OMCP common mistakes
OMCP · Understand the traps and how to avoid them.
Treating a rising bounce rate as proof the page is bad and recommending a redesign, when the page's purpose is to answer a question in one screen (definitions, store hours, a single-fact landing page).
Candidates memorize "lower bounce rate is better" and apply it universally, forgetting that a successful single-page visit is recorded as a bounce with zero engagement time.
What to do instead
Before judging bounce rate, state the page's conversion goal. For single-intent pages, pair bounce rate with a downstream metric (click-to-call, directions tap, next-page rate). For multi-step pages, use exit rate on the step after entry instead.
Reporting a 40% lift in conversion rate after a test and recommending a full rollout, without checking whether the result reached statistical significance or ran a full business cycle.
Stakeholders reward decisive answers, and analytics dashboards display point estimates with no confidence interval, so small-sample noise looks like a win.
What to do instead
Report the sample size, the confidence interval, and the test duration. Require at least one full business cycle (typically two weeks) and a pre-computed minimum detectable effect. If the interval crosses zero, call it inconclusive rather than a win.
Adding "best," "cheap," or "near me" to every target keyword and calling the resulting list a keyword strategy.
Modifier-stuffing feels productive and produces long lists, but it ignores intent segmentation and search volume reality.
What to do instead
Group keywords by intent (informational, commercial investigation, transactional, navigational) and map each group to a page type. Validate volume and difficulty in a keyword tool, and check the live SERP to confirm the intent matches the content you plan to build.
Writing ad copy and landing pages that promise different things - e.g., the ad says "free trial, no card" and the landing page asks for a card number.
PPC and landing-page work are often studied as separate domains, so candidates optimize each in isolation and miss message-match as a Quality Score and conversion driver.
What to do instead
Audit ad-to-landing-page alignment as one unit: headline, offer, and CTA must repeat the ad's promise. In PPC answers, name ad relevance, expected CTR, and landing page experience as the three Quality Score components and tie the fix to the relevant one.
Answering an email deliverability question by recommending a better subject line.
Subject lines are the most visible email lever, so candidates reach for them even when the scenario describes spam-folder placement, high bounce rates, or unsubscribes.
What to do instead
Diagnose by symptom: inbox placement → authentication (SPF, DKIM, DMARC), list hygiene, and sender reputation; low open rate with good deliverability → subject line and sender name; high unsubscribe → frequency and relevance. Match the fix to the stated symptom.
Listing every social platform as appropriate for every business, instead of choosing channels from the audience and content format.
Platform checklists are easy to memorize, and candidates fear omitting one, so answers become generic coverage rather than a recommendation.
What to do instead
Justify channel choice with audience demographics, content format fit (short video vs. long-form vs. professional network), and resourcing. Explicitly say which channels to deprioritize and why; a defensible "no" scores better than an undifferentiated "yes."
Recommending a mobile app when the scenario describes a small business needing bookings, or designing a mobile page that hides the phone number and primary CTA below the fold.
"Mobile" is equated with "app," and desktop layouts are shrunk rather than redesigned, so the mobile-specific constraints (thumb reach, one-column layout, tap targets, load time on cellular) are never applied.
What to do instead
Start from the mobile task: what does the user need in the first screen? Recommend responsive web for discovery and low-frequency tasks, an app only for repeat high-frequency use. Specify tap target size (about 48px), above-the-fold CTA placement, and click-to-call.
Producing a content calendar of publish dates with no mapping from each piece to a funnel stage, audience, or distribution plan.
Calendars look like deliverables, so candidates stop at scheduling and skip the strategic layer the exam is testing.
What to do instead
For each planned piece, state the target persona, funnel stage, primary keyword or topic cluster, and the distribution channels and repurposing path. A content plan without distribution is half a plan.
Answering an attribution question by crediting the last click, or by asserting that one model is "correct."
Last-click is the default in most analytics tools, so it is absorbed as the truth rather than recognized as one model among several with known biases.
What to do instead
Name the model and its bias: last-click over-credits closers (branded search, retargeting), first-click over-credits discovery. Recommend a model matched to the question (position-based or data-driven for mixed funnels) and note that no model is objectively correct.
Treating technical SEO as a content problem - recommending more blog posts for a site that is not being indexed or is suffering from duplicate URLs and slow mobile load.
Content is the most familiar SEO lever, and candidates cannot see crawl or index data in a text-only scenario, so they default to what they know.
What to do instead
Triage in order: crawlability and indexation (robots.txt, canonical tags, sitemaps), then site speed and Core Web Vitals, then on-page relevance, then content volume. If pages are not indexed, new content cannot help; fix access first.