TTD AE common mistakes
TTD AE · Understand the traps and how to avoid them.
Treating the platform's fee/tech fee as a line item cost and subtracting it twice when estimating net media spend, or forgetting it entirely when reconciling budget to working media.
Candidates memorize the fee percentage but never map where it is applied in the cost chain, so they apply it to the wrong base or double-count it.
What to do instead
Write the cost chain explicitly: gross budget → platform/tech fee → working media → bid → win → impression. Practice converting a gross budget into working media and back, and state the base each percentage applies to before calculating.
Confusing advertiser, campaign, and line item hierarchy - e.g., expecting pacing, budget, and targeting to be set at the same level, or putting a budget where only a pacing cap exists.
The platform uses a three-tier hierarchy with different controls at each level; candidates who learned a two-tier model (campaign/ad group) map fields incorrectly.
What to do instead
Build a one-page hierarchy map listing which settings live at advertiser, campaign, and line item level. For any question, first identify the level being asked about, then recall only that level's controls.
Choosing a bidding strategy by name preference rather than by the objective and data available - e.g., selecting a performance/CPA-style strategy on a new advertiser with no conversion history or sparse pixel data.
Strategy names sound outcome-oriented, and candidates assume the 'smartest' option is always best without checking volume and learning-data prerequisites.
What to do instead
Create a decision table: objective, available conversion volume, required KPI, and bid control. Match each strategy to minimum data thresholds and fall back to a base/impression-based strategy when data is thin.
Treating all inventory as equivalent - quoting open-market CPMs for a deal question, or assuming a private marketplace deal guarantees a fixed price and guaranteed delivery.
Candidates collapse open auction, PMP, preferred deal, and programmatic guaranteed into 'deals' and miss the price/delivery guarantees that distinguish them.
What to do instead
Memorize a four-column matrix: price fixed vs. auction, inventory guaranteed vs. non-guaranteed, who sees the deal ID, and typical use case. Answer deal questions by locating the deal on that matrix first.
Building audience segments without checking match rates, scale, or data-source suitability - e.g., proposing a first-party CRM segment with tiny reach as the primary targeting for a broad awareness campaign.
Candidates focus on segment logic and forget that onboarded segments shrink through match and overlap, so the addressable audience may be far below the plan.
What to do instead
Before recommending a segment, estimate matched reach (segment size × expected match rate, minus overlap), compare it to required impression volume, and pair narrow first-party segments with broader contextual or third-party layers.
Misreading pacing scenarios: assuming a line item that is behind pace should always get a higher bid, or that a budget cap alone controls daily delivery.
Candidates conflate budget, bid, and pacing levers and do not separate 'not spending enough' from 'not winning enough' or 'not eligible enough'.
What to do instead
Diagnose in order: is it eligible (targeting, frequency, deal active), is it bidding competitively (win rate, bid vs. clearing price), is it budget-constrained (daily cap hit time)? Apply the lever that matches the diagnosed cause.
Reporting on the wrong metric layer - using platform-reported impressions/clicks as if they were deduplicated, cross-device conversions, or treating view-through and click-through conversions as interchangeable.
Candidates learn metric definitions in isolation and never reconcile platform numbers against advertiser analytics or attribution windows.
What to do instead
For each report, state the source, the attribution model, the window, and the deduplication rule. Practice explaining why platform conversions and advertiser-side conversions differ before answering measurement questions.
Skipping QA on targeting intersections - publishing a line item with conflicting include/exclude rules, empty segments, or geo/language combinations that yield zero eligible supply.
Candidates answer conceptual questions correctly but have not internalized that targeting rules AND together, so overlapping constraints silently zero out reach.
What to do instead
Adopt a pre-launch checklist: read every include and exclude as a combined boolean, estimate reach after all constraints, and forecast eligible impressions before saving. Apply the same checklist mentally to exam scenarios.
Confusing supply path optimization goals - optimizing for the shortest path rather than the most efficient path, and ignoring fees, resold inventory, or duplicate bidding on the same impression.
Candidates equate 'fewer hops' with 'better' without accounting for take rates, auction dynamics, and log-level data needed to verify the path.
What to do instead
Evaluate each path on total cost (media + intermediary fees), transparency (log-level data, deal ID visibility), and auction efficiency. Practice ranking paths by effective CPM rather than hop count.
Answering troubleshooting questions with a fix before identifying the symptom class - e.g., rebuilding a creative when the real issue is a paused deal, expired flight, or frequency cap.
Under time pressure candidates jump to the most familiar fix and skip the eligibility-versus-performance distinction.
What to do instead
Use a fixed triage sequence: status and flight dates → budget → targeting and deal eligibility → creative approval → bid competitiveness → measurement lag. State the triage step before naming the fix.