Category: Attribution & Measurement

Last-click attribution gives all the credit for a conversion to the final thing someone clicked. For a purely digital funnel that is a defensible simplification. For a business whose revenue happens in a building, it is close to useless — and worse, it is systematically misleading in one direction.

What last-click actually credits

Last-click follows a chain of digital touchpoints and awards the conversion to whichever came last before the purchase. It is cheap, immediate and available in every analytics platform, which is why it became the default.

Its logic holds together only while every meaningful interaction is a click, and the conversion is a click too.

Where the chain breaks

Consider a real sequence. Someone passes a billboard on Tuesday. On Thursday they mention the brand to a colleague. On Saturday they search the brand name, click the top result, and then drive to the store and buy something.

Last-click credits branded search. The billboard that created the intent gets nothing, because no click exists to record. Neither does the conversation. And the purchase itself — happening in a shop, in cash or on a card that analytics never sees — may not register at all.

The failure is not that last-click is imprecise. It is that the thing which caused the outcome is invisible to it by construction.

The structural bias this creates

This is the part that costs money. Last-click does not distribute error evenly; it consistently over-credits the bottom of the funnel.

Branded search sits closest to the conversion, so it collects the credit. Retargeting sits close too. Anything that builds awareness — out of home, radio, print, most social — sits further back and collects nothing.

Budget then follows the reporting. Spend moves toward the channels that appear to convert, which are mostly the channels harvesting demand that something else created. The demand-creating channels look inefficient, get cut, and the harvest shrinks a quarter or two later — by which point the cause is no longer obvious in the data.

Why this matters more for physical retail

For an ecommerce business, last-click is a flawed but functioning model, because the conversion is at least observable. For physical retail the conversion happens outside the measurable system entirely.

That means a retailer using last-click is not measuring a distorted version of the truth. It is measuring a different thing: the digital traffic that preceded a purchase it cannot see.

What to use instead

None of these replaces last-click for the jobs it does well. They answer the question last-click cannot: whether the advertising caused anything to happen in the physical world.

The common thread is a control group

What separates all four from last-click is that they compare against a counterfactual. Last-click asks what happened before a conversion. A controlled measurement asks what would have happened anyway — and only the difference belongs to the campaign.

That difference is the only number worth putting in front of a finance director.

Common questions

Is last-click attribution always wrong?

No. For measuring the efficiency of demand-harvesting channels within a digital funnel, it is fast and adequate. It becomes wrong when used to judge channels whose effect is upstream of a click, or outcomes that happen offline.

What is the difference between last-click and multi-touch attribution?

Multi-touch spreads credit across several digital touchpoints rather than one. It is an improvement, but it shares the same blind spot: it can only see interactions that produced a trackable event, so offline exposure and offline conversion remain invisible.

How do you attribute out of home advertising?

Through exposure rather than clicks — identifying who was plausibly near the advert during its run and comparing their subsequent visit rate against a matched unexposed group. Our guide to out of home advertising covers the formats this applies to.

Can offline sales be connected to advertising at all?

Not deterministically for every transaction, and any vendor claiming otherwise is overselling. What can be measured reliably is the incremental change in visits between exposed and control groups, which is the basis of a causal claim rather than a correlational one.

Next steps

If most of your revenue is earned in physical locations, the attribution model in your analytics platform is not measuring most of your business. Our visit attribution and measurement pages set out how real-world outcomes can be measured against a control.

James Ewen

James is the head of marketing at Tamoco

Visit attribution answers a question digital advertising has always struggled with: did the people who saw an advert actually turn up? Not did they click, not did they search afterwards, but did they physically walk into a store, showroom, restaurant or venue.

For any business whose revenue happens in a physical location, that is the only conversion that matters. This guide covers what visit attribution is, how the measurement actually works, what data it requires, and where it commonly goes wrong.

What is visit attribution?

Visit attribution is the practice of connecting advertising exposure to real-world visits. It uses anonymised location data to establish whether people who were exposed to a campaign subsequently visited a relevant physical location, and whether they did so at a higher rate than comparable people who were not exposed.

The output is not a count of visits. It is a comparison — the difference between an exposed group and a control group. That difference is the part the advertising can reasonably claim credit for.

Why click-based attribution fails for physical outcomes

Standard digital attribution follows a chain of clicks ending in an online action. That chain breaks the moment the conversion happens offline.

A shopper might see a billboard on Tuesday, think nothing of it, and visit the store on Saturday having never touched their phone in response. No click exists to attribute. Last-click models resolve this by crediting whatever digital touchpoint happened to come last — usually branded search — which systematically over-credits the bottom of the funnel and under-credits everything that created the intent in the first place.

For out of home, radio, print and much of social, this is not a small measurement gap. It is the whole picture.

How visit attribution works

The method is closer to a controlled experiment than to conventional ad tracking.

1. Define the exposed group

Identify anonymised devices that were plausibly exposed to the campaign — within range of a billboard during its run, or served a mobile impression. Exposure is probabilistic for physical media, and honest measurement treats it that way.

2. Define a control group

Identify a comparable unexposed group, matched on the characteristics that would otherwise explain a difference: location, movement patterns, time of day, device type. Without a matched control, any visit figure is meaningless — busy areas produce visits regardless of advertising.

3. Define the locations that count

Draw accurate boundaries around the places a visit could occur. This is harder than it sounds. A polygon that overlaps a neighbouring unit, a car park or a pavement will record visits that never happened.

4. Measure and compare

Compare visit rates between the two groups over a defined window. The difference is the lift attributable to the campaign.

Measurement approaches compared

What the measurement requires

Visit attribution is only as good as the data underneath it, and three things determine whether the result means anything.

Positional accuracy. Consumer location data varies enormously in precision. Data accurate to within a few metres can distinguish one shop from its neighbour; data accurate to a hundred metres cannot, and will quietly attribute visits to the wrong business.

Panel scale and consistency. The exposed and control groups must both be large enough that the difference between them is a signal rather than noise, and the panel must be stable across the measurement window.

Consent and provenance. The data must be collected with valid consent and be traceable to its source. This is a compliance requirement, and it is also a quality signal — supply chains that cannot explain their provenance usually cannot explain their accuracy either. Our guide to data transparency and consented collection covers what to ask for.

Where visit attribution goes wrong

Claiming correlation as causation. Reporting raw visits from an exposed group without a control group produces a number that looks impressive and means nothing. People near a billboard were always going to visit nearby shops.

Sloppy location boundaries. Polygons drawn around a rough address rather than the actual premises are the most common source of inflated results, particularly in dense retail environments and shopping centres.

Measurement windows that flatter. A window long enough will capture visits that would have happened anyway. The window should reflect a realistic decision cycle for the category — same-day for convenience, weeks for considered purchases.

Ignoring seasonality. A campaign running into a peak trading period will show lift that the calendar, not the creative, produced. Controls need to account for it.

Visit attribution: common questions

What is visit attribution in advertising?

Visit attribution is a measurement method that connects advertising exposure to physical visits, using anonymised location data to compare visit rates between people exposed to a campaign and a matched group who were not.

How is a store visit actually detected?

By comparing anonymised device location against a defined boundary for the location, applying a dwell-time threshold so that passers-by are not counted as visitors. The threshold matters: a few seconds indicates someone walking past, while several minutes indicates a genuine visit.

Can out of home advertising be attributed this way?

Yes, and it is one of the main reasons the channel has become measurable. Exposure is inferred from presence near the advert during its run rather than from a click. Our guide to out of home advertising covers the formats this applies to.

What is visit lift?

Visit lift is the difference in visit rate between the exposed and control groups, usually expressed as a percentage increase. It isolates the campaign’s contribution from visits that would have occurred anyway.

How long after a campaign should visits be measured?

It depends on the purchase cycle. Convenience and food categories tend to use short windows of a day or two; considered purchases such as furniture or vehicles need weeks. The window should be set before the campaign runs, not chosen afterwards to suit the result.

Next steps

Visit attribution turns offline advertising from an act of faith into something measurable — but only when the underlying location data is accurate, consented and matched against a proper control.

If you are evaluating measurement for a real-world campaign, our visit attribution and measurement product pages set out how Tamoco approaches each of the requirements above, and our guide to location data explains the data types involved.

James Ewen

James is the head of marketing at Tamoco

Marketers now have a bigger arsenal than ever before. It’s a digital arsenal, an outdoor advertising arsenal, a location-based, personalized, tailor-made arsenal of marketing channels that attract customers.

All this power comes with a problem; it’s hard to tell which marketing channels have the most significant impact on your business. Creating dozens of campaigns across several channels might get the word out about your business, but you’ve got to tie leads back to the marketing channel that delivered them if you want to know what’s working.

That’s where marketing attribution comes in. Attribution means tracking each touchpoint customers have with your brand and assigning credits to those touchpoints according to how much they impacted the customer decision. According to data from HubSpot, companies that use attribution to show marketing ROI have a 12% bigger marketing budget than those that don’t.

Without attribution, you won’t know how to adjust spend and allocate resources towards the most effective, efficient marketing channels. Let’s make sure that doesn’t happen.

The Issue of Offline Attribution

Attribution can solve the problem of how to give credit to offline conversions. Imagine that someone calls your business’s call center after seeing your billboard, bus stop ad, or brochure. How do you assign credit to that marketing source if you can’t track a click?

Call tracking numbers can solve this issue. Simply put a tracking number on the ad, link it to your real number, and use call tracking software to capture that lead and track it back to the marketing source.

Offline attribution methods like call tracking allow you to quantify revenue from non-digital ads. As for the digital ads? Many tracking tools use first-click attribution; when credit goes to the first click that brings a user to a website.

A Simple Guide to First-Click Attribution

Picture this; someone performs a Google search for “what is CRM software.” That user finds your site in the search results and clicks. They visit your site, see your number and eventually call your salespeople. This is a lead, and the credit for this lead goes to the first marketing source they clicked on; Google organic search results.

Now let’s imagine something different happened. Instead of calling your sales team, that user ended up leaving your site without making contact. A few days later, they saw one of your PPC ads while searching for CRM software again. Recognizing your company name, they clicked the ad, called the tracking number on the landing page, and got in touch with a salesperson.

The credit for this lead still goes to Google Organic because that was the first click this user made before interacting with your site. There’s a problem though; PPC marketing also played an essential role in converting this customer.

Without the ability to track and account for all your marketing sources, you’d risk overvaluing Google Organic and undervaluing the PPC campaign simply because first-click attribution didn’t split the credit evenly between the two marketing sources. Marketers have recognized the need for more data and multi-channel attribution; a 2016 study found that 39% of marketers planned to use six or more marketing channels over the next two years.

Expanding Attribution with Better Data

Landing Page reports and Lead Page Reports give you the extra insights you need to assign lead credit across all marketing sources accurately. Landing Page reports tell you which web page the lead first landed on when coming to your site, a tidbit that can help you identify the keyword they used to search for your product. For instance, if they land on a specific product page, they likely used a keyword related to that product.

Landing Page reports can also reveal the other marketing methods leads used to visit your site. You may be sending out emails with links to pages on your website, or you may have backlinks from other sites, and landing page reports can reveal whether or not the people who clicked on those links ended up converting.

Lead Page Reports are a little different. They reveal which page the lead was on when they took a conversion action (calling, filling out a form, or initiating a chat). These reports reveal how effective various web pages are at turning prospects into leads.

First-touch attribution may only reveal one marketing channel’s effectiveness, but combining attribution data with a landing page and lead page reports will reveal how the various marketing channels intersect and impact the customer journey. In fact, you can even bring location data into the mix.

Realizing the Power of Real World Store Attribution

Attribution isn’t just for digital businesses. Integrating a real-world store attribution tool into your marketing stack will tell you when leads saw your marketing, when they came into your business and when they made a purchase. For brick and mortar businesses, this is a must. It’s also no longer a futuristic technology reserved for a select few; real-world store attribution tools are available.

With these tools, you can see how to drive traffic to your Shopify store, as well as your real0world store. You can see how long people stay in your stores, how often they visit, and demographic information about each customer. Maybe some marketing efforts drive people to your stores, but those people don’t make large purchases. You might be able to identify the marketing efforts that drive high-value customers to your stores, who make large purchases and come back for more.

The prevalence of mobile devices among the population has made real-world store attribution possible. However, these devices are just one part of the multi-device ecosystem that marketers must follow to capture every lead and trace it back to the marketing source.

The Need for Cross-Device Attribution

According to this Google research project, 51% of consumers will search for products across multiple devices, and 48% will directly navigate to their destination website from a different device than they used to find the site initially. Someone might search for a new CRM software on their phone, find a product they like, and then use their desktop computer to navigate that website directly and buy the product. Attribution has to be able to keep up with these changes.

That’s a lot of information to keep track of. Luckily for us marketers, that study was conducted in 2012, and attribution technology has caught up since then. While most consumers probably use multiple devices to shop even more frequently in 2020, it’s easier to track all searching, browsing, and shopping with cross-device attribution tools.

A Wrap-up on Attribution

Attribution is one of those marketing phrases that gets thrown around a lot, but it boils down to how well you can understand what marketing works. It’s the process of capturing data, tying it all together, and creating simple reports that explain what’s going on. If you can do that, you can calculate the exact impact your marketing has on the business’ bottom line. That’s valuable information, and it can make you, as a marketer, invaluable to your company.

James Ewen

James is the head of marketing at Tamoco