# Case Studies | Crank Source: https://wearecrank.com/case-studies What twenty-seven account audits and Crank's own managed accounts found — real findings, real figures, sourced and named where the account can be. --- Case Studies # What the accounts actually showed Findings drawn from Crank's own audits and managed accounts — every figure traced to a source document or a database record. Named where the account can be, described plainly where it can't. [Beauty and skincareEcommerce2023–2026What Three Skincare Brands Found When They Stopped Reading One Blended NumberRetention now returns roughly one and a half to three times what acquisition returns across every brand and market measured — and 2025 revenue grew 38–79% while blended return fell, which the separation shows is the growth working, not a decline.Read the full story →](/case-studies/is-clinical-revitalash-deesse-pro)[Beauty and home fragranceEcommerce2016The Retailer Whose Cheapest Clicks Were Its Most Expensive SalesShopping campaigns turned out to cost four times as much per sale as brand search despite looking cheaper per click, and mobile cost more than double desktop per sale with basket completion nearly half — two findings invisible until the wrong metric was replaced.Read the full story →](/case-studies/beauty-home-fragrance-measurement-2016)[Beauty and home fragranceEcommerce2018The Site Relaunch That Cost Half the Revenue While Traffic Barely MovedThe new navigation had restructured around content ahead of products — only a third of visitors ever reached a product page — and the basket abandonment email, one of the highest-returning things in the account, had been left to underperform through the exact quarter it mattered most.Read the full story →](/case-studies/beauty-home-fragrance-replatform-2018)[Beauty and wellbeingEcommerce2017When Neither the Catalogue Nor the Checkout Could Be SeenBasket completion had already fallen from 73.9 to 62.2 per cent in a year, and most product categories were missing from the data entirely — meaning the business had been unable to see either what wasn't selling or why people were leaving at the till.Read the full story →](/case-studies/beauty-wellbeing-tracking-checkout-2017)[Men's groomingEcommerce2017The Account Where 'At Best a Guess' Was the Verdict on AttributionOne of the clearest findings was neither a tag nor a guess: a delivery cost that rose in step with average order value pointed to an unannounced minimum order threshold — order value rose and conversion fell as a direct, explicable consequence of a decision nobody had flagged to marketing.Read the full story →](/case-studies/mens-grooming-attribution-2017)[Beauty and skincareEcommerce2024Three Separate Failures, None Visible From Inside the Advertising PlatformCoverage was restored, a starved campaign acquiring customers at £13.94 was funded properly against one running freely at nearly double the cost, and the account had 303 additional monthly sales of modelled headroom at £16.02 each once the daily budget stopped switching off the best-performing days.Read the full story →](/case-studies/beauty-skincare-operational-audit-2024)[Games and collectablesEcommerce2026Ten Things Wrong, None Visible From Inside the Advertising PlatformNone of the ten issues found were visible from inside either advertising platform — reported performance could have been doubled, halved, or simply wrong, and the account had been making decisions on all of it as though it were reliable.Read the full story →](/case-studies/games-collectables-full-audit-2026)[SkincareEcommerce2026The Customer Database That Was One Setting Away From Being UsableA cleanly isolated cold-prospecting campaign — audience contamination already excluded — still lost money on every sale at $208.40 per acquisition and 0.32x return, confirming the problem was the offer and creative rather than who was being targeted, once everything else had been ruled out.Read the full story →](/case-studies/skincare-consent-retention-creative-2026)[Luxury department storeEcommerce2022A Growing Top Line Was Hiding a Worsening Unit CostShoppers who used site search converted three times better and content viewers converted 44 per cent better — both reaching under a quarter of visitors — while conversion varied nearly six-fold between departments of the same shop, none of it visible from the headline revenue and cost figures alone.Read the full story →](/case-studies/luxury-department-store-behavioural-2022)[Consumer services selling high-value appointmentsEnquiry-led2017The Decline That Wasn't a Marketing ProblemA steep performance decline traced back to a business decision — consultation locations cut from ten to four — that had been analysed as a marketing problem until the change was surfaced; unusually, this business could also state its enquiry-to-order rate and average order value, letting revenue be modelled directly from enquiry volume rather than guessed.Read the full story →](/case-studies/high-value-appointments-mediated-2017)[Financial technologyEnquiry-led2022Beating Every Published Benchmark and Still Unable to Say It Made MoneySeventy-one per cent of spend was going to two adverts costing over twice what the cheapest ones cost, and the target audience — after salespeople had already been excluded — still included marketing and research job functions who would never buy, quietly inflating volume and depressing quality.Read the full story →](/case-studies/fintech-benchmark-vs-profit-2022)[Speciality chemicalsEnquiry-led2025What a Business's First Digital Campaign Actually Costs to BuildCost per enquiry ranged four-fold between five campaigns run side by side on the same platform to the same market — from £226 to £886 — turning one blended average into a clear case for which product was worth scaling and which was worth stopping.Read the full story →](/case-studies/speciality-chemicals-first-time-advertiser-2025)[Data infrastructure softwareEnquiry-led2025–2026A Two-Year B2B Programme, and the Same Finding TwiceThe format finding repeated almost exactly a year later on the next campaign — direct messaging converting at 18.97 per cent against 0.94 for single-image ads — and a warm audience of 78,661 people, built at a penny each, sat completely unused because nothing was retargeting it.Read the full story →](/case-studies/data-infrastructure-b2b-two-year-programme)[Social impact softwareEnquiry-led2026The Channel That Looked Eight Times Cheaper and Was a Third More ExpensiveOnce qualification was traced through, the cheap channel converted at 5.3 per cent against 54.5 for the expensive one — cost per qualified enquiry came out at £901 against £665, meaning the channel that looked eight times cheaper was in fact a third more expensive.Read the full story →](/case-studies/social-impact-software-qualification-2026)[Commerce softwareEnquiry-led2020The Month a Quarter of the Traffic Disappeared and Enquiries TripledThe following month, traffic fell 25 per cent and enquiries rose 218 per cent — a quarter of the traffic disappeared in the same month that produced the best enquiry volume on record, which reported on volume alone would have read as a failure.Read the full story →](/case-studies/commerce-software-integrated-campaign-2020) ## More findings from the evidence base Single findings, from accounts too small a slice of the evidence for a full story on their own — but real, sourced, and worth stating plainly. Executive education 2019 ### A quarter of confirmed enquiries had no matching analytics record at all The CRM recorded 25% more brochure-download enquiries than the analytics platform for the same event. Of 10,647 confirmed CRM touchpoints, only 77% had a matching form-submission event and 74% a matching goal completion — six separate root causes were individually diagnosed, from a '#' used instead of '?' in email links to campaign names that differed between systems. Food and beverage 2021 ### A consent-handling bug corrupted a year of traffic data Without an accepted cookie, no persistent ID was set, so every page view created a brand-new visitor. Direct traffic share climbed from 18% to 70% of the whole site over the year — not a measurement gap, but data actively manufacturing a false picture. Footwear retail 2019 ### A rare example of customer segmentation done properly, before any campaign was built Most accounts in this evidence base have no separation between new and existing customers at all. This one had four segments — new, active, lapsed, and in-store-only — each with tracked year-on-year movement, forecasted revenue, and a named action. Home and garden retail 2020 ### The headline conversion number was wrong by eleven per cent The site's conversion goal fired on reaching a page, not on completing a transaction. It recorded 158,000 completions against 142,000 actual transactions in the same period — the number the account was being judged against was overstated before any campaign decision was made. Technology infrastructure 2026 ### The same advert cost three times as much depending where it ran Three campaigns promoting the same product, on the same platform, over the same five weeks: Europe cost $3.81 a click, North America $2.36, Asia $1.31\. In the same account, narrow B2B lead generation into Asia and India cost $18.96 and $18.12 a click against $5.86 into Latin America — geography moved price roughly threefold, and audience narrowness moved it further still, in the opposite direction. Business systems consultancy 2026 ### Half the enquiries were not people you could sell to Every contact from two months of automated-campaign enquiries was reviewed by hand. Of 68 enquiries, 11 were a real company with a relevant job title. Headline cost per enquiry was around £58; cost per genuine business enquiry was £385. Workforce software 2023 ### Qualification came back at 70%, and revenue never did Of 35 enquiries, 25 were confirmed valid — an unusually good 70% qualification rate. But pipeline value and deal value were never filled in, so a cost of £255 per valid enquiry couldn't be judged against what a customer was worth. Separately, of 368 people who opened the same business's enquiry form, only 35 finished it — a 9.5% completion rate, with cost per open at £11.13 and cost per finish at £182. Data security software (B2B) 2024 ### A format that won at one B2B account lost, by more than double, at another Document ads had beaten single-image ads by a wide margin at a data-infrastructure account. At this data-security account, the opposite happened: single-image ads produced enquiries at £303 against £712 for document ads, same platform, same month. Separately, fifteen named target accounts were tracked individually for engagement rather than reported as one blended figure — genuine account-based measurement, rare in this evidence base.