THE SALE
BEFORE
THE CLICK.

The click is visible. The reasons behind it are not.

Digital advertising has given marketers an extraordinary ability to observe behaviour. We can see who clicked, what they clicked, when they clicked and whether they bought something afterwards.

What we cannot necessarily see is whether the advertising caused the purchase.

This is the distinction at the heart of one of marketing's most expensive misunderstandings: confusing the ability to measure a sale with the ability to explain it.

The distinction matters particularly in search advertising, where marketers can end up paying to intercept customers who were already on their way to buy.

01 / THE SHORT ANSWER

A customer searches for a brand they already know, clicks its sponsored result and buys something. The advertising dashboard records a conversion. The agency reports a return on advertising spend. Everyone congratulates themselves.

But the customer may have bought anyway.

Research by Beknazarova and colleagues, involving nine field experiments across eight small, online-only direct-to-consumer brands, found no statistically significant sales effect from stopping branded search advertising.

When both branded and generic search advertising were stopped, website visits fell in two of three experiments, but there was no statistically significant effect on sales.

This doesn't mean search advertising never works. It means the relationship between clicking an advertisement and buying something is not necessarily the relationship between being influenced by an advertisement and buying something.

Consumer psychology helps explain why.

Our decisions are shaped by memories, existing preferences, circumstances and goals that precede the search. The sponsored result may simply be the final, highly visible step in a journey that was already heading towards a purchase.

02 / THE SALE BEFORE THE CLICK

Imagine somebody needs a new pair of running shoes.

They have worn the same brand for years. The shoes fit, they're comfortable, and the customer knows which model they want.

They open Google and type the brand name.

At the top of the results appears a sponsored link. They click it, find their shoes and complete the purchase.

The advertising platform reports a successful conversion.

But what did the advertisement actually accomplish?

It didn't necessarily introduce the brand. It didn't necessarily create the need. It didn't necessarily persuade the customer that this particular manufacturer was preferable to another.

Those things may already have happened.

Perhaps the customer originally discovered the brand through a recommendation. Perhaps years of wearing the shoes established a preference. Perhaps distinctive advertising made the name easy to remember. Perhaps the shoes were simply available when they first needed them.

Whatever the combination, the customer arrived at Google already equipped with a brand name and a reason to buy.

The search advertisement may have provided a convenient doorway.

It did not necessarily build the house.

These are not interchangeable.

The question isn't whether the sponsored link received the click. It did.

The question is whether the purchase depended on it.

03 / WHAT THE EXPERIMENTS ACTUALLY FOUND

The distinction between branded and generic search matters.

Branded search involves searches containing a particular brand name, such as "Nike running shoes" or "John Lewis washing machine".

Generic search involves category or product terms without a specified brand, such as "running shoes for beginners" or "best washing machine for small kitchen".

The first often reveals that a particular brand is already in the buyer's mind. The second may offer an opportunity to introduce brands the buyer hasn't yet considered.

In their paper Do brands need search advertising?, Ilmira Beknazarova, Justin Cohen, Carl Driesener, Nicole Hartnett and Steven Bellman investigated what happened when search advertising was withdrawn.

Across nine field experiments involving eight small online-only brands, stopping branded search advertising produced no statistically significant sales effect.

In experiments withdrawing branded and generic advertising together, visits declined significantly in two of three cases, but sales did not show a statistically significant change.

The researchers also conducted a supplementary online experiment measuring intended clicks, not completed purchases.

The study's scope matters. The field experiments used non-random assignment, involved a limited group of online-only businesses, and cannot establish that the true effect of branded search is precisely zero.

The findings are nevertheless a reason to question the assumption that attributed search sales are necessarily additional sales.

THE EVIDENCE, SEPARATED FROM THE INTERPRETATION

These results also have an important predecessor.

In research published in Econometrica in 2015, Thomas Blake, Chris Nosko and Steven Tadelis reported large-scale field experiments at eBay.

They found no measurable short-term benefit from branded keyword advertising.

Generic search advertising had a more interesting pattern: new and infrequent customers responded positively, while frequent customers accounted for substantial advertising expenditure without corresponding incremental purchasing effects.

For somebody who rarely uses eBay, a generic search advertisement might introduce a relevant shopping destination.

For somebody who visits eBay habitually, the same advertisement might simply provide a paid shortcut to a website they already intended to visit.

The commercial value of an advertising exposure depends partly on what the buyer already knows and intends to do.

04 / WHAT THE BUYER'S MIND IS DOING

Marketing science tells us what happened in the experiments. Consumer psychology can help us understand why such results are plausible.

There are three important processes.

MEMORY: THE BRAND WAS ALREADY THERE

When somebody types a particular brand name into Google, something significant has already happened.

They have retrieved that brand from memory.

The brand may have been encountered through advertising, previous purchases, recommendations, packaging, sponsorship, social conversations or simple familiarity.

It may be connected to a particular buying situation through what Ehrenberg-Bass researchers call a Category Entry Point .

A Category Entry Point is a cue that can bring a category, and potentially particular brands within it, to mind. It might be a need, an occasion, a location, an activity or an emotional state.

For example, somebody realises they're hosting friends at the weekend. That situation prompts thoughts about food, drinks and perhaps particular supermarkets or delivery services.

The brand retrieval may happen before a search engine is opened.

The search engine then records what is effectively a later stage of the decision.

This doesn't mean the brand's earlier advertising necessarily caused that particular purchase. It means that the sponsored search click alone cannot tell us how the buyer came to prefer or remember the brand.

GOALS: PEOPLE ARE OFTEN LOOKING FOR A DESTINATION

A search query isn't always a request to be persuaded.

Sometimes it's simply a way of getting somewhere.

If I type the name of my bank into Google, I'm probably not inviting competing banks to make their case. I'm trying to reach my bank's website.

If the sponsored link disappears and the official organic result remains easy to find, I may simply click that instead.

The route changes. The destination doesn't.

This is consistent with the experimental results, although the studies don't establish the precise psychological mechanism behind every substituted click.

The important distinction is between helping somebody complete an intended action and changing what they intend to do.

Both can have value. They are not necessarily worth the same amount.

EFFORT: CONVENIENCE IS NOT PERSUASION

Human decision-making is sensitive to effort.

We often choose an easy, obvious route rather than undertake unnecessary work.

A prominent search advertisement can make a website easier to reach, particularly when organic results are difficult to locate or competitors occupy valuable space.

That convenience can be commercially useful.

But reducing the effort required to reach a chosen brand is not the same as persuading somebody to choose that brand.

A sponsored result may save a customer a few seconds without generating an additional sale.

Or, under different circumstances, it might remove enough friction to prevent an abandoned purchase.

Which happened is an empirical question, not something the click count can settle.

There's a further distinction worth making. Clear purchase intent does not necessarily mean high involvement.

Somebody can know exactly which brand of toothpaste they want while giving the purchase almost no conscious thought. Another person might spend days comparing insurance policies without having settled on a provider.

Clarity of destination and depth of deliberation are different things.

A search for a specific brand can represent a habitual, low-effort action just as easily as the conclusion of an elaborate decision.

05 / WHY THE DASHBOARD TELLS SUCH A CONVINCING STORY

The appeal of performance marketing is that it appears to make causality visible.

An advertisement appeared. Somebody clicked it. A purchase followed.

Three events. One tidy narrative.

And a return on advertising spend calculated to two decimal places.

The difficulty is that the most convincing-looking measurement can still be answering the wrong question.

Imagine a shop assistant standing just inside a supermarket entrance.

Every customer who walks past is already heading into the store. The assistant smiles, hands them a leaflet and directs them towards the aisles.

At the till, the supermarket records every customer who received a leaflet and made a purchase.

The assistant's conversion rate is magnificent.

But what would those customers have done without the leaflet?

This is the missing counterfactual.

A search advertisement can face the same problem. It reaches people who have already expressed interest, sometimes very strong interest, in a brand or product.

That makes them unusually likely to convert.

It does not necessarily make them unusually likely to be changed by the advertisement.