How to Turn Intent Data Into Campaigns, Alerts and Tests

Intent data is useful only when a change in it changes what your team does that week: which accounts see ads, which ones sales calls and what the next email says. A score that sits in a dashboard is a cost. This guide covers how to read weekly intent scores, which signal deserves which action, how to get the accounts into LinkedIn ads within LinkedIn's own list rules, and how to test whether any of it adds pipeline.

We use intent data from Fokals as the worked example throughout, because its documentation states plainly what a score is and what it is built from. Disclosure: Fokals is an EGGKNITE partner.

What does a weekly intent score measure?

Fokals scores buying intent for each company, topic and week on a 0 to 100 scale. Its data dictionary describes 69 topics in nine groups, its intent page names CRM and data warehousing among the topics, and it writes a row only when a company scores at least 5 on a topic. Weeks run Monday to Sunday in UTC.

Three details in that documentation shape how you should use the data:

• Every score carries dated evidence. Each score lists the signals behind it, with when and where each one was seen. Fokals names its sources as company websites and public DNS records, company job boards and careers pages, regulatory filings and company announcements. A salesperson can open the evidence before a call instead of guessing why an account surfaced.
• Scores are written once and never revised. A score from three months ago is what the data said three months ago. That makes an honest backtest possible: you can check which accounts scored high before they became opportunities in your CRM.
• A surge is separate from the level. The surge flag is true in a week when a company's score for a topic rises sharply against its own recent weeks. We found no published threshold for the flag in Fokals' documentation.

Read the level and the surge differently. A high, steady score says a company has been active on a topic for a while. A surge says something changed this week. The surge is the trigger for time-sensitive work such as a sales alert, and the level decides who stays in always-on audiences.

Fokals delivers the scores by API or as CSV and JSON Lines files, so the weekly batch can land in a warehouse or CRM on a schedule rather than in someone's inbox. From there, the same signals can drive the account-level messaging covered in B2B personalization.

Which topics should you track?

Track the topics your product's purchase depends on, plus one or two that you believe come just before it in your market. A reporting tool might watch data warehousing and CRM if its sales team sees CRM projects turn into reporting projects; the backtest below tells you whether that belief holds. Keep the list short. Every topic you add multiplies the accounts that cross your threshold, and sales stops reading alerts that rarely lead anywhere.

Filter by fit before you look at intent at all. A company outside your ideal customer profile with a score of 90 is still outside your ideal customer profile.

Which signal should trigger which action?

How do you get intent accounts into LinkedIn ads?

LinkedIn's company list targeting is the most direct route for B2B paid media, and its rules shape how you build the list:

• A list needs at least 300 rows to upload, and it can hold up to 20 MB or 300,000 companies.
• To match a list to LinkedIn Pages, LinkedIn requires at least one of company name, company website, company email domain, LinkedIn Company Page URL or stock symbol. LinkedIn recommends adding fields such as website, industry and location to improve matching.
• LinkedIn recommends 1,000 or more companies, and the list must match at least 300 member accounts before an active ad set can use it.
• Generating the audience can take up to 48 hours, and on rare occasions longer.

Fokals' company records carry the company name, a domain where a website is on file and, where the company or its parent is listed, a ticker, and all three map to columns LinkedIn accepts. Two consequences follow. A single week of surges can fall short of LinkedIn's 300-row minimum, so build the paid list from recent surges plus steady high scorers, and keep the weekly surge list for sales. And with up to two days of processing, a list uploaded on a Monday may not serve until Wednesday, so treat the paid audience as a weekly tier and leave same-day reaction to sales.

A worked example with assumed numbers

Every number in this section is an assumption for illustration. None is a result or a benchmark.

A B2B software company sells a CRM add-on. Assume its ideal customer profile covers 4,000 companies and it tracks the CRM topic. In one week:

• 120 companies in the profile have a surge on CRM.
• 960 more have scored 60 or above on CRM in each of the last four weeks without a surge. The threshold of 60 is our own starting point, to tune once results come in.

The weekly loop then runs like this:

1. Sales. The 60 surge accounts in the treated group (see step 3) go to sales as alerts with the evidence attached. Even all 120 surge accounts are below LinkedIn's 300-row minimum, so surges alone cannot be the paid list.
2. Paid. The paid list is the 960 steady accounts plus the 60 treated surge accounts, or 1,020 companies, which clears LinkedIn's recommended 1,000. It is uploaded on Monday with a plan for it to serve from Wednesday.
3. Test. Half of the 120 surge accounts, picked at random, are held back from ads and outreach for the test period. Meetings and pipeline from the 60 treated accounts are compared with the 60 held back. Groups of 60 only show large differences, so the comparison is pooled over several weeks before anyone acts on it.
4. Cleanup. An account whose score stays below 60 for three weeks leaves the paid list.

Where does intent data mislead a campaign?

Intent data earns its cost in some situations and backfires in others.

• Crediting intent for deals that were coming anyway. This is the biggest pitfall. Accounts that surge are already moving, so they are likely to convert better whether or not you advertise to them. Without a holdout, intent will look better than it is, which is counterintuitive for a tool meant to find buyers. The holdout is how you measure incrementality, the part of the result the campaign itself caused.
• Hiring signals cut both ways. A careers page that lists a CRM administrator role can mean a new purchase, or a team that already runs a CRM and needs staff. The catch is that the score cannot tell you which; the dated evidence can, so read it before a sales call.
• Large companies light up everywhere. In practice, a big enterprise can show activity on many topics at once, so treat a high score from one with more caution than the same score from a small company. However, the surge flag helps, because it compares each company with its own recent weeks, and filtering by ideal customer profile and by topic reduces the rest of that noise.
• Weekly data meets slower ad platforms. On the other hand, sales can act on a surge the same day, while LinkedIn may take two days to build the audience. The trade-off is to let sales own speed and let paid own reach.
• Reading the score as a probability. It is an easy mistake to make. Fokals defines the score as buying intent on a 0 to 100 scale, and we would not recommend reading 80 as an 80% chance of purchase. Use it to rank accounts against each other, unless your own backtest shows a calibrated link to conversion.

How we run this on engagements

On client engagements we start from the funnel and the ideal customer profile, then decide which intent topics map to which buying stage before a list is built. Accounts are tiered: the surge tier goes to sales with the evidence in the CRM record, and the steady tier feeds LinkedIn and nurture audiences. The holdout is set before launch, so the first report can say whether intent-led accounts produced more meetings and pipeline than the ones held back. Topics and thresholds are reviewed each week against what sales says about the alerts. This sits inside our lifecycle and demand generation work, where account-based programs belong, with the LinkedIn side run as part of paid media. Our account-based work for Smiota, a B2B smart-locker platform, built on audience research, account-based marketing and lead nurture, contributed $56M in pipeline in under six months.

Where should a team start?

Start with a backtest. Take the scores from the weeks before your recent opportunities were created and check whether those accounts stood out. Point-in-time scores, written once and never revised, keep that test honest. If high scores came before real deals, set up the weekly loop above with a holdout from the first week. If they did not, change the topics before you spend on ads.

For the paid side, see how the two main B2B ad platforms compare in Google Ads vs LinkedIn Ads. For what account-based work costs a small team, see the economics of agentic ABM.

Sources

• Fokals, Intent: weekly 0 to 100 scores by topic, surge flag, dated evidence, sources, delivery formats.
• Fokals, Data dictionary: 69 topics in nine groups, rows only for scores of at least 5, Monday to Sunday UTC weeks, signal sources, listing identifiers.
• Fokals, API reference: company records with name, domain and ticker.
• LinkedIn, Company list targeting in Campaign Manager: list size limits, accepted fields, minimum match and processing time.
• LinkedIn, Upload company and contact targeting lists: upload steps in Campaign Manager.