If you built a LinkedIn campaign in the last year, you were probably offered Accelerate: a campaign type where you handed over a URL and LinkedIn assembled the budget, the schedule, the audience, the ads and the placements for you. It was one clearly labeled decision. Say yes to the AI, or say no.
That decision no longer exists. LinkedIn's own help center puts it plainly: Accelerate campaigns as a standalone type are no longer available as of August 4th, 2026. Campaigns already running keep serving, but from January 5th, 2027 you will not be able to extend their run dates or add budget to them.
Here is the part that matters more than the retirement notice. In LinkedIn's words, Accelerate's "core creation capabilities are fully available in Classic campaigns." Nothing was switched off. The automation was unbundled and distributed through the ordinary campaign build, where you now meet it one screen at a time, in four different places, with nothing telling you these pieces used to be the same product.
The Campaign Type Is Gone. Nothing Else Is.
A product retirement is usually a small thing for an advertiser. This one is not, because of what replaced it.
When Accelerate existed, the AI came with a label on it. You knew you were choosing automation, you knew roughly what you were giving up, and you could reason about it in one go. Now the same capabilities are options inside a normal campaign: one in the audience step, one next to it in the same step, one in the ads step, one in bidding. Each looks like a small convenience on its own. Together they are the thing you used to be able to accept or refuse deliberately.
There is nothing sinister in this. It is how platforms usually fold a bet that did not land back into the main product. But the practical effect for a small advertiser is that you can now end up running most of Accelerate without ever having decided to.
Two of Them Are Already On
Start here, because this is the part that costs money while you are reading about the rest.
Audience Expansion. LinkedIn's documentation is direct about the default: "Audience expansion is automatically enabled on ad sets where it's available." You do not switch it on. You switch it off, or you do not, and if you have never looked at that checkbox then it has been on in every ad set you have built.
The LinkedIn Audience Network. This one takes your ads off LinkedIn entirely and delivers them across third-party apps and sites, using the same targeting, bid and budget as the rest of your ad set. It is available for single image, carousel, document and video ads. You can apply publisher-level category exclusions, which are supplied by a third-party partner rather than by LinkedIn itself.
Neither of these is a scandal, and both have legitimate uses. But a business paying LinkedIn prices is usually paying for one specific thing: reaching an exactly defined professional audience on the platform where their job title is a matter of public record. Both defaults spend some of that money somewhere looser. If you have never audited them, do that before you read the next section.
Auto-Targeting Builds the Audience From Your Website
This is where the biggest piece of Accelerate ended up.
Auto-Targeting asks you to supply a company URL in the "What you're marketing" step, and then builds the audience for you. According to LinkedIn's documentation, it combines four things: your product name plus AI-generated insights drawn from that URL, your own targeting settings such as location, profile language and exclusions, any optional signals you attach (Lead Gen Forms, contact or company lists, conversions, retargeting audiences), and LinkedIn advertising data combined with the historical performance of your account's previous ad sets.
That last input is worth pausing on, because it decides who Auto-Targeting is actually for. A brand new account has no history to learn from, so it is working from your website and LinkedIn's general model of who buys things like yours. An account with two years of conversion data behind it is a genuinely different proposition. The same feature is much smarter for an established advertiser than for a new one, and LinkedIn does not warn you about that anywhere in the interface.
The control you keep is real. You can review what it generated, click Edit to adjust the targeting settings, and if you would rather do it yourself you can switch back to building your own audience before you save the ad set as a draft. What you cannot do is see the reasoning. You get an audience, not an explanation.
Our own position on LinkedIn targeting has not changed: on this platform, targeting is the lever, not the bid. Handing the lever to a system that has not seen your last twelve months of closed deals is a different act from handing it to one that has.
Expansion Widens What You Built. That Is Not the Same Thing.
These two get confused constantly, including by people selling LinkedIn services, so it is worth being exact.
Auto-Targeting builds an audience from nothing. You give it a URL, it produces a target.
Audience Expansion widens an audience you already defined. LinkedIn's own example: if your ad set targets people with the skill "Online Advertising," expansion might add people who list "B2B Marketing" instead. Your exclusions are still respected, but your inclusions get loosened.
One is a starting point, the other is a modifier, and you can run both at once without realizing that the audience you carefully specified was first replaced and then widened. LinkedIn positions expansion as an upper-funnel tool for building awareness with a broader group, which is a fair description. It is also the wrong tool if you are paying premium prices precisely because you want a narrow group.
The sizing question underneath both of these is one we have written about at length: an audience of 300,000 and an audience of 300 are both wrong, and expansion pushes you toward the first of those by default.
Predictive Audiences Are the One Worth Switching On
Not all of this is a tax. One piece of LinkedIn's AI is genuinely worth an owner's attention, and it is the one nobody enables by accident because you have to go and build it.
Predictive audiences take data you already own and use LinkedIn's model to generate an audience predicted to behave like it. The source can be a Lead Gen Form, a contact or company list you upload, your conversion data through the Insight Tag or the Conversions API, or a retargeting audience. LinkedIn frames it as a bottom-funnel tool, which is the opposite end from expansion.
The distinction that matters: expansion guesses from attributes, predictive audiences learn from outcomes. One says "these people have similar skills listed." The other says "these people resemble the ones who actually converted for you."
This is also the official replacement for lookalike audiences, which LinkedIn discontinued on February 29th, 2024. If you have a dormant lookalike sitting in your account, it stopped refreshing years ago and is serving static data.
The catch is the obvious one and it is not really a catch: it is only as good as the data you feed it. Which is the same argument we made about teaching LinkedIn which leads are worth having. If the only outcome your account has ever reported is "someone filled in the form," then a predictive audience will faithfully find you more people who fill in forms.
Maximum Delivery Is an Instruction, Not a Discount
The fourth piece sits in bidding, and it was the default long before Accelerate existed.
Maximum delivery uses machine learning to bid on your behalf, aiming to produce as many results as your total budget allows. It is available for every objective except the Dynamic ad formats. LinkedIn recommends it for understanding what your whole budget can produce, and recommends cost cap instead when you need to control what each result costs.
One documented behavior deserves more attention than it gets, because it describes a lot of small B2B accounts exactly: if your ad set pairs a large budget with a small audience, LinkedIn may show your ad to the same person repeatedly, producing higher frequency and lower reach. That is not a malfunction. It is what "spend this budget" means when there are not many people to spend it on.
We have covered the three bidding strategies and why the default is the priciest in full elsewhere, so the only thing to add here is the framing: maximum delivery is a fifth piece of automation, sitting in the same build, making a decision on your behalf, and it arrived before the other four.
LinkedIn Will Draft the Ad. It Also Warns You About Slop.
The creative half of Accelerate survives as Draft with AI, which will generate introductory text, a headline and images for a single image ad.
What is genuinely interesting is the guidance LinkedIn published alongside it. Its best-practices page tells advertisers that content which "appears generic, repetitive, or lacking in substance may be considered and selected as 'AI slop' by our members."
That is the platform offering you a generation tool and warning you in the same breath that its members will recognize and reject the output if you ship it unedited. The recommendation attached to it is to treat Draft with AI as a starting point, review and edit everything, and use the brand kit so the tool has your logo, colors, fonts and a description of your voice to work from.
This is consistent with what we see: the generated draft is a reasonable way past a blank page and a poor way to finish. The part of an ad that works is the specific claim only your business can make, and that is the exact part a model trained on everyone else's ads will smooth away.
What to Accept on a Small Budget
If you are running LinkedIn on a few thousand a month, here is the position we would take on a client account.
Turn off Audience Expansion. You are paying LinkedIn's premium for precision. Paying it and then loosening the definition is the one combination that makes no sense. If you want broader reach, buy it somewhere cheaper.
Turn off the Audience Network, at least until the on-platform version works. Judge it as a separate experiment with its own numbers, not as a default riding along inside a campaign you are trying to read.
Use Auto-Targeting only if your account has history. With a year of conversion data behind it, it is a reasonable first draft you then edit. On a new account it is guessing from your homepage, and you know your buyer better than your homepage does.
Build a predictive audience as soon as you have real outcomes to build it from. This is the piece with genuine upside, and it gets better exactly as your conversion tracking gets better.
Leave bidding on maximum delivery while you are learning what a result costs, then move to cost cap once you know. Watch frequency if your audience is small.
Use Draft with AI to start, never to finish.
The theme underneath all six: automation is worth accepting where it has your data to learn from, and worth refusing where it is only smoothing you toward the average advertiser. Accelerate made that one decision. Now it is six, and nobody is going to ask you to make them.
The Switches Are Not Going to Announce Themselves
The retirement of a campaign type is a footnote. The reason it is worth twenty minutes of your time is that the replacement is quieter: an interface where the automated path is the path of least resistance, two pieces of it are on before you arrive, and none of them are labeled as parts of the same system.
Open your ad set setup this week and look at four places: the audience step, the expansion checkbox next to it, the placements, and the bidding section. You are looking for what has been decided for you. You do not have to switch any of it off. You do have to know it is there.
We run LinkedIn accounts for small businesses every day, and the work is rarely exotic. It is mostly this: knowing which defaults are earning their place and which are quietly spending your money on reach you did not ask for. If you would rather have someone watching the settings while you run the business, that is what we do. If you would rather do it yourself, the list above is where to start.
