Questions to Ask an Ad Network Before Committing Your Budget

Weak creative is only one way an ad budget can go to waste. A poorly matched network can undermine a campaign before the first placement runs. These questions help get clear answers on supply, fraud, forecasting, and exit terms before money moves.

Where does the inventory come from?

Start with supply. A network is a middleman between advertisers and publishers, and the quality of that work depends on where impressions originate.

Ask if inventory comes from direct publisher deals or resold placements bought elsewhere in programmatic advertising. Both models carry different risks. Direct deals give the network more control over placement quality and brand safety. Resold traffic adds extra hops and blind spots that make bad placements harder to catch. It also makes credits harder to get when something goes wrong. Ask what share is direct versus aggregated from other sellers, and how that mix affects price and visibility.

Ask for a representative list of top domains or apps where ads would show for your vertical and formats. A sample is enough to judge fit; the full publisher list isn't necessary.

Don't accept vague promises.

If the rep can't name sources or show placement-level data, there's little basis to trust later quality claims. A straight answer names sample publishers and explains how you can block unwanted placements. Placement-level transparency is the real test because it shows whether the network knows where ads run or is just passing through volume.

How is invalid traffic detected and filtered?

Fraud is the fastest way to burn a test budget. Invalid traffic includes bot clicks, fake impressions, and other non-human activity that never had a chance to become a lead or sale. In its 2023 report, Juniper Research estimated global ad fraud losses at $84.2 billion for that year and projected $172.3 billion by 2028. That number is reason enough to put fraud controls high on the call agenda.

Ask what antifraud technology runs before billing. Strong networks use in-house detection plus independent third-party verification. In-house tools can flag data-center IPs and odd click patterns, along with placement spikes in real time. Independent verification adds a check that isn't graded by the seller. Ask which verification vendors are supported and whether their reports can trigger credits without a long dispute.

Ask what bot or invalid rate the network will stand behind after filtering, and how it's measured. There's no perfect traffic source. There should be a clear process for detecting and filtering invalid traffic, with refunds when appropriate. Vague quality claims mean little compared with a written policy on measurement, filtering windows, and credit terms. Ask how often blocklists are updated and whether suspect placements are paused automatically or only after an advertiser complains.

Can you forecast volume for your targeting?

Media buying breaks without a forecast. Traffic forecasting tells you whether a GEO and format combination has enough daily impressions to test or scale, or should be avoided. It also tells you if bids need to rise to win inventory.

Ask for an estimator showing expected daily impressions by GEO, device, format, and OS. Those breakdowns can help assess whether the available supply matches the campaign. GEO shows country-level depth. Device and OS show where mobile demand sits. Format shows whether push notifications, pop-unders, interstitials, and in-page display have real depth or only token volume. Without those four splits, planning turns into guesswork and test budgets get spread too thin.

Ask how fresh and granular the data is. Last quarter's country averages won't help a performance marketing buyer who needs daily reads by region, carrier, browser, language, and time-of-day slot. Ask if forecasts reflect recent win rates at your bid range or total available impressions at any price, and whether they adjust for frequency caps and dayparting. A useful answer cites recent windows and shows ranges by targeting layer. An evasive answer offers one big number with no breakdown or date.

Which networks should make the shortlist?

After the first three questions, most networks sort themselves. Some document supply, fraud controls, and forecasts in writing. Others stall or deflect, promising follow-ups that never arrive. That gap is the short list.

Many buyers search for a top ad network for advertisers that can show direct supply sources, fraud filtering, and realistic forecasts in the same pitch. The useful comparison is how well each candidate meets the campaign's requirements. Build a small set of candidates, then ask each the same questions and compare answers side by side. Note who provided placement samples and explained filtering plainly, and who shared a dated forecast with GEO and format splits.

Don't move forward with more than two or three options at this stage. Too many parallel tests dilute data and slow interpretation. Pick networks that answered in specifics and put terms in writing. Drop those relying on slides and slogans. The remaining vetting decides where the first deposit goes and how much control stays with you after launch.

What does pricing and reporting actually include?

Price structure decides whether a test can stay disciplined. Ask how each format is sold across CPM, CPC, CPA, and hybrid deals. Those four cover most cases, but the fit depends on the goal. CPM fits reach and retargeting pools where impression depth matters. CPC ties payment to clicks, but does not establish that those clicks carry search-like intent. CPA and hybrid deals fit affiliates and direct-response buyers who need payout tied to actions. Ask which formats support each model; networks don't price every format alike.

Ask directly about minimum spend, bid floors, and added charges. Ask about the deposit needed to activate, the minimum bid to win usable volume in your GEOs, and any fees for targeting, optimization help, or postback setup. There shouldn't be surprise line items after the insertion order. Ask if unspent funds roll over and whether currency conversion or wire transfers carry fees. Get the fee list before funds move.

Reporting determines whether optimization is possible. Ask if the dashboard updates in near real time and splits results by campaign, placement, GEO, device, and format. Those five cuts help buyers find what works without waiting for a managed report. Ask if attribution and postbacks connect to an outside tracker or analytics platform so conversions tie back to clicks. Ask if postbacks fire on multiple events or only the final sale, since trial starts, app installs, form submits, and purchases tell different stories. Delayed, sampled reporting or a single fixed view slows optimization and lets bad placements eat spend longer.

Who controls optimization after launch?

Ask who touches the controls after launch. Some networks auto-rotate creative, shift budget to top placements, and pace spending through the day. Others leave those tasks fully manual. Either setup can work, but you need to know which you're buying.

Ask if auto-rules can be capped by placement, GEO, and daily budget so a winner doesn't swallow the test. Ask if creative rotation favors click rate, conversion rate, or revenue per mille, and whether pacing spreads spend or front-loads it. Manual control suits buyers with strict brand rules. Assisted control suits small teams that can't watch dashboards all day. Either way, get the defaults in writing and ask how to turn them off.

What can you run and how much control do you have?

Creative fit deserves attention before launch. Ask which formats are live across push notifications, pop-unders, interstitials, and in-page display. Those four behave differently. Push can drive fast volume but needs tight frequency control. Pop-unders can scale but need careful caps and placement blocks. Interstitials demand clean design and fast loading. In-page display needs viewability checks and placement review. Ask which verticals or content categories are restricted on each format and what that means for your angle.

Ask about compliance and moderation before building assets. Ask which claims or images trigger rejection, including problematic landing page elements. Ask how long creative review takes and whether resubmits go to the back of the queue. Ask if landing pages are scanned after approval, since post-launch flags can pause a winner without warning. Clear rules save cycles. Fuzzy rules create rework.

Targeting controls determine how spend can be directed. Ask for GEO, OS, browser, language, carrier, and time-of-day options, plus IP, device type, and connection type where mobile matters. Then ask about placement-level whitelist and blacklist controls. Whitelists limit delivery to proven spots. Blacklists cut losers without killing the campaign. Ask if lists can be built from your data and saved across campaigns, with controls applied before spending starts. If placement controls require managed service or high spend tiers, small tests won't have the tools to improve.

Can you start small and stop fast if results disappoint?

Support quality shows up when things break. Ask what help is included: dedicated manager access, technical onboarding help, written docs, and realistic response times for urgent blocks. Those four cover most needs. A named contact helps with strategy and escalations. Onboarding help matters for pixels, postbacks, and feed setup. Docs matter when the manager is offline. Response time matters most during overspend or tracking drops. Ask how urgent tickets are triaged and which channel gets the fastest reply during campaigns. If support is email-only with no severity levels, plan tests accordingly.

Ask for a limited-budget pilot with no long-term lock. It should have a spend cap and date range, plus a clear success metric tied to your tracker rather than dashboard clicks alone. Ask about budget rollover, refunds of unused balances, instant pausing, and account closure. Those four terms define the exit. A capped pilot with a clear kill switch keeps risk contained while data collects. It also reveals how the network behaves when results are soft.

Ask how to pause in one step if fraud spikes or tracking breaks. There should be a visible pause that stops bidding immediately, not a request sitting in a queue. Ask if rules can auto-pause placements crossing cost or invalid-rate thresholds, and whether change logs show who paused what and when, so tests stay auditable. Fast stops don't fix bad traffic, but they limit damage while the cause is investigated.

Strong networks welcome this scrutiny. They answer with names, dated forecasts, written policies, and tools you can test on a small budget. That paper trail turns a sales call into a workable plan and keeps the next deposit tied to proof rather than promises.