Six Gyms, 18,000 Sessions, and a Seven-Field Form
Summit Line operates six indoor climbing gyms and sells a $9 intro day pass online. The pass is the main path to an $89/month membership, and 22% of pass holders convert within 60 days. Last month the site received 18,000 sessions. Paid search cost $4,200 and drove 60% of those sessions, including bids on the brand name even though the homepage already ranks first organically for it. Mobile accounts for 78% of sessions but converts at 1.4%, versus 6.8% on desktop, producing 466 pass sign-ups overall (2.6%). The sign-up form asks for six fields, and 62% of people who start it never submit. The marketing lead proposes three moves: cut paid search by half, add a seventh form field asking 'how did you hear about us', and spend the savings on a billboard near gym #3.
1. Quantify where the funnel is leaking and state what you would fix first, showing the arithmetic.
Read the worked answer
Split the traffic before diagnosing. Mobile: 18,000 × 0.78 = 14,040 sessions × 1.4% = 197 passes. Desktop: 3,960 × 6.8% = 269 passes. If mobile converted at the desktop rate, mobile alone would produce 14,040 × 6.8% = 955 passes - a gain of roughly 758, which dwarfs every other lever on the table.
The form is the second leak. 466 completions at a 38% completion rate implies about 1,226 starts. Halving abandonment to 31% would yield 1,226 × 0.69 = 846 completions, about 380 more passes at zero media cost.
Order of operations should follow cost-to-fix against size of leak: (1) strip the form to two or three fields - no spend change, same day; (2) fix mobile friction (tap targets, load time, keyboard type on phone fields) and re-measure mobile conversion; (3) only then touch media budget. The seventh 'how did you hear about us' field is the wrong move at this stage: it adds friction to the step that is already losing 62% of starters, and self-reported attribution is weak data. Collect attribution through UTMs and a post-signup survey instead.
2. The lead wants to halve paid search and fund a billboard. Evaluate that trade.
Read the worked answer
The proposal conflates two different jobs. Paid search on non-brand terms captures existing demand at the moment of intent; a billboard creates awareness with no click, no landing page and no same-day attribution. Halving the demand-capture budget to fund the demand-creation channel removes the very conversions that make the billboard measurable.
The brand-name bidding needs splitting out first. If a meaningful share of the $4,200 goes to brand terms while the homepage already ranks first, some of that spend is defending against competitors conquesting the term rather than buying incremental clicks. Test it: pause brand bidding in one metro for four weeks and compare total branded sessions and pass sign-ups in matched control metros. If total volume barely moves, that spend is cannibalizing organic and can be cut.
For the billboard, agree the measurement before signing: a dedicated promo code, branded search lift in gym #3's metro versus a matched metro, and pass sign-ups by location. Judge it on incremental passes and cost per pass, and accept that it needs a longer window than paid search. Cut only the spend you have proven is non-incremental.
3. Name two metrics you would add and one you would stop reporting, with justification.
Read the worked answer
Add cost per pass by device and by query type (brand versus non-brand). Right now $4,200 is reported as a single lump, so nobody can see whether the mobile-heavy paid traffic pays for itself. If non-brand cost per pass is, say, $15 and expected 60-day revenue per pass is 0.22 × $89 = $19.58 before retention beyond 60 days, the channel is thin but plausible; the device and query split tells you where to trim rather than guessing.
Add step-level form completion, not just completion overall. Knowing that 62% abandon is useless without knowing which field kills them - start, field three, submit. That single view is what tells you whether the mobile problem is page performance or form design.
Stop reporting blended conversion rate as a headline number. A 2.6% blend hides a 1.4% mobile rate and makes every month look stable while 78% of traffic underperforms. Report mobile and desktop separately, or better, report passes per 1,000 sessions by device. Blended figures protect the status quo; segmented figures point at the fix.