Every new road is cheap while it is being paved. The owner wants the traffic before the toll, because a toll on an empty road collects nothing, and the habit of driving a route is worth more than any single fare collected on it. So the early years of a road are a quiet subsidy, paid by the owner to the people willing to try it, and the subsidy ends on a schedule the drivers never get to see. The toll comes after the road. It always has.
In 2012 the new road was the phone. People were searching on small screens in enormous numbers, and advertisers were paying less to reach them, partly because a click on a phone was worth less to most businesses and partly because most businesses had not yet worked out what it was worth at all. The effect showed up in Google's own numbers. Its average cost per click had not risen year over year since the third quarter of 2011; in the third quarter of 2012 it fell 15 percent, and it kept falling through the first half of 2013, down another 6 percent in the second quarter even as paid clicks grew 23 percent. Analysts read those figures as a problem in the business. They were a construction site.
The Booth on the On Ramp
The toll arrived in February 2013, under a name that sounded like a feature. Google introduced enhanced campaigns, and Larry Page called the change the biggest ever made to the company's advertising system. Every campaign would be upgraded automatically by July 22 of that year. Device targeting went away: an advertiser could no longer buy computers alone, and the only way to stay off phones entirely was to cut mobile bids by 100 percent. Phones were folded into every campaign and priced through the same auction as the desktop clicks advertisers already valued. Two years later Google reported that more searches took place on mobile devices than on computers in ten countries, including the United States. The road was finished, the traffic depended on it, and the booth had been built into the on ramp rather than beside the road, where a driver might have chosen to go around it.
Nobody was cheated. The sequence was simply the business model, and anyone paying attention could read it in advance.
The Same Sequence, Announced
On Wednesday the sequence was announced again, this time in plain words. Alphabet's third quarter results showed search and other revenue up 12 percent to $63.35 billion, below the $64.96 billion analysts expected and roughly flat with the $63.27 billion of the second quarter, after growth of 19 percent in the first quarter and 17 percent in the second. In the same call, Sundar Pichai said query growth was the highest it had been in five years. More questions, flatter revenue: the signature of a road carrying traffic it does not yet charge for.
The company was candid about why. Philipp Schindler, its chief business officer, gave three reasons for the slower growth: currency first, then a normalization in retail and media spending, and then lower monetization in the conversational search Google calls AI Mode. He also described what is traveling on the new road. Many of the new questions sit early in the buying process, the majority are incremental rather than replacements for old searches, and they run about three times longer than a traditional query. Then Pichai supplied the historical parallel himself: "Just like our shift to mobile over a decade ago, we are focused on building the best Search user experience, followed by monetization."
Experience first, monetization second. The owner of the road has told us where we are on the schedule.
And the booth is already framed. The same day, Google said it would place ads in follow-up questions within those conversations in the United States and bring ads in conversational search to ten more countries. Its own advertising documentation, as summarized by Securities.io, fills in the design. Existing search, shopping and fully automated campaigns are eligible automatically. Advertisers cannot target the new placements on their own or opt out of them separately from search, though negative keywords and brand exclusions still apply, and the results are reported in aggregate with the rest of the campaign. The short explanation shown beside some ads is generated by Google from the user's question, the AI's answer and public sources, including the advertiser's website, and advertisers cannot write, edit, approve or opt out of it. Google says its internal data shows sponsored and organic product listings in these conversations drawing similar click rates. It is 2013 again, almost line for line: no separate lane, no separate price, the new traffic folded into the campaigns you already run.
What the Discount Is For
Lower monetization is not generosity; it is the price of a habit. A question three times longer than the old query, asked earlier in a purchase, tells the platform far more about the buyer than three words ever did, and an auction that knows more about a buyer can eventually charge more for her. Today that information is underpriced for two reasons. The platform has not finished building the inventory, and advertisers have not yet learned what a buyer in the middle of a long conversation is worth to them. Both conditions are temporary, and only one of them is under the advertiser's control.
That is the part worth dwelling on, because it decides who benefits from the cheap years. The advantage in an introductory period does not go to whoever buys the most of it. It goes to whoever learns what it is worth, since when the price rises, the advertiser who knows the value of the traffic can decide how much toll to pay, and the advertiser who does not simply pays it. In 2013 the businesses with a clear view of what a mobile customer was worth could set their bids with confidence when phones arrived in every campaign. Everyone else learned the value of a phone click the expensive way, after the price had already been set by the people who had done the work.
There is a second trap, specific to this road. A buyer who asks a long, early question is not ready to buy; she is forming the shortlist. Judged on cost per lead or on the last click, she will always look expensive, and a team that measures her that way will either cut the spend just as it starts to matter or keep paying without ever knowing what it bought. The early question is worth exactly what it contributes to the pipeline a quarter later, and that number has to be built on your side, because the new placements arrive inside campaigns you already run, reported in the same totals, rather than as a line you can buy and read on its own.
Learn the Road While It Is Cheap
The work is not exotic. Read the search terms in your accounts for the long, question-shaped queries that now arrive alongside the short ones, and follow those visitors into the pipeline over a full quarter rather than a session. Decide in writing what an early-stage buyer is worth to you relative to one who is ready to talk, so that the number exists before the platform proposes its own. Keep negative keywords and brand exclusions current, since they are the only controls you have been left. And make the pages those buyers land on answer the questions they were asking, plainly, with prices and proof, because when the words beside your ad are written by a machine that reads your website, the most reliable influence you keep over them is the clarity of what you have published.
Then plan next year's budget as if the discount ends, because it will. The platform has said so in its own earnings call, in the same sentence that compared this moment to mobile.
Every road is cheap while it is being paved, and the owner of this one has told us, with unusual candor, that the paving comes first and the toll comes second. The toll is not a betrayal; it is the business, announced in advance. The only question left to the driver is whether, when the booth opens, she will already know what the trip is worth.
