Marketing interviews test whether you can connect activity to numbers — channels, funnels, attribution, and unit economics. Below are the most common questions with model answers. Middle: deeper understanding, optimization, and real-world situations.
CAC is total acquisition spend divided by customers acquired in the same period, including salaries and tools, not just ad budget. LTV is average revenue per customer times the expected lifetime, minus the cost of serving them. A ratio around three to one is the usual rule of thumb, but the number that decides survival is payback period — how many months until a customer has returned what they cost.
It gives all the credit to the final touch, which is usually a branded search or a direct visit — the channel that closed, not the one that created demand. So content and social look worthless and search looks brilliant, and the budget moves accordingly until growth stalls. Any multi-touch model is better, and the cheapest sanity check is turning a channel off and watching what happens elsewhere.
When it reaches the sample size you calculated before starting, not when the result first looks good. Stopping early on a promising number is how teams ship changes that do nothing — with small samples the difference is noise most of the time. Also run it for whole weeks, because Tuesday traffic and Sunday traffic behave differently and a partial week biases the result.
Start from what you already know: the channel with the shortest payback gets the base, and a fixed share — often ten to twenty percent — is reserved for testing the ones you have no data on. Reallocate on a schedule rather than continuously, because daily fiddling reacts to noise. The discipline that matters is deciding in advance what result would make you stop a test.
Segment before theorising: is it all traffic or one source, one device, one country, one landing page. A drop confined to a single channel is usually a campaign or a tracking change; across all channels it is usually the site — a deploy, a broken form, a slow page. Check whether anything shipped, and check whether the analytics itself broke, because a tracking bug looks exactly like a conversion drop.
From the intersection of what people search for and what you can honestly answer better than what already ranks. Volume without intent is a trap — a keyword with ten thousand searches and no buying intent brings traffic and no customers. The other input is sales: the questions asked in every demo are the articles that convert, and nobody has to guess at them.
That people asked for it and open it. A large list built from a giveaway is mostly people who wanted the giveaway, and it shows up as a falling open rate and rising spam complaints, which then damages deliverability for everyone on it. Prune the inactive, measure by revenue per email rather than by list size, and make unsubscribing easy — a bad address costs more than an absent one.
The ideal customer profile describes the company worth selling to, not the person — industry, size, geography, tech stack, funding stage, and the trigger that creates the need. It is built from the customers you already have who renew and are cheap to serve, not from who you wish you sold to. Its value is that it makes disqualification possible, which is what stops a pipeline filling with deals that never close.
Who the product is for, what problem it displaces, why now, how it reaches them, and what the first hundred customers look like. The failure is treating it as a launch plan — a channel list without a positioning decision is just spending. It also has to say what you will not do, because a small team pursuing every segment reaches none of them convincingly.
It groups users by when they arrived and follows each group over time. Averages hide the direction: a product whose retention is worsening can show a stable average for months while new cohorts perform worse and old ones prop up the number. Cohorts answer whether the changes you shipped made anything better, which a single retention figure cannot.
ROI measures return on the whole investment; ROMI isolates the marketing spend — revenue attributable to marketing minus that spend, over the spend. ROMI is the more honest number for a marketing team because it does not take credit for product or sales, and the argument it starts is which revenue is attributable, which is the same attribution problem in another form.
With a written definition and examples of what it is not, because adjectives alone do not survive contact with three writers. The practical artefact is a short document with real before-and-after rewrites, plus the words you use for your own product and the ones you never use. Consistency is enforced at review, and the first thing to check is whether the voice changes when the channel does.
In a real interview you speak under pressure. Cam asks these same questions, scores every answer, and shows exactly what to fix.
Practice a Middle Marketing interview →