Instagram Is a Search Engine Now

Open Google tonight and search for what you sell. Not your company name. Search the way a customer searches when they do not know you exist yet. Something like best hard seltzer for camping, or where to drink on a patio in Vancouver.

Look at what comes back.

There is a decent chance one of those results is an Instagram Reel. A thirty second vertical video, shot on a phone, ranking on the first page of Google. Sometimes above a website that cost twenty thousand dollars to build.

That is not a glitch. Those are the rules now, and they changed a year ago.

The change already happened. The memo did not go out.

On July 10, 2025, Instagram stopped keeping the door shut. Public content from professional accounts became available for search engines to index. Photos, videos, carousels, Reels. And it was not just new posts. Eligible content reaches back to January 1, 2020.

Read that again with your own account in mind. The Reel you shot in a parking lot in 2023 and forgot about is on the board.

Software has a name for what Instagram did. A silent behavior change. The interface looks identical, nothing on your screen moved, and the thing underneath now works completely differently. I spent twenty five years in software and usability, and silent changes were always the ones that ate people alive, because nothing prompts you to adapt. You just quietly start losing to whoever noticed.

A year on, most brands have not noticed.

The audience did not wait for permission

Search stopped being one box a long time ago. Sprout Social's 2025 research found 41 percent of Gen Z go to social platforms first when they want an answer, against 32 percent who still start at a traditional search engine. GWI data shows more than half look up a brand on social rather than on search.

For years that was framed as a threat to Google. It was reported like a war. Social versus search, pick a side, place your bets.

That framing is dead. The two surfaces merged. Your Reel and a competitor's blog post are now in the same auction for the same person asking the same question, and the person asking has no idea which platform the answer will come from. They just want the answer.

Virality has a half life. Search traffic has a shelf.

Scott Graffius runs an annual study on how long social posts stay alive, built on an analysis of more than 5.6 million posts. His 2026 numbers put the average Instagram post at 18.27 hours before it has collected half of every interaction it will ever get. That is down from 19.04 hours the year before and 19.75 the year before that. Blog posts in the same study sit at 2.03 years.

Be careful with that comparison, because it measures engagement, not search. Nobody has published a clean study on how long an indexed Reel keeps pulling Google traffic, and I am not going to invent one.

But the direction is not ambiguous. On the algorithm side, your content is measured in hours and the window is getting narrower every single year. On the search side, it is measured in something much closer to years. Indexing does not extend the feed life of a post. It gives the post a second life on a completely different clock.

Your best performing Reel of last March is dead in the feed. It may still be answering a question today.

Three Google launches, and almost everybody is blending them into one

This is where the takes online fall apart, so slow down here.

There have been three separate announcements, and they are routinely treated as a single feature. They are not.

December 2025. Search Console Insights started showing social channel data, but only for people with a verified website that Google had already linked to a social account on its own. Limited, automatic, experimental.

June 2026. Search profiles arrived, giving larger creators and publishers a way to connect their platforms and surface that content inside Search.

July 7, 2026. Platform properties. This is the real one.

Platform properties are a new property type inside Search Console. You add Instagram, TikTok, X or YouTube the same way you have always added a domain, authorize the connection, and Google starts reporting on it. LinkedIn is not in yet.

What you get is the report SEO people have had for twenty years and social people have never had once: clicks, impressions and the actual queries driving traffic, at the post level and the profile level.

Three details worth holding onto.

First, there is no website requirement. Verification happens on the account rather than a domain, so a brand that lives entirely on Instagram can finally see its own Google Search performance.

Second, and this is the part getting lost everywhere, access to the data is not the same thing as authority. Connecting a platform property is a reporting feature and nothing more. It measures. It does not boost. Whether your Reel ranks in the first place is a separate question, and it comes down to whether Google understands who you are. Google now builds that picture out of everywhere your brand appears at once: a website, a YouTube channel, an Instagram account, a Google Business Profile, other people writing about you. A brand with nothing but social is not disqualified, but it is asking Google to trust one witness instead of five. That gets harder the more commercial the query is and the more established the competition is. Your website did not become optional. It became one source among several, and the actual work is making all of them tell the same story about the same company.

Third, the rollout is uneven and Google has published no completion date. On launch day, Barry Schwartz, who has covered Search Console longer than almost anyone alive, could not add a platform property in his own account.

So if someone promises you a live search dashboard for your Instagram by Friday, they are selling you a rollout they do not control.

Views were never the scoreboard. Now there is a better one.

I will keep saying this until it stops being necessary, which appears to be never.

A view count is not a result. It is a number that makes a deck look good. The goal is the scoreboard, and the goal is almost always selling more. Sometimes it is bookings, foot traffic, sign ups, or awareness ahead of a launch. Name that first, then measure against it: conversions, sales, cost to acquire on the money end. Brand lift on the front end, meaning ad recall, awareness, consideration, favorability and purchase intent. Lift is the leading indicator that feeds the sale, not the finish line.

What search adds is a column nobody in social has ever had. Intent.

An engagement rate tells you people liked something. A query tells you what a stranger typed at eleven at night when they had a problem and money to solve it. One of those is a feeling. The other is a shopping list.

One round is not a test

Here is where most content programs quietly fail. Somebody runs a concept once, it does fine, and that becomes The Way We Do It. One round is a single data point wearing a pattern's clothing.

The way I work it is organic first, because organic is cheap and honest. Small tests, no ad budget, real audience. Find what works. Then adjust it and run it again against the winner, and try to beat your own best result. Then again. Only after that does paid budget make sense, because now you are amplifying something with evidence behind it instead of buying reach for a guess.

Search makes this better, not harder. Trends were never really testable. By the time you had data the sound was dead. A query is stable. You can build against it, measure it, rebuild it and measure it again, and the target does not move underneath you.

The reframe the whole thing hangs on

Stop asking what will do numbers this week.

Start asking what somebody will type six months from now.

That single swap changes the brief, the shot list, the caption, the title, and what counts as a win. Your content stops being a broadcast and becomes an answer. Broadcasts expire. Answers accumulate.

What to actually do

None of this is difficult. It is just different from what your team has been doing.

1. Pick the query before the concept. Write the question a real buyer would type, then build the video that answers it. Not the other way around.
2. Say the keyword out loud in the first few seconds. Audio gets transcribed. Transcripts are text. Text is what a crawler reads.
3. Put it on screen too. Your on screen text is doing double duty now, hook and label.
4. Write the caption like an answer, not a caption. First line carries the question and the payoff. Save the personality for line three.
5. Use the name field on your profile, not just the handle. That field is indexable real estate and most brands waste it on a tagline nobody searches.
6. Fill in alt text. It takes nine seconds and almost nobody does it.
7. One question per post. A video that answers four things ranks for none of them.
8. Choose evergreen over trending for this job. Trend content is still worth making. It is just doing a different job on a different clock.
9. Check your searchability setting is actually on. It is a toggle in your privacy settings, and I have found it switched off on accounts whose owners swore it was not.
10. Connect the platform property when it appears in your account. Then wait. Early numbers are for orientation, not for reporting to your boss.

What good actually looks like

An honest assessment before you spend anything, including the part where somebody tells you this is not your problem right now. A real brief and a real kickoff, so there is a reason behind the creative instead of a vibe. Testing that runs more than once. Measurement tied to the thing you sell rather than the thing that flatters the deck.

And a creator who will turn down your product if they would not use it. That is not a moral flex. Authenticity is the performance mechanism. An audience can smell a rented endorsement through a screen, and a search result that gets clicked and immediately abandoned teaches Google that your answer is a bad one.

Questions worth asking anyone you hire

• Which of the three Google launches are you talking about, and what does it actually do?
• What query is this piece of content built to answer?
• How will we know it worked, and which number is the goal rather than the vanity metric?
• How many rounds are we testing before we spend on paid?
• What are you going to tell me if the results are bad?
• Would you use this product if I were not paying you?

If they answer all six without stalling, you are in good hands. If they answer them before you ask, that is us.

Instagram spent a decade training everyone to chase the feed. The feed still pays, and it still pays fast. But the feed forgets you in about eighteen hours, and Google does not.

Build for the person typing the question.