Ask a marketer for a comprehensive analytics strategy and you get a dashboard with forty tiles. Ask an owner what they did with it last month and you get a pause.
Owners ask me for a comprehensive analytics strategy the way they would ask for a comprehensive insurance policy, as if more coverage were always better. I have built measurement from the ground up for companies more than once: deciding what to measure, setting up the tracking and reporting, using the results to change what gets done. The part that surprises people is how little of it is tracking.
A measurement strategy is a list of decisions, not a list of metrics. If a number on the report would not change what somebody does next month, it should not be on the report. Everything below follows from that one sentence.
A 38-tile dashboard nobody had opened in four months
The company manages about 1,800 residential units across 60 buildings in and around Detroit. Two kinds of customers walk through the website. Tenants looking for a place, who fill vacancies. And property owners looking for someone to manage their buildings, who are the actual business. One new management contract is worth more than a year of leasing fees on a single unit.
Marketing spend went to Google Ads, paid listing syndication for vacancies, and an SEO retainer. Reporting was a Looker Studio dashboard with 38 tiles, built by a previous agency. Sessions, pageviews, bounce rate, top pages, device split, a map. GA4 had 41 events configured. The dashboard’s own usage data said the owner had last opened it four months earlier.
What the owner wanted to know was one thing: “Which of the money I spend brings me a new owner?” None of the 38 tiles answered it. The site was recording everything and measuring nothing.
Before: 38 tiles Opened 4 months ago
After: five numbers Read every month
Start from the decisions, not the data
Before I opened GA4, I sat with the owner and the operations manager and asked what they decided every month. Not what they wanted to know. What they decided. It came down to three things: where next month’s ad budget goes, which vacancies get a rent reduction, and whether to hire a second leasing agent.
Then, for each decision, one question: what number would have to move for you to decide differently? That question produced the five metrics. It also killed most of the dashboard, because no decision depended on bounce rate.
| Decision made monthly | The number that drives it | Where it comes from |
|---|---|---|
| Where the ad budget goes | Qualified owner inquiries and cost per inquiry, by channel | GA4 key events, call tracking, ad platforms |
| Which vacancies get a rent drop | Applications started per vacancy in the first 14 days | GA4 key event joined to the listing ID |
| Whether to hire a second leasing agent | Days to lease, and the applications-to-showing gap | Property management software, not the website |
| Whether to trust the other four | Share of leads with no known source | CRM source field, checked against tracking |
Two of those numbers do not come from the website at all. That is on purpose. A marketing analytics strategy that stops at the browser tells you what people clicked. The owner does not get paid for clicks. Days to lease lives in the property management software, and it belongs on the same page as the ad numbers, because it is the thing the ad numbers exist to move.
What we deliberately did not measure
This list took longer to agree on than the first one. Every metric we cut had someone who liked it.
| Cut from the report | Why |
|---|---|
| Bounce rate, time on site | No decision changed when they moved. Listing pages have a high bounce by design: the tenant found the unit, called, and left. |
| Sessions and pageviews | Went up every spring regardless of anything we did. Useful for a trend line once a year, not for a monthly decision. |
| Keyword rankings | Kept in the SEO retainer’s own report. The owner never needed to see them. |
| Social followers and reach | Nobody could name a management contract that started on Facebook. Until someone can, it is not a number. |
| 35 of the 41 GA4 events | Scroll depth, video, outbound clicks on the footer. Retired. Six key events remain, and each one is a lead or a step toward one. |
Tracking is a week of work, and it goes first
I have written about this before and will keep writing about it: build the tracking before you touch anything else, and record a baseline. Here it was a clean Google Tag Manager container, six GA4 key events, and call tracking with a different number for each channel so a phone call could be traced to the ad, the listing site or the organic search that produced it.
The one rule that made the whole thing work: a lead only counts as qualified if the form was for owner services or the call ran longer than 90 seconds. The company’s inbound calls were mostly tenants asking about maintenance. Counting every call would have made the ad numbers look wonderful and meant nothing.
Four weeks of baseline, then the first report.
One report, three readers, three different pages
The mistake in most reporting is one dashboard for everyone. The owner, the marketing lead and the leasing team make different decisions on different schedules. They got different pages, built from the same data.
Owner
One page, monthly
- The five numbers, each with a status: on track, watch, act
- One sentence per number explaining what moved and why
- No charts. Twelve-month table underneath for anyone who wants it
Marketing lead
Channel table, weekly
- Qualified inquiries and cost per inquiry by channel and campaign
- Unknown-source share, so tagging problems show up before spend decisions
- Budget recommendation for the following week, written, not implied
Leasing team
Vacancy list, every Monday
- Each vacancy with listing views, applications started, days listed
- Units past day 14 with fewer than three applications flagged for a rent review
- Nothing about marketing spend. They do not decide it
The owner’s page has no charts on it. That was his request and it was the right one. A status word and one sentence beats a line graph for someone who is going to spend two minutes on it in the truck between buildings.
The rule for when a number moves
A metric without a threshold is trivia. For each of the five numbers we wrote down, in advance, the level that triggers a conversation and the level that triggers an action, and who owns the action. That last part matters most. A number that moves and has no owner just sits there being interesting.
| Number | Watch | Act | Who acts, and how |
|---|---|---|---|
| Cost per qualified owner inquiry | Over $450 for two weeks | Over $520 for two weeks | Marketing lead pauses the worst campaign and moves the budget to the best one |
| Qualified owner inquiries | Under 10 a month | Under 7 for two months | Owner and I review the offer and the landing page, not the ads |
| Applications per vacancy | Under 4 at day 14 | Under 3 at day 14 | Leasing manager reviews rent and photos on that unit |
| Days to lease | Over 28 average | Over 35 average | Owner decides on the second leasing agent |
| Unknown-source share | Over 20% | Over 25% | I fix tagging and call routing before anyone touches spend |
The last row is the one I would defend hardest. If a quarter of your leads have no known source, your channel numbers are fiction, and any budget decision made on them is a guess with a spreadsheet attached. Fix the measurement first, then decide.
What attribution could not prove, and how we said so
Property management leads come from everywhere. Yard signs. A referral from an owner who already uses the company. A tenant’s landlord who liked how the maintenance calls were handled. A call from a number a contractor had on a business card for three years. None of that shows up in GA4.
So the report says what it cannot see. The unknown-source number is on the owner’s page, in the same font as the others, and in the first month it was 41 percent. That was uncomfortable and it was the most useful number on the page, because it stopped a decision to double the Google Ads budget until we knew whether Google Ads was working.
Three honest limits stayed on the report every month: last-click attribution credits the final step and undercounts the first one; the “how did you hear about us” field disagreed with the tracking about a fifth of the time and we reported both; and referrals from existing owners were counted by asking, not by tracking, because there is no tag for a conversation at a hardware store.
What changed in six months
With the thresholds written down, the decisions made themselves. Two Google Ads campaigns aimed at tenants were paused in month two, about $2,400 a month, because tenants were arriving fine through the listing sites and the ads were producing maintenance calls. That budget moved to owner-services search terms and a landing page written for a landlord with three buildings and no time.
Same total marketing budget. Owner-acquisition spend went from about $3,800 to $6,200 a month; inquiries grew faster than spend because the landing page did its job. Applications per vacancy and days to lease held steady, so moving money away from tenant ads did not hurt leasing.
Fourteen contracts from about a hundred qualified inquiries is a 14 percent close rate, up from 11. The number I care about most is that fourth one, and it is the one the website did not produce. The contracts came from the owner’s sales conversations. The measurement strategy did not close them. It told him which conversations to have, and it stopped him spending money on the ones that were not going to happen.
The method, so you can run it on your own business
- List the decisions, not the questions. What does each person change every week or month? Three to five decisions is normal.
- One number per decision. What would have to move for you to decide differently? If nothing, the decision is not data-driven and that is fine, just do not pretend.
- Write the cut list. Every metric you drop, with the reason. This is the document people argue about, so have it.
- Include the numbers the website cannot give you. Revenue, days to lease, closed deals. Put them beside the marketing numbers or the marketing numbers will get worshipped.
- Build the tracking, then wait. A clean tag container, key events that mean a lead, call tracking by channel, four weeks of baseline.
- One report per reader. Match the page to the decision and the schedule. The owner’s page can have no charts.
- Write the thresholds and the owners in advance. Watch level, act level, who acts, what they do.
- Put the unknown-source number on the page. If it is high, fix measurement before you fix marketing.