Adverge Sales FAQ
Straight answers to the questions BD hears.
A canonical reference for the technical, commercial, and architectural questions prospects raise. Each entry leads with the short answer, then the reasoning, then a paired Don’t say / Do say to land the right framing on a live call.
Should we trim our ads.txt entries to cut down on reseller hops?
No. Adverge does not recommend trimming ads.txt as a yield strategy, and doing so would put a revenue guarantee at risk.
Cutting ads.txt entries removes authorized demand paths. On a publisher of any meaningful scale, that means dropping fill and yield, not improving it. The conventional advice around aggressive ads.txt pruning comes from publishers without sophisticated wrapper-level optimization, where the ads.txt file itself is the only lever they have to control supply path quality.
Adverge’s approach is different. Yield Cortex (our wrapper) performs traffic shaping at request time. The ads.txt stays broad, preserving every authorized demand path. The wrapper scores each bidder per request on win rate, latency, and revenue history, then fires only on bidders likely to perform on that specific impression. Non-performant paths are excluded at the wrapper layer, not at the file.
This achieves what aggressive ads.txt pruning is trying to achieve (cleaner supply paths, better SPO scores in buyer-side algorithms) without sacrificing demand availability. Buyers see clean supply chains via Sellers.json and the SupplyChain object. The auction sees fewer redundant bids. The publisher keeps every authorized demand option in reserve.
Specifically: Smartadserver (now Equativ), TripleLift, Sharethrough, and similar SSPs with large reseller networks are not trimmed in our setup. We let the wrapper decide per request which of those paths to call.
“We’ll trim your ads.txt to 2 or 3 entries per SSP for a 4-8% RPM lift from SPO scoring.”
“Our Yield Cortex wrapper handles supply path optimization in real time. Your ads.txt stays broad. The wrapper decides who bids on each impression based on performance scoring, so you get the clean-supply-chain signal buyers’ SPO algorithms reward, with the demand floor preserved.”
Will we have to rip out our existing stack to use Yield Cortex?
No. Adverge’s standard onboarding is an A/B test alongside your current setup, not a migration.
Yield Cortex runs as an alternate path through a wrapper layer (client-side, server-side, or hybrid) that sits next to whatever you have today. The standard setup is a session-level A/B split: control traffic continues to route through your current stack, treatment traffic routes through Yield Cortex. Both groups see the same ad placements. Both are tracked against the same discrepancy baseline so the comparison is clean from day one.
Most publishers run a 50/50 split for 30 to 60 days, then increase Adverge’s share once the data supports it. Some go faster (Sofascore did a full migration after a short comparison window). Some go slower (Boredpanda ran a 90/5 A/B for 74 days, then expanded). The pace is yours to set based on what the lift looks like.
If the lift doesn’t show in the data, there’s no obligation to continue.
“We’ll replace your existing setup with Yield Cortex.”
“We run Yield Cortex alongside your current stack for the first 30 to 60 days so the lift is measurable against your real baseline. Most clients only move more traffic to us after they’ve seen the numbers.”
How do you handle our existing demand partner relationships?
We don’t disrupt them. Your demand partners stay yours. We add to the auction, not subtract.
Adverge’s value isn’t replacing your demand stack. It’s optimizing how the demand competes. Specifically:
- Direct deals stay direct. Buyers who transact with you directly continue to. We don’t intermediate.
- Existing SSPs stay live. Your ads.txt stays broad (see Q1). We don’t trim authorized lines.
- We add complementary paths where there are gaps in your coverage (new SSPs, identity solutions, signal vendors).
- The wrapper decides which partners compete on each impression based on performance scoring, not based on who Adverge favors.
The publisher keeps the relationships, the data, and the reporting visibility. Adverge runs the optimization layer underneath.
“We’ll consolidate your demand stack.”
“Your existing demand relationships stay yours. We add complementary paths where there are gaps, and the wrapper makes intelligent decisions about which partners compete on each impression.”
Do you offer a revenue guarantee, and how does it work?
Yes, on qualified engagements. The structure is a floor against your established baseline, and terms vary by publisher size, vertical, and stack complexity.
The general shape:
- We establish a baseline using the A/B comparison window in your first 30 to 60 days on Yield Cortex.
- We commit to a guaranteed minimum lift over that baseline for the contract period.
- If actual lift falls below the guaranteed floor, Adverge makes up the gap per the engagement terms.
- Revenue guarantees are available above a minimum inventory threshold and are evaluated case by case.
We don’t quote a guarantee in cold outreach. The actual number is sized to your inventory, your category, and the optimization headroom we see in the comparison window.
“We guarantee X percent lift.”
“We do offer revenue guarantees on qualified engagements. The structure is built around your real baseline once we’ve measured it, not a number we quote on the first call. Happy to walk through how we’d think about it for your specific inventory.”
What’s the difference between Yield Cortex and Data Cortex?
Yield Cortex is the publisher’s monetization infrastructure (auction, wrapper, demand orchestration). Data Cortex is the data and identity layer that feeds it. They work together, but either can ship standalone.
- Yield Cortex is the auction infrastructure: wrapper logic, bidder weighting, refresh strategy, floor pricing, supply path control. It’s what publishers run as their monetization stack.
- Data Cortex is the data and identity layer: 10-plus live identity solutions (LiveRamp, ID5, SharedID, Lotame, UID2, 33across, and more), first-party data activation (IP, device, geo, ortb2 fields), and contextual signal on the roadmap. All of these flow into the bid request payload.
The two reinforce each other. Data Cortex feeds richer signals into the bid request, which means buyers can identify, target, and measure better, which means higher CPMs in the auction Yield Cortex runs.
Both are publisher-owned infrastructure, not services. Both ship as code that lives on your stack.
“Yield Cortex is for revenue. Data Cortex is for data.”
“Yield Cortex is the auction infrastructure. Data Cortex is the data and identity layer that feeds it. They’re built to work together, and most publishers run both. Either can ship standalone if that’s the right fit.”
How is Adverge different from Freestar, Mediavine, or Raptive?
Those are managed yield services. Adverge is publisher-owned infrastructure. Different commercial model, different control model, different operational tolerance.
The clearest way to draw the line:
- Managed yield services (Freestar, Mediavine, Raptive, and similar): you outsource the auction to a third party. They control your stack, often your demand relationships, sometimes your supply path. You see the result, not the math.
- Adverge: you run Yield Cortex and Data Cortex on your side. We provide the engineering and the optimization, but you keep the data, the relationships, and the operational visibility.
What that means in practice:
- You own the demand stack. We don’t intermediate buyer relationships.
- You see the auction math. We publish the optimization logic and log decisions at the impression level (see Q7).
- You can swap any component. Yield Cortex is infrastructure, not a managed bundle.
- Direct sales stays direct. Your direct deals continue unchanged.
This is a fit question, not a quality question. If you’d rather hand the stack to a service provider and just see results, Adverge isn’t the right fit. If you want infrastructure you own and can audit, we are.
“Managed yield is bad.”
“We’re built for publishers who want infrastructure they own and control. If you’d rather hand the stack to a managed yield service and just see results, we’re not the right fit. If you want the math visible and the relationships still yours, we are.”
How does the AI decide which bidders to fire? Can we see the math?
Yes. Every bidder selection decision is logged at the impression level, with the scoring inputs visible.
The wrapper scores each bidder per request on a weighted set of inputs:
- Recent win rate on similar inventory
- Median CPM on win
- Bid latency (penalizes slow bidders)
- Discrepancy track record
- Geo and ad-unit relevance
These weights are documented and visible in the operations dashboard. For every impression, the wrapper selects a subset of bidders based on the score, fires the bid requests, and logs which bidders were called, which weren’t, and why.
Publishers can adjust the weights for their inventory if they want the optimization to favor a different target (highest CPM versus highest fill rate versus lowest latency, for example).
This is the answer to Q1‘s ads.txt question too: the wrapper does at request time what aggressive ads.txt pruning tries to do at the file level, with full audit visibility.
“The AI just figures it out.”
“Every bidder selection decision is logged at the impression level with the scoring inputs visible. You can see exactly why each bidder was or wasn’t called on any given impression. Nothing is a black box.”
What happens to our first-party data? Do you sell or share it?
No. We don’t sell or share publisher data. Data Cortex is publisher-owned infrastructure that surfaces your data to buyers through the bid request. We don’t operate as a data broker.
Specifically:
- Your first-party data stays your first-party data.
- Data Cortex passes the signals you authorize (geo, IP, ortb2 fields, contextual signal where you’ve activated it) into the bid request payload.
- We don’t aggregate publisher data across our network.
- We don’t sell publisher data to third parties.
- We don’t use one publisher’s data to inform another publisher’s optimization.
Buyers see your data in real time during the auction. That’s how the lift happens. The data flows into the bid request, not into an Adverge database.
“Trust us with your data.”
“Your data goes into the bid request, where buyers see it during the auction. It doesn’t go into our database, and it doesn’t get sold or shared with anyone else. Data Cortex is publisher-owned infrastructure, not a data broker.”
What’s required to participate in the AdCP beta?
First-party audience data, an internal champion who can move at pilot speed, and a willingness to scope and learn alongside the other beta participants. Vertical, geography, and stack are flexible. Pace isn’t.
We’re inviting six publishers to the Q3 cohort. What we’re looking for:
- First-party audience data (the signals we surface to buyer agents)
- An internal champion who can move at pilot speed
- A willingness to participate in weekly cohort working sessions and share findings
What Adverge provides:
- Seller’s agent integration on top of your existing Yield Cortex setup, no rip-and-replace
- Direct introductions to active buyer agents and the budgets they’re pointing at agentic inventory
- Inventory packaging and signal-schema design with our team, drawing on the Data Cortex publisher data layer
- Session-level reporting on agentic versus baseline, so the lift is measurable from day one
- A seat in the weekly working session where the cohort shapes the next sprint
If you’re not on Yield Cortex yet, that’s fine. The beta can fold the Yield Cortex setup into the kickoff.
Apply: calendly.com/chriscyriax-adverge/30min
Context: adverge.ai/agentic-adcp-stopped-being-theoretical
“Apply and we’ll tell you if you qualify.”
“We’re prioritizing six publishers for Q3. The criteria are first-party data, an internal champion who can move at pilot speed, and a willingness to scope with the cohort. If that’s you, let’s grab 30 minutes.”
How long until we see lift, and what’s the rollout sequence?
Most publishers see measurable lift within 30 days. Full optimization takes about 8 weeks. Case studies show 30 to 60 percent revenue lift in the first quarter.
The standard sequence:
- Weeks 0-2: Clean migration. Yield Cortex live alongside your existing stack with key-value reporting active. Baseline discrepancy tracking starts. Day one of new setup also becomes day one of clean comparison data.
- Weeks 2-6: Brand safety + demand stack tuning. New SSPs onboarded with day-one discrepancy tracking. Identity activation (LiveIntent and similar) where relevant. Geo restrictions reviewed and progressively reopened.
- Weeks 6-8: Price floor strategy + refresh logic + rich media tuning. AI-powered dynamic floors replace static ones. Smart Refresh logic calibrated by geo and ad unit. Rich media formats activated on premium placements.
- 8+ weeks: Compounding optimization. Each new lever adds incremental lift. The first three months show the biggest gains. Year-over-year compounding continues.
Real reference points from the case studies:
- Major sports publisher: +62 percent US revenue and +48 percent US RPS in the first three months, ~8 weeks to full optimization.
- Chatib: +13 percent average revenue uplift in 33 days, from a 50/50 A/B test.
- Boredpanda: +40 to 90 percent eCPM versus incumbent across 74 days of A/B testing.
- Flying Carpet rich media: +69 percent RPM and +38 percent viewability on enabled units.
“You’ll see results immediately.”
“Measurable lift in the first 30 days. Full optimization in about 8 weeks. Our case studies show 30 to 60 percent revenue lift in the first quarter, with compounding gains after that as each new lever activates.”
What happens if we want to leave? Data portability, contract terms.
You can leave. We don’t lock publishers in beyond standard minimum terms, and your data stays yours.
The general posture:
- Standard contract terms include an initial commitment period, then month-to-month after that.
- Data portability: your data exports cleanly. Yield Cortex and Data Cortex don’t hold proprietary data structures that lock you in.
- Demand relationships stay yours. If you leave, your direct SSP deals continue.
- Offboarding is the reverse of installation. No long sunset process. We hand you back the keys.
We don’t believe in lock-in. Publishers who stay should stay because the lift is real, not because the contract makes leaving painful.
“It’s a long-term commitment.”
“We have a standard initial commitment period, then month-to-month. Your data and your relationships stay yours. We don’t build the contract to make leaving painful. Publishers stay because the lift is real.”