Lux Signal · Early access
Your ad accounts are missing a bidding engine that owns the number they bid on.
Google and Meta already run world-class bidding algorithms. They just bid on whatever number your tracking hands them. Lux Signal is the bidding engine that owns that number, deciding what every conversion is really worth, which ones shouldn’t count at all, and who you should stop bidding on entirely, then feeds it back through server-side conversion APIs so the platforms bid on your model instead of their guess.
Built by Lux Marketing: managing $100K+/mo in paid media, with 50+ pixel & API implementations.
Every bidding algorithm is only as good as the number it bids on.
Google and Meta bid brilliantly. They just bid on a number nobody has looked at since the day the tracking was installed.
The number is wrong
A form fill, a free trial, a $400 order and a $120 order all report as conversions worth whatever your tag says. The bidding algorithm can’t tell them apart, so it bids as if they’re the same.
The number is incomplete
The platform sees the checkout. It never sees the refund, the churn, the lead that never closed, or the customer who came back four more times. It bids on the fraction of value it can observe.
Nobody owns the number
It was set once by whoever installed the tracking and never revisited, and each platform bids on its own siloed version of it. The highest-leverage input in your account is running unowned, on autopilot.
Each conversion gets a real value, so Smart Bidding runs on your model.
It doesn’t replace Smart Bidding or touch your budgets. It owns the signal the bidding algorithms run on.
- 1
Ingest your data
We pull conversions, margins, refunds, pipeline and LTV from your warehouse and CRM: the clean data your Foundation already collects.
- 2
Score and value every conversion
Proprietary models assign each conversion a real worth (predicted profit, lifetime value, lead quality), combining your own value rules with enriched third-party data the bidding algorithms never see.
- 3
Feed the bidding algorithms
Those corrected values flow back to Google and Meta through server-side conversion APIs, the same rails your Clean Data Foundation runs on. No platform API approvals, no risk to your account.
- 4
Bidding runs on your model
Smart Bidding keeps doing the bidding, at the platform’s full speed and scale, but now it’s bidding on your definition of value, across both channels.
- 01Warehouse + CRMMargins, LTV, pipeline
- 02Lux SignalValue + quality per conversion
- 03Server-side eventsCAPI + Enhanced Conv.
- 04Google + MetaBid on your model
It does four jobs no ad platform can, because the data lives in your business.
Four jobs no ad platform will ever do for you, because the data they need lives inside your business.
What is this conversion worth?
Predicted profit, lifetime value, pipeline value: whichever definition fits your business. The bidding then chases the customers worth having instead of the ones that look biggest at checkout.
Should it count at all?
Every lead and conversion is scored before it ever reaches the platform, so junk conversions stop training the bidding algorithm to go find more of them.
Who should we stop bidding on?
Existing customers, low-fit segments, accounts that will never close. They’re sent through at near-zero value, so the platforms’ own bidding stops chasing them.
What happens when something breaks?
Anomaly detection, tracking-break alerts, value caps and guardrails, built to catch a broken tag or a freak order before it distorts a week of bidding.
“Valuable” means something different in every business, so the engine bids on yours.
The engine doesn’t impose one definition of value. It bids on yours.
- Bid on contribution margin, not order value
- Repeat-purchase and cohort LTV, not first order
- Discount conversions by refund and return probability
- Bid harder on the first orders that become second orders
- Lead quality score sent as the conversion value
- Closed-won revenue imported back from the CRM
- Suppress tire-kickers, existing clients and bad-fit geos
- Bid on booked revenue, not form fills
- Churn-discounted predicted LTV, not signup count
- Trial-to-paid quality scoring at the point of conversion
- Firmographic fit: company size, industry, role
- Bid on accounts that retain, not trials that register
It works with the platforms’ algorithms instead of fighting them for control.
Not another bid-management tool fighting the platforms for control. A bidding layer that makes their algorithms work your objective.
Your bidding model, not theirs
You define what a conversion is worth, and the platforms bid to that definition. The single input that decides where every dollar goes finally belongs to you.
Cross-platform
One bidding model across Google and Meta, measured on blended after-margin return: the view no single platform will ever give you.
Guardrails & control
Value caps, rules, and anomaly protection. You tune the model and approve the logic; no bidding decision runs as a black box.
No account risk
It never sets bids or moves budgets directly. It only corrects the values the bidding algorithms consume, so the platform’s own safeguards always stay in force.
Proprietary models + data moat
Proprietary models combine your first-party data with enriched third-party signals the platforms can’t see. The richer the inputs, the bigger the bidding edge competitors can’t replicate.
Works with Smart Bidding
No rip-and-replace. Keep tROAS, Max Conversion Value, or Performance Max; the engine just gives your existing bidding strategy a smarter target.
Machine-learning models and a rules layer run on your data, enriched with outside signals.
A hybrid of machine-learning models and a rules-and-guardrails layer, fed by your data and enriched with third-party signals. Here’s what’s under the hood of the bidding engine.
- Identity & match data (higher match rates, householding)
- Firmographic & B2B (company size, industry, revenue, role)
- Intent & behavioral signals (in-market audiences)
- Demographic & financial (geo, income, household value)
- Predicted lifetime value (pLTV), not just first purchase
- Lead & conversion quality scoring
- Refund & churn probability discounting
- Blended cross-platform attribution
- Anomaly detection & tracking-break alerts
- Audience suppression (exclude existing or low-fit buyers)
- Offline & CRM conversion import (closed revenue)
- Value caps, rules & guardrails you control
The bidding engine is only as good as the data underneath it.
Bidding on profit, LTV or lead quality means the engine needs server-side tracking, conversion APIs, and that data wired into one place. That’s exactly what our Clean Data Foundation builds. If your tracking and data are already solid, we plug straight in. If they’re not, the Foundation is step one, and the base your bidding runs on.
The engine comes from years of tracking builds and paid media run on them.
Early partners get a model built for their business and founder pricing locked in.
We’re opening early access to advertisers spending on Google and Meta. Founding partners get a bidding model configured to their business, founder pricing locked in, and direct input into the roadmap.
A good fit looks like
- Spending $15K+/mo across Google and/or Meta
- Some conversions are worth materially more than others, and you know which
- Running value-based bidding (tROAS, Max Conversion Value, or PMax)
- Server-side tracking in place, or willing to build the Foundation first
Pricing is a percentage of managed ad spend, aligned to profitable growth. Founder rates locked for early partners. Full pricing published after the early-access cohort.
It never touches your bids or budgets; it only corrects the values they run on.
How the bidding engine actually works, in plain English.
Does it take control of my ad accounts or budgets?
No. Lux Signal never sets bids or moves budgets. It only corrects the conversion values that flow into the platforms, and their own bidding algorithms do the rest. That means the platform’s safeguards always stay in force and there’s no risk of an external tool overspending your account.
Do I need the Clean Data Foundation first?
You need its building blocks: server-side conversion tracking and your value data (margin, LTV or CRM outcomes) accessible in one place. If you already have reliable server-side tracking and can supply that data, we connect directly. If not, the Clean Data Foundation puts it in place, and it’s what makes the bidding edge defensible in the first place.
Which platforms does it support?
Google Ads and Meta at launch, through Enhanced Conversions and the Conversions API. Because the engine works by sending conversion values rather than managing bids directly, any platform whose bidding accepts server-side conversion values can be added over time.
I run lead gen or B2B. Does this work without ecommerce margins?
Yes; that’s one of the strongest use cases. Instead of margin, the engine bids on lead quality and closed-won value: it scores each lead, imports the outcome back from your CRM, and sends that as the conversion value. The platform stops bidding for form fills and starts bidding for the leads that turn into revenue.
What if I don’t know my margins?
You don’t need perfect cost accounting to start. The engine works with whatever value signal is most reliable in your business: product-level margin if you have it, category-level estimates, average order profitability, or a quality score. We start with the best available input and sharpen it as the data improves. Making that data reliable is exactly what the Clean Data Foundation is for.
What does early access involve?
We onboard design partners by hand. You get the bidding model configured to your business, founder pricing locked in, and a direct line into the roadmap. In exchange we ask for feedback and, where results allow, permission to share anonymized performance. Apply and we’ll see if it’s a fit.
Own the number your bidding runs on.
Apply for early access to Lux Signal. We’ll review your setup, tell you whether the bidding engine will move the needle for you, and what it would take to go live.
Or call us: +1 855 589 6150