Leads & Marketing

ARPI holds no benchmark data for response time, contact rate, show rate, cost per lead, cost per sale or gross return, so it publishes no target for any of them and nothing on this page is coloured good or bad. The response bands are descriptive bins. Compare stores, sources and periods against each other; this project has no industry standard to offer you.

Each lead is counted once, at the furthest stage it reached, and the counts sum exactly to the valid leads in scope. They say where progression stopped. They say nothing about why, and nothing in this project could: no communication content, activity detail, note or disposition is modelled anywhere in it. Some leads that bought have no modelled showroom visit — a walk-in later matched to a lead — which is why the chain of stage rates only approximates lead-to-sale conversion rather than equalling it. That difference is expected and is not an error.

Spend is recorded by calendar month while leads arrive daily, so dividing a month of spend by part of a month of leads would produce a figure that looks precise and means nothing. Cost measures are computed only over whole months in the selected period, and spend is never prorated: this project governs no proration rule. For organic and internal sources, cost per lead, cost per sale and gross return are NOT APPLICABLE rather than zero — a walk-in is not a zero-cost advertising campaign, and a $0.00 cost per lead would rank it as the most efficient channel the group operates. Where spend produced no attributed lead or no attributed sale, that is stated as its own result rather than shown as an infinite or missing cost.

Cost per lead, cost per sale and gross return are ratios of sums at whatever group they are formed over: total spend over total attributed outcomes, never an average of per-campaign ratios, which would weight a campaign that produced two leads the same as one that produced two hundred. The same governed arithmetic is applied at three groups on this page — per source for the comparison, per source and campaign for the table, and over every cost-attributable row for the headline figures. KPI-MKT-001, KPI-MKT-002 and KPI-MKT-003 keep their published definitions at all three.

It is attributed gross divided by spend: a contribution measure. It nets out the cost of the vehicle and nothing else — no personnel, facility, floor-plan, overhead, tax or agency cost beyond the modelled spend is deducted anywhere in this project — so it is not profit, not net profit and not return on investment. A revenue-based return is deliberately not shown: vehicle revenue includes the cost of the vehicle, which inflates such a ratio by roughly an order of magnitude. Clicks and impressions are vendor-reported activity rather than value, and are not presented as marketing outcomes.

Attribution is single-source and first-touch: a lead is credited to exactly one source and one campaign, and a customer who arrived through three channels is credited to one. No multi-touch, linear, time-decay, last-touch, position-based or view-through model exists in this project. Sales and gross are attributed through the originating lead’s creation month rather than the sale date, so this month’s spend is never credited with last quarter’s leads — and the consequence is that the most recent cohort has had the least time to convert and always looks worst. ARPI defines no maturity horizon, so recent cohorts are shown in full and labelled rather than hidden or adjusted.

They are different populations on purpose. Vendors count leads their own way and typically count duplicates that the CRM de-duplicates, so the two figures are published side by side and never reconciled to each other. The gap is something to raise with a vendor; it is not automatically a data-quality failure, and the vendor figure is never substituted for the governed valid-lead count.

No lead, customer, salesperson, message, note, call recording, phone number or email address appears anywhere on this page, because none of it is exported: the response distribution is a histogram of counts with no identity attached to any bin. Every source, vendor and campaign named here is fictional, and the data is synthetic throughout. There is no employee ranking — that is a separate surface — and no recommendation: this page reports evidence and does not tell you what to do about it.