG, and Vizio
- App store discovery instrumentation: row placement and impression tracking tied to install and activation events
- Full browse-to-watch event taxonomy covering row impressions, autoplay starts, and playback abandonment
- Experimentation framework built for batched-telemetry SDK constraints, not ported from web tooling
- SSAI-to-growth-analytics integration for ad-pod-level exposure and conversion measurement
- Data governance map documenting what each OEM's terms actually allow you to join and retain
If your attribution pipeline can't tell you which OEM's discovery surface drove a subscription, you're not making growth decisions – you're making platform-relations decisions and calling them growth.
Our Methodology
We run growth engineering engagements for CTV companies as 90-day infrastructure sprints. Days 1-30 are a data and access audit: we inventory every ACR license, device-ID feed, SSAI log, and app-store partner API you currently have access to, identify the gaps against what a real household-level attribution model requires, and produce a prioritized build list ranked by which gap is costing you the most in misallocated acquisition or reactivation spend.
Days 31-60 are pipeline design and build: schema, identity-resolution logic, and event taxonomy get specified and implemented alongside your data engineering team, with special attention to the platform-specific data-use terms that make CTV attribution legally different from mobile attribution. We build the experimentation harness in parallel so it's ready to consume the new event streams the moment they're live.
Days 61-90 are validation and handoff: we run the pipeline against a real campaign or content-placement change, confirm the attribution reads match what you can independently verify (upfront commitments, known FAST distribution deals, direct-sold campaigns), and document the system so your team can extend it without us in the room. By day 90 you have infrastructure that answers questions your growth team couldn't answer on day 1.
How We Work
Growth engineering engagements typically run 3-6 months, structured around the 90-day sprint above with an option to extend into ongoing platform additions (a new OEM launch, a new FAST distribution partner, an SSAI vendor migration). We embed with your data engineering, growth, and ad-ops teams rather than working from the outside – attending your existing sprint planning and standups rather than asking you to adopt ours.
Your team owns production infrastructure and implementation; we own the data-pipeline design, the identity-resolution logic, the event taxonomy, and the experimentation framework, and we stay through build and validation rather than handing off a spec. For teams without a dedicated data engineer, we can also help scope what hire or contractor capacity you need to sustain the pipeline after the engagement ends.
Weekly cadence is a working session with your engineering lead and whoever owns growth or marketing analytics, plus async progress against the build list. Every artifact – schema, taxonomy, test design – is documented and owned by your team, not held in our heads.
Expect us to push back on vanity build requests (a real-time dashboard nobody will act on) in favor of the infrastructure that actually changes a decision – which OEM to negotiate harder with, which content row to test next, which FAST deal is or isn't pulling weight.
- Attribution accuracy across fragmented OEM platforms instead of platform-siloed guesswork
- ASO decisions grounded in actual discovery-surface performance data
- Valid experimentation results from infrastructure built for real CTV SDK constraints
- SSAI exposure data usable for actual reactivation and AVOD/FAST campaign measurement
If your ctv / connected tv company needs growth engineering leadership, we should talk.
Expand your marketing team output with our experts
Let us take a custom approach to your growth goals by assembling and leading the best-in-class marketing team to support your next stage.
Frequently asked questions
How much does growth engineering cost for a CTV company?
Engagements typically run $12K-$25K per month depending on how many OEM platforms you're reconciling and how much existing pipeline infrastructure you already have. A company distributing on three platforms with no existing attribution stitching sits at the higher end; a company adding one new SSAI integration to an existing pipeline sits lower.
How long until we see attribution improvements from this work?
The audit alone, in the first 30 days, usually surfaces at least one concrete misattribution that's costing you money right now – that gets flagged immediately, not held for the final report. The full pipeline is typically live and validated by day 90.
Do you replace our data engineering team, or work alongside them?
Alongside. We design the identity-resolution logic, event taxonomy, and pipeline architecture; your engineers implement and own the production systems.
What makes Winston Francois different from a CTV ad-tech vendor or measurement platform?
Vendors sell you a product with their own data-access limits and lock-in. We're fractional operators who build the pipeline as your infrastructure, using whatever combination of vendor feeds and direct OEM access actually gets you a complete picture – including telling you when a vendor's offering won't get you there and what to negotiate for instead.
How do you measure ROI on growth engineering investment?
We track it against decisions the new infrastructure makes possible that weren't possible before: correctly attributed acquisition spend by platform and campaign, ASO changes with a measurable before/after on install rate, experimentation results your team trusts enough to act on, and SSAI exposure data that lets you defend or cut a FAST or AVOD reactivation line item with evidence instead of instinct.
Is our company a good fit for this engagement?
This is built for CTV and streaming companies at Series A/B or growth stage, roughly $5M-$100M ARR, distributing across at least two or three OEM platforms, where the growth or marketing team is making spend decisions without a reliable cross-platform attribution view. If you're pre-launch on a single platform, this is premature – come back once you're live on multiple storefronts and the fragmentation problem is actually costing you money.
