Your Buyer Journey Flipped: Instagram, TikTok and YouTube Are the Whole Funnel Now

Think about the last time you tried a new restaurant.

You saw the food first. Somebody filmed it, the cheese pulled the way cheese is supposed to pull, and you thought about it for the rest of the afternoon. You did not read a description of the food. Reading about food is not the same as watching somebody eat it, and your brain knows the difference.

So you searched the name on TikTok, because you wanted to see other people eat there too. Then you asked your phone whether it was actually any good, and something answered you with a name and a sentence. Then you Googled the place for the hours and the parking, glanced at the site for about three seconds, and booked.

Four surfaces. Under five minutes. You never once decided which platform to use.

Now the part that should keep a brand owner up at night. In those five minutes, you were sold to at every single step, and none of it looked like an ad.

The funnel you were taught is upside down

The old model was a triangle standing on its point. Wide attention at the top, friction all the way down, a thin trickle converting at the bottom. You bought reach, you paid rent to sit on somebody else's list of ten blue links, and you hoped enough strangers clicked that the math worked.

That model ran on volume. Get in front of everybody, lose most of them, keep the survivors.

Watch what your buyer just did instead. They did not get shown ten options. They got shown food, then proof, then one recommendation. My partners at Plate Lunch Collective map this shift better than anyone I have read, and their framing of the buyer journey before and after AI is the clearest version of it: attention used to be wide and conversion narrow. Now attention is narrow and conversion is wide.

Narrow attention means fewer chances. There is no page two of an AI answer. There is no scrolling past the first recommendation to give you a fair look.

Wide conversion means the ones who do arrive are already sold. The traffic that comes from a recommendation converts at four to five times the rate of the traffic you used to buy, because a trusted source already did the vetting and everything after it is just confirmation.

Fewer at bats. Far better odds on each one. That is the whole game now, and it is the opposite shape of every funnel diagram hanging in every marketing department in this country.

Your feed is not awareness anymore

This is where most brands are still stuck. Social gets filed under awareness. Top of funnel. The fun stuff you do before the real selling starts.

That filing is wrong, and it is expensive.

Gen Z searches Instagram, TikTok and Google at almost exactly the same rate, 67%, 62% and 61%. Not one platform beating the others. All three, all the time. And YouTube, TikTok and Instagram together now account for over 60% of product discovery.

Your feed is not the thing that happens before the funnel. It is the funnel. It is where the craving starts, where the proof gets checked, where the how to gets answered, and increasingly where the decision gets made before your website is ever loaded.

A gap sits inside that behaviour and nobody talks about it. Roughly two thirds of Gen Z have used TikTok to search. Only about a quarter said it actually answered them well.

Two thirds asking. A quarter satisfied. That is not a reason to leave the platform. That is an open lane with a line of people standing in it, and the data on whether creator content closes that gap for Canadian brands is not subtle.

Your buyer is not the mark anymore

The other thing that changed, and this one is uncomfortable for a lot of agencies: people got good at this.

Your buyer has watched ten thousand ads pretending not to be ads. They can spot a paid post from the first frame, they know what a script written by a brand sounds like coming out of a creator's mouth, and they clock a stock photo faster than you can say brand guidelines. The tricks that worked in 2015 do not just fail now. They actively cost you, because getting caught trying is worse than not showing up. It is the same reason the cheap synthetic version quietly loses sales even when it looks fine in the export.

Two things follow from that.

Aesthetic matters, more than it ever has. Not polish, exactly. Taste. When somebody lands on your grid after hearing your name, they make a judgment in about a second and a half, and they are judging whether you seem like you know what you are doing. A sloppy feed reads as a sloppy product. This is not vanity. It is the only impression most of your buyers will ever form of you.

Trust cannot be manufactured in post. 92% of consumers trust a recommendation from a person over an ad, and creator style content is trusted by 69% of consumers against 52% for brand produced video. That gap is the whole business. It is also why I turn down products I would not use. Not because I am precious about it, but because the moment I fake one, every video I make afterward is worth less to every brand I work with.

The machines are reading your feed too

Now the part almost nobody has caught up to.

When somebody asks an AI for the best place to eat in Windsor, or the best canned cocktail for a patio, or which brand of anything is worth the money, that system is not making it up. It is pulling from what exists about you and deciding whether you are the answer.

It reads your site. It reads what other people say about you. And it reads your social, which is why video is now part of being the answer rather than a separate department. Instagram opening its doors to Google in July 2025 was a much bigger deal than the coverage suggested. Your public posts, your captions, your Reels and your alt text became indexable. I wrote about that in detail when Instagram became a search engine, and a year on, most brands still have not checked what of theirs is ranking.

Notice the order of operations there, because it runs backwards from what most people assume. Your website gets read by the machine before a human ever lands on it. Often instead of a human landing on it. The page gets crawled, weighed and boiled down into one sentence for somebody who never clicks, and that sentence is either your pitch or your obituary. A page can do the most important work it will do all year and show up in your analytics as a quiet Tuesday.

Which brings up the thing that quietly excludes more brands than anything else: the machine has to understand what you sell and who it is for.

If your captions are poetry, if your bio is a vibe, if nothing anywhere on your account plainly states what the product is, who it is for and where somebody can get it, then you are not a candidate for the answer. You are not being rejected. You are not being considered. There is a difference, and the second one is worse, because it does not show up in any dashboard you currently look at.

Your competitor with the boring, clear, consistently stated commercial intent gets named. You get skipped. Both of you spent the money.

Regulated categories feel this worst. Alcohol and cannabis brands already fight platform limits on what they can say, which leaves the machines with the thinnest possible picture of them. If you sell either one, the rules on what you can actually say about alcohol in Canada shape your visibility as much as any content strategy does.

Write the site for both, or it works for neither

Everything above applies to your captions. It applies just as hard to every page you own, and this is where brands tend to pick one side and lose.

Write purely for the machine and you get a spec sheet. Every heading a keyword, every claim tagged, technically flawless and completely dead. A human arrives after hearing your name, reads two lines, and leaves, because the page sounds like it was written by a filing cabinet.

Write purely for the vibe and the machine cannot use you at all. Clever headlines with no nouns in them. A homepage where the actual word for the thing you sell never appears once. Beautiful, unquotable, invisible.

The version that works does both jobs in the same sentence. Say the plain thing in plain words, then let the personality live in how you say it. Each section should answer one real question completely enough that somebody could lift it out and quote it on its own, because that is exactly what a machine does with it. And it should sound like a person wrote it, because that is what the human needs when they land there to confirm what they were told.

None of which removes the plumbing, and this is where a lot of genuinely good writing still loses. Plain language is what gets you understood. The setup underneath is what gets you read at all.

Your meta title and description are the first thing a machine sees and often the only thing it shows a person, and a startling number of sites still have theirs blank or auto filled from a page name. Your headings have to describe the section underneath them instead of being decorative. Images need alt text. Pages need structured data that plainly declares what kind of thing they are, a business, a product, a menu, an article, an event. URLs should be readable by a human being. Nobody puts any of that in a brand deck, and all of it decides whether a well written page gets filed or skipped.

Write it for the person. Wire it for the machine. Skip either half and you have wasted the other.

Those two goals fight far less than people assume. Clear writing was always the right answer. The machines just made the penalty for skipping it immediate, and gave it a checklist.

If that last stretch made you realise nobody has looked at yours since the site went live, that is a conversation worth having before you spend another dollar on content pointing at it.

You cannot be your own only source

Everything up to this point has been about what you control. Now the limit, and it is the part brands hate most.

You can say whatever you like on your own website. That is what a website is. The machine knows this too, which is why what you say about yourself is treated as a claim rather than a fact. A claim with one source, and that source is you.

AI does not weigh confidence. It weighs consensus. It wants to see the same thing said about you in places you do not own.

Take a restaurant. Your site says you have the best shawarma in Windsor. So does every other site. Nothing about that sentence tells a machine anything, because it has read ten thousand identical sentences written by people with an obvious interest in the outcome. Now put a creator on camera eating it and saying it, get a local food blog to write the place up, collect a few hundred people saying the same thing in reviews and comments, and something changes. It stops being a claim. It becomes a pattern, agreed on by sources with no reason to agree.

That is authority. Not a score in a tool. Just the unglamorous fact that other people said it and the machine noticed.

Which reframes what you are buying when you hire a creator. You are not only renting an audience for a weekend. You are producing a source. A video of a real person eating your food, in your room, saying your name out loud is doing two things at once: it is an ad a human chooses to watch, and it is a citation a machine can use. One asset, both audiences, and the second job keeps working long after the reach is gone.

It is also the part you cannot shortcut, which is the whole reason it is worth anything. Bought mentions read as bought, to people and increasingly to machines. If it could be faked at scale it would already be worthless.

It still has to be fun

Everything above could be read as an argument for turning your feed into a spec sheet. Do not do that. You would win the machine and lose the human, and the human is the one with the wallet.

You get about three seconds before a thumb decides. Ads built on humour are 6.1 times more efficient at driving market share than the safe and flat alternative, and most brands spent two decades backing away from exactly that. Nobody forwards an interruption. People send each other things that made them laugh, all day long, usually with a caption like this is so you.

And the bridge between those two worlds is a number I keep coming back to: people who see a product inside content are 39% more likely to go search for it afterward than people who saw it in a commercial.

Read that as a funnel and the whole thing clicks together. The fun video makes somebody want the thing. Wanting the thing makes them search. The search hits Google, or an AI, or your own grid. And whether you are there when they land is decided by work you did months earlier.

Entertaining and legible are not in competition, they are consecutive. Entertaining is what sends somebody looking. Legible is why the search ends on your name instead of the guy down the street.

The scoreboard is not views

Views are vanity, including the ones I post about my own channel.

Name the goal first. More cases moved, more covers on a Tuesday, more bookings, awareness before a launch. Then measure both ends: hold rate and reach on the front, brand search and site traffic in the middle, orders and cost to acquire at the end. Brand lift is the leading indicator, not the finish line.

Under the new shape, one metric deserves a line of its own: branded search. When people start typing your name instead of your category, the recommendation engine is working. That is the signal that you got named somewhere you cannot see, and most brands are not tracking it at all.

One round is not a test

None of this is guesswork you do once.

Post it organically, cheap, no budget. Watch what holds, what gets saved, what people ask in the comments. Then change one thing and run it again, trying to beat your own winner. Then again. Paid money goes behind the version the market already voted for, not the version that sounded good in a meeting.

One round is not testing. It is a single data point in a lab coat. I managed teams for years before I ever managed a content calendar, and it was true there too: whoever ran the test once and declared victory left the best version on the table.

What to actually do

• Say what you sell in plain words, in your bio and in your captions. The product, who it is for, where to get it. Boring beats invisible.
• Answer real questions on camera. Take the ten things buyers actually ask and make one video each. Question spoken in the first two seconds, answer in the video, keywords in the caption.
• Read your homepage out loud. If it never plainly names what you sell, a machine cannot quote you and a human cannot place you.
• Check your meta titles and descriptions. Every page, written on purpose. This is a boring afternoon that pays for years.
• Treat the grid like a storefront. Somebody will land on it after hearing your name. Decide what they see in the first second and a half.
• Count who talks about you that is not you. Creators, local press, blogs, reviews, forums. If the honest answer is nobody, that is your first project, not your last.
• Ask an AI about your category and see if you come up. If a competitor gets named and you do not, that is a content gap, not bad luck.
• Keep the funny. Legibility gets you eligible. Entertainment gets you chosen.
• Track branded search. It is the closest thing you have to a receipt for the recommendation you cannot see happening.

Red flags

• A strategy deck that still calls social top of funnel.
• Captions written as poetry with no statement of what the thing is.
• Meta titles and descriptions left blank, duplicated, or auto generated across the whole site.
• Content that only says facts, with nothing anyone would watch voluntarily.
• Reporting built on followers and views with no line for sales or brand search.
• Anybody promising to guarantee you become the AI answer.
• Every source describing your brand is a source you wrote.
• Nobody on the team has asked an AI about your category this month.

What good looks like

An honest read before you pay anybody, including the part where the answer is that you do not need me yet. A brief and a kickoff so the creative exists for a reason a buyer would recognise. Testing that runs more than once. Measurement against the goal you named before you started. And a creator who says no to products, because the belief is the mechanism, not the marketing. If you are still deciding who makes the work, the tradeoffs between an agency, a marketplace and hiring direct are worth reading first.

Questions to ask anyone making your content

• Is this built for discovery, for proof, or for the moment somebody is deciding?
• Does anything we publish plainly say what we sell and who it is for?
• If someone asks an AI about our category next month, why would our name come up?
• Would a stranger watch this on purpose?
• What is the goal, and what does failure look like?
• How many rounds before paid budget goes behind it?
• Does this create a source somebody else owns, or only more content we own?
• Are we tracking branded search, or just views?

If they need to go check before answering any of those, you are not buying strategy. You are buying footage with a colour grade. There is a longer version of this diligence in how I vet a creator portfolio before anyone pays.

I answer them before you pay me.

Let's talk

Your Instagram, your TikTok, your YouTube. All of it is doing two jobs now. It makes a human want the thing, and it leaves behind the evidence a machine needs before it will say your name out loud. Get one right and you are half a business. Get neither and the answer comes back with somebody else's name in it, in front of a buyer who was ready to spend, who will never know you were an option.