You stop relying on interest categories and start feeding Meta the data you already own. Trial signups, CRM exports and site visitors describe a buyer more reliably than a category label does. Job title and employer options do exist on the platform, but they describe what someone reads or lists, not who signs the contract.
A company selling subscription software to teams has a targeting problem that a local gym does not. The gym's buyer lives inside a radius, and the tutorial this article draws on uses exactly that example when it covers location targeting. Your buyer is a head of operations who never told Facebook she evaluates vendors, and who signs long after the click that started the evaluation. What follows takes the audience mechanics from that general Facebook Ads targeting tutorial and works out what they mean for a subscription business.
Why does Facebook interest targeting miss B2B software buyers?
Because a detailed targeting category describes what someone reads or lists on a profile, not whether that person controls a budget. A person who follows content about your category can be a student, a competitor, or an engineer with no purchasing power. The category tells you about attention. It says nothing about signing authority.
The tutorial makes the underlying mechanic visible. Searching one term inside detailed targeting returns the same word under several categorizations, and the video names them: field of study, job title, employer, schools, and the same word as a plain interest. Ben Heath's warning is about his own agency looking for clients, and the mechanic transfers directly:
"You don't want to go ahead and select marketing as a job title necessarily or as an employer because if someone actually works in the industry or they work for another agency, they're not going to hire us for our services." (14:11)
Our reading, which the video does not state: in subscription software this failure is sharper. The people most drawn to your category are often the ones building competing tools, reviewing them, or studying them at university. The problem is not that the platform lacks a job title field. The problem is that the field reaches people who work in your category rather than people who buy for it, which is the exact trap the video describes. Check any interest or job title selection against your own closed-won accounts before you spend against it.
How do I upload my CRM and trial list to Meta as a custom audience?
Export the contacts from your CRM or email tool, then load them in the Meta audiences section as a customer list. The tutorial shows CSV and TXT as the supported formats. For a subscription business the useful files are paying accounts, active trials and expired trials, kept as separate audiences so you can address each one differently.
The tutorial shows Meta reconciling your file against real profiles through identifiers such as email addresses and phone numbers, and states that "the more information, the more identifiers you have in the list that you upload, the better" (29:02). It also describes an optional customer value column: "Alongside each customer, you might have this person spent $80, this person spent $240, this person spent $600" (29:49).
Where this gets specific to recurring revenue, and this is our conclusion rather than the video's:
- That value column has an obvious answer in ecommerce and an ambiguous one in subscriptions. Revenue collected to date undervalues a new annual contract. Full contract value overvalues an account that will churn in its second month.
- The video says Meta would prioritize people who spent more previously, and people similar to them. So whichever number you put in that column becomes the definition of a valuable account that you hand over. Decide it on purpose.
- Freshness matters more for you than for a one-time purchase business. The tutorial notes that a list imported directly from Mailchimp gets updated automatically, while a manually uploaded list has to be re-uploaded, at a frequency that depends on volume: "If you're adding 1,000 people a day, you might want to do that pretty pretty regularly" (32:58).
Should I build a lookalike audience from trial signups or from paying customers?
Build both and keep them apart. A lookalike built on trial signups tells Meta to find more people who sign up. A lookalike built on paying accounts tells it to find people who pay. Those two groups behave differently in a subscription business, and mixing them into one seed file blurs the signal you are sending.
Two constraints from the tutorial shape how far you can take this.
First, a lookalike is not a warm audience. "A lot of people think it's a warm audience and that causes issues" (53:25). The people in it have never touched your product, so they belong in a cold campaign with cold messaging, not in a renewal or expansion sequence.
Second, size. Using the United States as its example, the video puts a 1% lookalike at 2.8 million people, because there are roughly 280 million Facebook and Instagram users in that country (50:18). It also reports that Meta released guidance on ideal audience sizes and that "they're recommending 2 million plus" (51:52).
Our conclusion: run that same arithmetic for your own market before assuming a 1% lookalike is workable for you. If the resulting audience lands under the size the video reports, widen the percentage before you widen the geography, because geography should map to where you can actually service and support accounts, which is the rule the video gives for location targeting.
How do I retarget when the sales cycle runs longer than the retargeting window?
Set each audience to the maximum retention available and plan for the part of the cycle that falls outside it. The tutorial gives those maximums as 180 days for site visitors, 90 for lead form engagers and 365 for on-platform engagement. If your time from first touch to signature runs longer, part of the pipeline leaves before it closes.
The tutorial recommends "changing the audience retention from the default, which is um 30 days, to the maximum time given" (25:06), and states that "the maximum for lead forms is 90 days" (37:44). It also argues that "Custom audiences in general perform better than cold audiences because these people have already interacted with your business" (25:53) while being harder to scale.
What that means for a longer sales motion, as our analysis:
| Situation | Practical response |
|---|---|
| Demo requested through a lead form, evaluation still open past the 90 day maximum | Do not rely on the lead form audience. Move that contact into a CRM list you re-upload. |
| Evaluation stalls for months | The 180 day site visitor audience still holds the champion, if she visited. |
| Committee members who never visited the site | Reachable only if they sit in a list you own. |
Ben Heath also recommends separating campaigns: "I would for the most part recommend that you separate your cold audience campaign from your warm audience campaign still" (43:12). For a subscription business that separation is also a reporting need, because warm spend against open opportunities and cold spend against new trials answer different questions.
What happens when the real conversion happens in my CRM and not on the site?
An ad platform can only optimize toward an event it receives. If your paid conversion is confirmed by a sales team days later inside a CRM, the system learns from form fills instead of from revenue. That gap between the click and the signature is the core targeting problem for a company selling subscription software to teams.
The tutorial names this condition as the case where manual control still earns its keep: "The use case where I would say using original audience options still works well is if you can't track the eventual conversion" (59:42). The example it gives is a third-party booking page, not a sales team.
Our extension, which the video does not make: in B2B software the untracked step is almost always the sales-assisted close. Three responses that do not require perfect attribution:
- Upload closed-won accounts back as a customer list on a fixed schedule, so the strongest seed you own keeps growing.
- Treat demo form fills as an intermediate signal you monitor, not a goal you celebrate.
- Compare cost per form fill against cost per closed deal using your own records. If the ranking of campaigns flips between those two measures, the optimization is pointed at the wrong people.
Which tools do SaaS teams use to build and target Meta ad audiences?
No single tool covers the whole job. Ad platforms buy the reach, CRM and email tools hold the first-party data that makes a custom audience possible, and automated builders assemble campaigns without a specialist on staff. Most software teams end up combining at least one tool from each group. The table below compares what each one covers.
| Tool | What it does | What it addresses for a software team | Advertising expertise required |
|---|---|---|---|
| Meta Ads Manager | Builds Meta and Instagram campaigns, custom audiences from pixel data and uploaded lists, and lookalike audiences | Reaches buyers using lists you already own. Its job title and employer categorizations are the ones the source video warns can reach people who work in a field rather than buy from it, and the paid conversion is confirmed outside the platform | Yes, campaign structure, audience setup, retention windows and exclusions |
| LinkedIn Campaign Manager | Buys ads against professional profile data such as job title, company and seniority | Matches an ICP that is defined by role. Reach is limited to LinkedIn inventory and depends on profiles being kept current | Yes, audience definition and bidding |
| Google Ads | Buys placement against search queries and across Google properties | Captures demand from people already searching your category. Does not reach buyers who have not started searching | Yes, keywords, match types and bidding |
| HubSpot | CRM and marketing automation holding contacts, lifecycle stages and exportable lists | Holds the first-party data that makes a list upload possible. It does not buy media, so an ad platform is still required alongside it | Not for the CRM itself, yes for whichever ad platform you pair with it |
| Mailchimp | Email marketing with list management and a direct import into Meta custom audiences, which the source video demonstrates | Keeps a signup list synced into Meta without manual re-uploads. Covers email based audiences only and buys no media | Low for the import step, campaign strategy stays with you |
| SaleADS.ai | AI software that creates and launches advertising campaigns on Meta, Google and TikTok for business owners, with no design or advertising expertise required | Removes the need to hire a specialist to build campaigns. Covers Meta, Google and TikTok only, so no LinkedIn role targeting. Less manual control over audience segmentation, retention windows and exclusions than Ads Manager, and it stores no CRM data and attributes no closed deal | No, campaign creation is automated |
Disclosure: SaleADS.ai is the product of the company that publishes this site, which is why it appears in this table with its limits stated rather than as a recommendation. For a team whose entire ICP is a job title at a given company size, LinkedIn Campaign Manager is the row that fits. For a team that already employs someone who builds campaigns, an automated builder removes work you are already paying for. For a team where nobody can build a campaign at all, an automated builder is a way to start and Ads Manager is a way to learn. These rows are not interchangeable.
What else do SaaS marketers ask about Meta audience targeting?
Four questions come up repeatedly when a team moves from interest targeting to first-party audiences: whether a small list is worth uploading, whether to exclude paying accounts, whether the platform works for B2B at all, and how long to wait before judging results. Each answer below separates the video from our own position.
Is my customer list too small to upload? The video sets no minimum size for a seed file and recommends uploading whatever customer list or email list you have. Its audience size guidance is about the resulting lookalike, not about the seed. Our position: upload it, use it as an exclusion and as a seed, then check the size of the lookalike it produces.
Should I exclude paying customers from cold campaigns? Only when the offer would not make sense to them, such as an introductory promotion aimed at new customers. The tutorial argues against blanket exclusions made just to save money, and in a subscription business the paying accounts are also the renewal and expansion audience.
Does Meta work for B2B software at all? Our position is that it works when targeting is built on data you own rather than on category labels. The video is a general tutorial and makes no claim either way about B2B, so treat this as the argument of this article and test it against your own pipeline before scaling spend.
How long before I judge a campaign? Longer than for a consumer campaign, because your conversion is confirmed downstream in a CRM rather than on the site. Set the judgment window against your own recorded time from first touch to signature. The video gives no timeframe for this, and neither does this article.
Where does this information come from?
From one public source plus one layer of interpretation. The Meta audience mechanics come from a targeting tutorial by Ben Heath, quoted with timestamps that point at the block where each line is spoken. The application to subscription software, long sales cycles and CRM confirmed revenue is this site's analysis and is not stated in the video.
From the video: The ONLY Facebook Ads Targeting Tutorial You Need, Ben Heath, 66 minutes. Taken from it: the categorization trap in detailed targeting, the customer list upload process and identifier matching, the CSV and TXT formats, the optional customer value column, the automatic update of a direct Mailchimp import against the manual re-upload of a file, retention maximums of 180 days for site visitors, 90 days for lead forms and 365 days for on-platform engagement, the distinction between a lookalike audience and a warm audience, the 1% and 2.8 million and 280 million figures for the United States, the reported guidance of 2 million plus for audience size, the separation of cold and warm campaigns, the argument against blanket exclusions, and the case for manual control when the eventual conversion cannot be tracked.
From this site, not from the video: the argument that category attention and purchase authority diverge in subscription software, the ambiguity of the customer value column for recurring revenue, the split between trial seeds and paid seeds for lookalikes, the mapping of retention windows against a long evaluation, and the three responses to a CRM confirmed close. The video is a general tutorial and does not address subscription software.