Threads Ads in 2026: Meta's Newest Ad Surface, Tested Against the Alternatives

The interesting thing about advertising on Threads is not that Meta opened the inventory. It is that the inventory sits on a network whose users behave in almost the opposite way to Instagram users, and most of the early campaign failures came from brands treating the two as one audience reached through two doors.

This guide is written for the person who has a Meta ad account, a working organic presence and a decision to make about where the next slab of budget goes. It covers what the Threads surface actually is in late 2026, why a text-first feed changes which creative works, who should test now and who should wait, and how to build a test that produces an answer rather than a chart of impressions.

What the surface actually is

Threads is a text-first network with an image and video attachment, not an image network with a caption field. The primary unit is a short post, usually under a hundred words, and the primary engagement is a reply. Likes exist and are cheap. Reposts exist and travel. But the behaviour that separates Threads from every other Meta surface is that people argue in public, at length, with strangers, and enjoy it.

Meta opened Threads advertising in stages. The programme started with a small set of advertisers in a handful of markets, widened through 2025, and by late 2026 the surface appears inside Ads Manager alongside Facebook and Instagram for most accounts in supported countries. Availability is still not uniform. Before planning anything, open Ads Manager, build a dummy campaign and confirm that Threads placements are offered for your account, your objective and your billing country. A strategy built on a placement your account cannot buy is a document, not a plan.

The practical consequence of the text-first design is that Threads inventory is cheap relative to Instagram inventory, and it is cheap for an honest reason: there is less of it, the formats are less visually imposing, and the audience is smaller. Cheap reach is not the same as good reach, but a low cost per thousand is a genuine advantage when the thing you are selling can be explained in a sentence.

Why Threads behaves unlike Instagram

Four differences matter more than the rest, and each one changes how a campaign should be built.

The feed is chronological-leaning and conversation-weighted. Instagram rewards a post that stops a scroll for two seconds. Threads rewards a post that starts an argument. A creative built to be admired in silence underperforms; a creative built to be answered does not.

Production cost is low and the audience knows it. A screenshot with a sentence over it can outperform a produced video on this network, and users reward that. Polished brand assets read as advertising here more sharply than they do on Instagram, because the surrounding content is so obviously not advertising.

Links are tolerated. Threads is far more forgiving of an outbound link than Instagram, where a link in the caption is inert and the bio link is the only path. That makes Threads the rare Meta surface where a direct-response campaign can end in a click instead of a profile visit.

Depth of comment is the visible signal. A post with forty replies reads as a live conversation. A post with forty likes and no replies reads as a press release. If your ad format cannot generate replies, you are buying the second thing.

Who should test it now

Five conditions. If a brand meets three or more, a test is worth the budget. Fewer than three, and the money belongs on a surface with more mature measurement.

• The product can be explained in one sentence. Threads has no patience for a three-part value proposition.
• The audience is English-speaking and under forty-five. The network skews that way, and the skew is stronger than Instagram's.
• There is an opinion worth having. Categories with live arguments — media, technology, marketing, politics-adjacent consumer issues, sport, work culture — carry far more organic conversation to sit beside.
• An organic presence already exists. Paid on Threads without organic is a cold start on a network that punishes cold starts. Twenty posts and a few hundred engaged followers is enough to make the paid layer legible.
• The team can reply within a day. If nobody answers the comments an ad generates, the ad has bought a billboard in an empty room.

Who should wait: brands selling on visual desire alone, luxury and jewellery houses whose equity depends on production values, local service businesses whose customers are searching rather than scrolling, and any account whose entire Meta history is Instagram-first. For those, Reels inventory remains the better use of the same budget.

The measurement problem, stated honestly

Threads ads inherit Meta's attribution, which means they inherit its weaknesses. A view-through window will claim credit for purchases that would have happened anyway, the platform-reported return on ad spend will be higher than the incrementally true number, and cross-device journeys will be partly invisible. None of this is unique to Threads, but it is worse here because the surface is new enough that there is no historical baseline to sanity-check a reported figure against.

The workable response is to decide, before launch, which number you will believe. For a lead-generating business that is cost per qualified enquiry, counted in the CRM and not in Ads Manager. For a content business it is engaged sessions and subscriber conversion. For an ecommerce brand it is a blended contribution margin measured in a geo holdout, not the platform's attributed revenue. Pick one, write it down, and refuse to change it after the results arrive.

Two smaller measurement notes. First, Threads campaigns generate a disproportionate share of their conversions on the same day, because the feed is chronological and posts decay quickly; a seven-day click window is generous and a one-day window is honest. Second, reply volume is a leading indicator that no dashboard surfaces clearly. Tag UTM parameters with the placement so that your own analytics can separate Threads traffic from Instagram traffic instead of receiving both as "meta".

A 30-day test that produces a decision

The failure mode in new inventory is a small budget spread across six audiences and nine creatives, which produces nine unreadable results and no decision. Do the opposite.

One objective, two audiences, three creatives. Hold the objective fixed. Build two audiences only: one broad interest or lookalike, one retargeting pool of people who have already touched the brand. Then produce three creatives that differ in kind, not in colour — one text-led post, one screenshot or document, one short vertical video under fifteen seconds.

Run for twenty-one days minimum. Threads volume is thinner than Instagram volume, so statistical noise takes longer to clear. Seven-day tests on this surface are a coin toss dressed as data.

Budget for significance, not for coverage. A useful floor is enough daily spend to accumulate at least fifty conversions or a hundred landing-page sessions per cell. If the budget cannot reach that, cut a cell rather than shrinking all of them.

Read three numbers at the end. Cost per the outcome you chose, reply rate per thousand impressions, and the share of clicks that came from the text-led creative. The third tells you whether the network wants words or pictures from your brand, which is the single most transferable finding.

Creative that survives a text network

Write the post, then add the asset. The reverse order is why most Threads ads look like Instagram ads with the picture removed. Start with a sentence a person would actually say out loud, then attach the least intrusive visual that supports it.

Lead with a specific claim or a specific number. Vague enthusiasm dies on this network. "We cut onboarding from eleven days to two" is a post. "Transform your workflow" is an ad, and everyone can tell.

Accept disagreement in the copy. A post that invites a counter-argument earns replies. Replies are the distribution unit. Copy that is closed and complete earns a like and stops there.

Keep the call to action conversational. "Reply and I will send the template" outperforms a button on this surface, because it produces a reply, and a reply is the thing the feed measures. Where a link is the goal, put it in the post and say plainly what is behind it.

Do not reuse Reels assets unedited. A nine-by-sixteen video with burned-in captions and a music bed is legible on Threads and reads as imported. Re-cut for a smaller frame and drop the music.

Five ways to waste the budget

Buying reach to a dead profile. The ad sends a stranger to an account with nine posts and a link in the bio. The click is wasted before the landing page loads.

Running the Instagram creative library untouched. Same assets, same audiences, same objective, and then comparing the two surfaces as though the creative were neutral. It is not.

Ignoring the comment thread. On a conversation-weighted network, an unanswered ad is worse than no ad: it shows a brand that started a discussion and left.

Judging a new surface on a week. The first week is the algorithm learning who to show the ad to. A verdict taken then measures the learning phase, not the campaign.

Using bought engagement to fake a conversation. Replies are the metric this surface actually reads, which makes them the most tempting thing to manufacture and the most conspicuous. A post with two hundred replies and no reply text longer than four words reads as exactly what it is.

Threads ads next to the alternatives

Versus Instagram Reels ads. Reels buys scale and visual persuasion at a higher cost per thousand. Threads buys cheap attention and conversation at lower scale. If the product needs to be seen, Reels. If it needs to be discussed, Threads.

Versus X ads. Similar audience profile, different temperament. X remains better for news-adjacent and B2B technology audiences; Threads is stronger for consumer and creator-economy categories and carries less brand-safety noise for mainstream advertisers.

Versus organic Threads. Paid amplifies what already works. Our guide to growing on Threads covers the organic side, and the honest sequencing is organic first, paid second, because the paid layer converts a conversation that organic started.

Where bought engagement fits is narrower than most sellers imply. Follower and like packages change how an account looks at the moment a paid click lands, which is a real if modest effect on conversion. They do not change what the algorithm does with the ad, and no purchase can substitute for an offer that is worth replying to. Our guide to social proof sets out the boundary between the two.

Getting started, in order

Confirm placement availability in Ads Manager for the objective you intend to run. Post organically for two weeks and note which of your posts earned replies rather than likes. Build three creatives around that pattern. Choose the single number you will judge the test by and write down the threshold that would make you scale it. Then run for twenty-one days without touching the campaign.

At the end, the decision is one of three: scale, iterate the creative and rerun, or stop. A test that ends with "it is too early to say" was not a test. On a surface this young the useful thing you are buying is knowledge, and knowledge only arrives if the question was specific before the money was spent.