Dotdigital Certification common mistakes
Dotdigital Certification · Understand the traps and how to avoid them.
Treating 'marketing automation' as a single on/off feature and assuming a program fires the moment a contact is created, instead of tracing the enrollment trigger, the entry conditions that filter contacts out, and the re-enrollment rule separately.
Candidates study program-building screens as a checklist of steps to click, so they memorize the order of nodes but never learn that trigger type (e.g. a date-based vs. an event-based trigger) and entry filters decide whether a contact ever enters the program at all.
What to do instead
For any automation scenario, write down three things before answering: what starts the program, what conditions must be true for a contact to be admitted, and whether a contact can re-enter and how often. Practice by drawing the flow as trigger → filter → delay → decision → action and checking which node the question is really testing.
Confusing suppression and exclusion logic with segment logic, and assuming that a contact who matches a segment will always receive the send.
Segments describe who a contact is; suppressions and exclusions describe who is removed at send time. Candidates learn segment rules well because they are visible in the UI, then answer as if segment membership alone guarantees delivery, forgetting address-book suppression, global suppressions, and per-campaign exclusions.
What to do instead
When a question asks who receives a message, compute the audience in two passes: first the included segment, then subtract every suppression layer (unsubscribes, hard bounces, global suppressions, campaign-level exclusions). Build a mental list of suppression types and check each one against the scenario.
Answering personalization questions by naming the feature (dynamic content, a block, a merge field) rather than identifying the data source, the fallback value, and what the recipient sees when data is missing.
Feature names are memorable and appear in menus, so candidates anchor on the label. Exam scenarios usually test the edge case - a contact with no first name, an empty product recommendation, or a mismatched locale - and the correct answer depends on fallback and default handling, not on the feature name.
What to do instead
For every personalization question, ask: where does the value come from, what happens if it is blank, and is the content resolved at send time or at open time. Learn the default/fallback setting for each personalization method you use and be able to state it without notes.
Treating SMS, MMS, WhatsApp, and mobile app push as interchangeable 'mobile channels' and applying email rules (like open tracking or a subject line) to them.
Candidates group channels by device rather than by protocol and consent model, so they miss channel-specific mechanics: SMS character segmentation and encoding, MMS media and carrier limits, WhatsApp template approval and 24-hour session windows, and push permission and payload constraints.
What to do instead
Build a one-page comparison table per channel covering consent capture, message anatomy, length or payload limits, tracking metrics available, and approval or registration requirements. Revise the table until you can state each channel's constraints from memory, then re-read any scenario and ask which channel's rules actually apply.
Assuming loyalty program points, tiers, and rewards update instantly and in the order events occurred, without considering processing latency, recalculation, and how tier changes trigger downstream automation.
Loyalty feels like simple arithmetic, so candidates compute points correctly but ignore the operational layer - when balances recalculate, how refunds and reversals are handled, and whether a tier upgrade fires an automation immediately or on the next evaluation cycle.
What to do instead
Separate the business rule (how points are earned, expired, or reversed) from the processing rule (when the balance and tier are recalculated). For each scenario, state both, and note which events should trigger a retention message versus which should wait for the next recalculation.
Confusing engagement metrics with business outcomes - reporting open rate as if it proved revenue impact, or comparing click rates across channels that define 'click' differently.
Dashboards surface familiar metrics first, and candidates answer with the metric they recognize rather than the one that answers the question. Open tracking is also affected by privacy features and image blocking, and each channel measures engagement with different definitions.
What to do instead
Before choosing a metric, restate the question as a decision ('are we losing subscribers?', 'which channel drives purchases?') and pick the metric that answers it. Learn which metrics are unreliable in each channel (e.g. opens affected by image proxying) and prefer revenue, conversion, and list-health metrics for outcome questions.
Describing AI features as magic that 'optimizes automatically' instead of explaining what data they learn from, what they predict, and what human configuration or guardrails they require.
AI marketing features are marketed in outcome language, so candidates repeat the marketing claim. Exam questions test the inputs and limits: which historical data trains a send-time or subject-line model, what happens with a new list or sparse data, and where a human must set goals or approve content.
What to do instead
For each AI capability you study, write three lines: the input data it needs, the output it produces, and the minimum data or configuration required to be useful. Be ready to say what happens when data is insufficient and which decisions remain with the marketer.
Mixing up consent models across regions and channels, assuming a single opt-in state or a single lawful basis applies everywhere.
Consent concepts look similar on the surface, but opt-in requirements, proof-of-consent expectations, and channel-specific consent (especially for messaging apps and SMS) differ by jurisdiction and by how the contact was acquired. Candidates learn one workflow and generalize it.
What to do instead
Map consent as a matrix: channel × region × acquisition source, noting what counts as valid consent and what evidence must be stored. When a scenario names a region or channel, check the matrix entry rather than defaulting to the workflow you know best.
Designing a retargeting or ads journey as if audiences sync instantly and one-way, ignoring match rates, refresh cadence, and how unsubscribes or exclusions propagate back.
Candidates think of ad audiences as a copy of a segment, so they miss that audience sync is periodic, subject to matching and platform minimums, and that failing to exclude converters or unsubscribers produces wasted spend and compliance risk.
What to do instead
For every audience-sync scenario, specify the source segment, the refresh frequency, the exclusion list, and the exit condition (e.g. purchase or unsubscribe). Learn typical match-rate and minimum-audience constraints and factor them into any question about expected reach.
Reading a scenario once and answering from the first familiar keyword instead of identifying the actual objective, constraints, and the single decision the question asks for.
Exam items are written with realistic detail that includes distractors - a channel, a metric, or a feature mentioned but not central. Under time pressure, candidates pattern-match a keyword to a memorized answer and skip the constraint that changes the correct option.
What to do instead
Use a fixed routine: underline the objective, list the stated constraints (channel, region, data availability, timing), then eliminate options that violate any constraint before choosing. Practice on long scenarios with a timer so the routine becomes automatic.