Here is one renter who signed a lease at one of your properties last month, and everything she did beforehand.

Nineteen days, seven touchpoints, four channels. One lease.

Now a simple question with a surprisingly contested answer: which channel produced it?

Four models, four right answers

• First touch gives the credit to the first tracked interaction. The social impression was never captured, because she did not click, so organic search becomes the first tracked touch.

• Last touch says paid search, because the guest card came through that ad. Everything before day 11 gets nothing.

• Linear divides the credit evenly across the tracked marketing touchpoints. If your system can see organic search, the listing portal and paid search, each receives a third. If it cannot see the portal activity, organic and paid split it between them.

• Your property management system says whatever went into the source field. Which is often not any of the above.

  Because think about how that field actually gets filled. The renter calls to confirm her tour. The agent asks how she heard about the property. She says "online." The agent picks Website from the dropdown, because that is the closest option, and moves on to the next call.

Four models, four answers, and all of them are defensible. None of them represents the entire journey. Each is applying a different rule to the interactions that were captured.

What falls out of every model

Look at what disappears.

The social post appeared early in the journey, but because she did not click it, that exposure may not appear in the tracked attribution path at all. The listing portal was part of the journey while she compared options, and appears in none of the models except linear, because the guest card did not come through it.

This is a systematic distortion, and it has a direction. Last-touch attribution rewards the channel present at the moment of inquiry, and gives nothing to touchpoints that came earlier in the journey.

That matters because renters research widely before they commit. Apartments.com's renter research finds that 85% use rental listing sites and that renters typically research around ten properties before seriously considering three. That kind of broad research behavior makes a multi-touch journey plausible, even though the survey does not tell us how many individual renters follow this exact path.

Now multiply that across a year of leases. Every lease credits one channel. The channels that assist rather than close get credited for almost nothing, their cost per lease looks terrible, and they become the obvious candidates when the budget gets cut. 

Cost per lead is the wrong denominator

Before getting to the fix, one more distortion worth separating out.

Most channel reporting shows cost per lead. It is easy to produce, because leads are counted where they arrive. It is also measuring the wrong thing.

A channel producing forty cheap leads that never tour is worse than a channel producing ten expensive leads that sign. Cost per lead cannot tell those apart, because it stops counting at the inquiry. The number that matters is cost per lease, which requires following each lead through to whether it signed. Our piece on the four conversion rates behind your headline number covers why the stages between inquiry and lease are where channels actually separate.

Which brings us to the obvious fix, and why it does not work the way it is sold.

The models being sold as the answer were retired by Google

The standard advice is to move from single-source to multi-touch attribution. Linear, position-based, time-decay, U-shaped. Spread the credit across the journey instead of handing it to one touchpoint.

It sounds right. It is also worth knowing that in 2023, Google retired first-click, linear, time-decay and position-based attribution from both Google Ads and Google Analytics 4, keeping only last-click and its data-driven model. A Google spokesperson cited low adoption, with fewer than 3% of conversions in Google Ads using the retired models. Google's own explanation, reported at the time, was that rule-based models do not provide the flexibility needed to adapt to evolving consumer journeys.

So the rule-based models that are sometimes presented to property managers as the modern answer are no longer available in Google Ads or GA4. Google moved to last-click and data-driven attribution instead.

The data-driven alternative has a different limitation, and for multifamily it is the decisive one. In a typical web analytics implementation, it sees the digital interactions available to that platform. It does not automatically see every listing-portal interaction, phone call, in-person tour or walk-in that happens outside it. For a leasing journey, a large share of what happens sits beyond any single platform's default view.

Neither path solves the problem, because neither addresses what is actually wrong.

Attribution is an identity problem, not a reporting problem

To give credit to the day 4 visit, the day 6 save and the day 11 click, you have to know they were all the same person.

That is harder than it sounds. She used her phone for the social feed, her laptop for the organic search, the portal's own messaging when she saved the listing, and her work email for the guest card. Listing portals often mask or relay contact details. The phone number she called from on day 12 may match nothing earlier in the record.

Without resolving those into one person, there is no journey to attribute. There are four unrelated events in four systems. We covered the mechanics of this, and why matching on email alone fails exactly where duplication is heaviest, in what counts as a lead.

So the order of work is the opposite of how it is usually sold. Upgrading the attribution model on top of fragmented identity produces a more sophisticated wrong answer. Fixing identity first makes even a simple model more useful than a complex one running on broken data.

When the lead and the lease happen at different properties

There is a version of this that portfolio operators face and single-property vendors almost never discuss.

A prospect inquires about a two-bedroom at your property on the east side. It is gone by the time they tour. Your agent offers a comparable unit at your property four miles west, and they sign there.

Which property's marketing gets credit?

The west-side property may show the lease under a new or differently attributed prospect record, while the east-side property's record still shows an unconverted lead. That can make the east property's cost per lease look worse and the west property's look better, while the cross-property journey that actually produced the lease remains invisible in both.

This matters most for operators who do it well. The better you are at matching prospects to units across properties, which is the argument of the leads you wrote off were just early, the more your property-level attribution understates what your marketing is achieving.

The fix is to record the originating property and the leasing property as separate fields on the lease, so the report can show both.

What an honest setup looks like

Not a model upgrade. Four changes, in this order.

• Capture source at the moment of inquiry, from the system, not the prospect. The dropdown filled in from a phone conversation is the least reliable data in your stack. Where a source can be captured automatically, from a tracked number, a tagged link or the portal feed itself, use that. Keep "how did you hear about us" as a secondary field, not the primary one.

• Keep one record per person across channels. This is the prerequisite for everything else. If the portal inquiry, the website guest card and the phone call create three records, no model downstream can reconstruct that renter's journey.

• Record every touchpoint, not only the first or last. Even if you report on a single-source basis for now, storing the full sequence means you can change how you analyse it later without having lost the data. You cannot attribute retroactively to touchpoints you never kept.

• Report cost per lease alongside cost per lead. Both numbers, side by side, by channel and by property. The gap between them is where your budget has been going to the wrong places.

None of this requires choosing a clever model. It requires the journey to exist as a single connected record. Once it does, the question of how to weight it becomes a reasonable analytical choice rather than a guess dressed up as data.

A note on what you can trust today

Until identity is resolved, treat your existing source report with care rather than suspicion.

It is not useless. For channels that genuinely close, it is probably directionally right. Where it misleads is on channels that assist: anything that builds awareness, keeps you on a shortlist, or reaches people before they are ready. Those will consistently look worse than they are.

So when a channel looks bad on the report, the right first question is not "should we cut this." It is "is this a channel that closes, or a channel that assists." The answer changes what the number means.

That lease will show up in your system with one source against it. Probably paid search, possibly Website, depending on who took the call.

That record will be used, eventually, to decide where next year's marketing budget goes. It will say the ad earned the lease. It will say nothing about the social post she saw, the organic visit, or the portal listing she saved.

Multiply that across every lease you signed last year, and the report on your desk is not a picture of what works. It is a picture of what happened last.

Fixing it starts before any model gets chosen, with the renter existing as one person across every channel she used. RIOO's leasing workflow brings inquiries and source information into a structured leasing process, while the unified customer view keeps inquiry, communication and leasing history together against the prospect, and dashboards and reports read from that same record. The journey becomes something you can look at, rather than four events in four systems.

Pick five leases signed last month. For each, try to list every touchpoint before the guest card. If you can find more than one for any of them, you already know your source report is incomplete. If you cannot find any, that tells you something more important.

Frequently asked questions

Q1. What is lead source attribution in multifamily?
The process of assigning credit for a signed lease to the marketing channels that contributed to it. A property management system may record a single source against a prospect or lease, which can give one channel full credit even when a renter interacted with several channels before signing.

Q2. What is the difference between first-touch and last-touch attribution?
First-touch credits the channel where a renter's tracked journey began. Last-touch credits the channel present when they converted, usually the inquiry. First-touch overvalues awareness channels, last-touch overvalues closing channels, and both ignore everything in between.

Q3. Should I switch to multi-touch attribution?
Not before resolving identity. Multi-touch models distribute credit across a journey, but they depend on knowing that separate touchpoints belong to the same person. Google also retired its rule-based multi-touch models from Google Ads and GA4 in 2023.

Q4. Why is cost per lease better than cost per lead?
Cost per lead stops counting at the inquiry, so it cannot distinguish a channel producing many leads that never tour from one producing fewer leads that sign. Cost per lease follows each lead through to the outcome that actually matters.

Q5. How do I attribute a lease when the prospect moved to a different property?
Record the originating property and the leasing property as separate fields. Otherwise the first property may show an unconverted lead and the second an unattributed lease, and the cross-property match that produced it disappears from both reports.