Nick QuirkEvery agency has a client who has been "growing impressions 40% year over year" for three years without a single additional sale. The reporting looks great. The relationship is one bad quarter from ending.
That's the cost of vanity metrics: they don't just fail to inform, they actively buy time for work that isn't working. Here's how to spot them, what to replace them with, and how to track the numbers that actually predict client retention and agency profit.
What makes a metric a vanity metric
Run every number in your reporting deck through three tests:
- The decision test. If this number doubled next month, what would you do differently? If the answer is "nothing," it's a vanity metric.
- The direction test. Can this number go down? Cumulative totals — total blog posts published, total backlinks, total keywords ranking — only ever climb. A metric that can't deliver bad news can't deliver good news either.
- The attribution test. Could this have moved without you? Brand impressions rise when a client runs a TV spot. Traffic rises in Q4 for e-commerce. If you can't separate your work from the tide, don't take credit for the tide.
Impressions fail all three. So does "tasks completed," which is the delivery-side equivalent — a number that rewards volume of activity over impact, and which we've written about at length in activity metrics vs ROI in client reports.
The five vanity KPIs agencies still report — and their replacements
1. Impressions → non-brand clicks by intent tier
Impressions inflate when Google tests you on page 8 for irrelevant terms. Split Search Console data into brand and non-brand, then bucket non-brand queries into commercial, comparison, and informational. A client going from 4,200 to 5,900 monthly non-brand commercial clicks is a real story. "Impressions up 340,000" is noise.
2. Total keywords ranking → tracked money keywords, segmented
Pick 20–40 keywords tied to revenue. Report position bands (1–3, 4–10, 11–20, 21+) and movement between bands. Six terms moving from page two to positions 4–10 is more meaningful than 900 new long-tail rankings on position 67.
3. Domain Rating / Domain Authority → referring domains that pass qualified traffic
DR is a third-party estimate you cannot bank. Track new referring domains from sites in the client's topical space, and how many of those links have sent at least one session in 90 days. An agency we know found that 71% of their "wins" for one client had sent zero referral traffic — which reframed the entire link strategy.
4. Social followers → assisted conversions and branded search lift
Followers are the purest vanity metric in marketing. If social is in scope, tie it to something: branded search volume over time, or conversions where a social touch appeared in the path.
5. Hours logged → outcome per hour
Clients don't buy hours; they buy results. Internally, hours are essential (see below). Externally, translate them: "18 hours on technical fixes → crawl errors down from 3,400 to 190 → 22% more pages indexed."
The three layers of metrics that matter
Useful agency measurement has three layers, and most agencies only build one.
Layer 1: Client business outcomes
One or two numbers the client's CFO would recognize: qualified leads, pipeline value from organic, revenue, cost per acquisition, or bookings. You will not always have clean access. Ask for it in onboarding, and if you can't get revenue, negotiate a proxy — form fills that reach a certain lead score, demo requests, or calls over 90 seconds.
Example: a B2B SaaS retainer at $9,000/month generating 31 demo requests from organic at a $4,100 average deal value and a 19% close rate is producing roughly $24,000 in monthly bookings. That's a 2.7x. Now the renewal conversation writes itself.
Layer 2: Leading indicators you control
Outcomes lag by 3–9 months in SEO, which is why you need honest mid-funnel signals: non-brand commercial clicks, position bands for tracked keywords, indexed page count, Core Web Vitals pass rate, content published against plan, referring domains earned. These are the metrics that let you say "the machine is working, the revenue is coming" — and back it up. Pair them with a clear narrative about timeline, as in explaining SEO timelines to impatient clients.
Layer 3: Agency health metrics (clients never see these)
This is where most agencies are flying blind. Four numbers:
- Effective hourly rate per retainer. Retainer fee ÷ hours logged. A $8,000/month retainer that consumed 74 hours is running at $108/hour. If your target blended rate is $150, that account is 28% underwater regardless of how happy the client is.
- Scope creep percentage. Hours logged outside the contracted deliverables ÷ total hours. Above 15% and you have a scoping problem, not a client problem. Tag time entries by deliverable so this is a report, not an investigation.
- On-time phase completion rate. Percentage of project phases that closed within their planned window. If audits ship on time but content phases slip 60% of the time, you know exactly where to add capacity.
- Net revenue retention. Revenue from existing clients this quarter vs. the same cohort last quarter, including expansions and downgrades. Logo churn hides the client who quietly dropped from $12k to $5k.
Layer 3 only works if time tracking is attached to the actual work rather than logged in a separate app someone forgets to open on Fridays. That's the practical case for keeping retainers, phases, and hours in one system — it's the reason PeakKR ties time entries to deliverables, and it's a real differentiator when you compare agency PM tools that treat time tracking as a bolt-on.
A worked example: choosing five KPIs for one client
Local multi-location home services client, $6,500/month, goal is booked jobs.
- Booked jobs from organic (outcome) — pulled from their CRM, monthly.
- Calls over 60 seconds from GBP and organic landing pages (outcome proxy) — because CRM attribution is incomplete.
- Tracked keywords in positions 1–3 across 12 service+city terms (leading) — the terms that produce calls.
- Location pages with complete NAP, reviews, and service schema (leading, and fully in our control) — 14 of 22 at baseline.
- Deliverables shipped vs. planned this month (delivery) — so the client sees velocity without seeing task counts.
Notice what's missing: impressions, DA,
