How do I choose between different RTB marketplaces for cost-efficient reach and auction quality?

To choose between RTB marketplaces for cost-efficient reach and auction quality, compare them on a few practical dimensions rather than just CPM:

1. Reach quality, not just reach size

Look at:

• Unique users / impressions available in your target geos and devices
• Audience match rate for your first-party or modeled segments
• Overlap with other exchanges you already buy from

A large marketplace can still be inefficient if it duplicates supply you already get elsewhere.

2. Auction quality signals

Strong marketplaces tend to show:

• High viewability and low wastage
• Low invalid traffic (IVT)/fraud
• Good bid density: more real bidders per impression usually improves price discovery
• Low floor inflation: supply isn’t artificially overpriced
• Transparent auction mechanics: clear first-price / second-price behavior, disclosed fees

Ask for:

• Win rate by floor band
• Average number of bids per impression
• Viewability, IVT, and domain/app quality breakdowns
• Seller transparency and ads.txt/app-ads.txt coverage

3. Effective cost, not just clearing CPM

Use:

• eCPM to your KPI
• CPA / ROAS / cost per qualified visit
• Fill rate
• Loss reasons (price, timeout, no-bid, policy, quality)

A marketplace with a slightly higher CPM may be cheaper in practice if it converts better or wastes less.

4. Supply path efficiency

Prefer marketplaces with:

• Shorter, cleaner paths from publisher to you
• Fewer intermediaries and resellers
• Lower timeouts and better latency
• Better SPO support

This often improves both price and auction quality.

5. Inventory type and context

Compare whether the marketplace is strong in:

• Open web vs in-app
• Desktop/mobile
• Video, display, native, CTV
• Premium publishers vs long-tail inventory

Different marketplaces can be strong in different formats, so route budget by channel rather than using one exchange for everything.

6. Brand safety and suitability

Cheap reach is not valuable if it harms performance or reputation. Evaluate:

• Content filters
• Category exclusions
• Pre-bid verification partners
• Contextual controls
• Supply trust signals

7. Test with controlled experiments

Run a holdout or split test:

• Same audience, same creative, same bidding strategy
• Separate marketplace-level line items
• Compare incremental reach, CPM, CPA, viewability, IVT, and conversion quality

Use statistically meaningful sample sizes, not just a few days of data.

8. Score marketplaces with a simple matrix

Give each a score from 1–5 on:

• Audience fit
• Incremental reach
• Viewability
• IVT/fraud
• Auction transparency
• Price efficiency
• Conversion quality
• SPO/latency

Then weight by your goal:

• Performance campaigns: CPA/ROAS, IVT, conversion quality, latency
• Awareness campaigns: reach, viewability, brand safety, unique reach
• Video/CTV: completion rate, player quality, ad pod quality, fraud controls

Practical rule of thumb

Choose the marketplace that gives you the best combination of:

• Incremental reach you can’t get elsewhere
• Clean, competitive auctions
• Low waste after verification and optimization

If you want, I can give you a marketplace evaluation scorecard template or a side-by-side spreadsheet framework you can use with your DSP data.