Intent data without alert fatigue: practices for sales reps

Intent data produces alert fatigue when every movement of a score is pushed to a sales rep. This guide sets out the practices that prevent it, for the revenue operations and marketing operations teams who decide what reps see: a threshold on the score, a trigger on change, a cap for each rep each week and the dated evidence with every alert. A worked example then counts the alerts each rule sends, and a query repeats the count on your target accounts before anything goes out.

What practices keep intent data from causing alert fatigue for sales reps?

Four practices keep intent data from causing alert fatigue for sales reps. Set a threshold on the score. Alert on a change, once, and not on a level. Cap the alerts each rep receives in a week. Show the dated evidence with every alert. Accounts below the line go to marketing as an audience, without an alert.

The US Agency for Healthcare Research and Quality describes alert fatigue in health care as busy workers becoming desensitised to safety alerts and, as a result, ignoring them or failing to respond appropriately. Its account has two causes: the sheer number of alerts, and that the vast majority of those generated by health care technologies are clinically inconsequential. The result it describes is that clinicians ignore the critical alerts together with the meaningless ones.

A rep's queue can fail in the same two ways: too many alerts, and too few worth opening. Each control answers one of them.

We found no recommended weekly number of alerts for a rep in the vendor documentation read for this guide, so the cap of five is a starting point derived from a rep's time: at an assumed twenty minutes each, five alerts are 100 minutes a week.

Intent products document levers of the same kind. Bombora's guidance on score and topic thresholds treats a business as spiking on a topic at a Company Surge score of 60 or more, and recommends a topic threshold of at least 25% of the topics in a report. 6sense's documentation of its Sales Intelligence alerts says the alerts are sent daily, that one alert can include up to a maximum of 40 accounts, and that each user chooses which account data points appear in it.

Setting the four controls on weekly intent scores

The intent dataset carries what each control reads. Intent Scores holds one row for a company, a topic and a closed week: a score from 0 to 100, a surge flag, the number of signals behind the score and the dated evidence behind it. Company Signals holds every dated signal and refreshes daily.

A threshold on the score

A company can carry rows on many topics in a week. That is the right list for marketing and far too long for a rep. Set the line for an alert at a score of 50 to start, beside the surge flag, and read the bands below it as audiences. Choose the topics before the line: of the 69 topics in 9 groups, alert only on the few that describe what your customers were doing before they bought, as the guide to prioritising accounts with intent scores shows.

The surge flag as the trigger

A level is a poor trigger. The evidence that lifts a score over a line this week can hold it there for several weeks, and a rule on the level sends the same account again. The surge flag is a rule on change: it is set when the score has risen sharply against the company's own normal for the topic. It compares a company with itself, so a large company that is always active on a topic is not flagged for its size.

Add one rule of your own: keep a log of the alerts you send, and alert an account again only after four weeks without one.

A cap for each rep each week

The first two controls set the average week, and the cap handles the week when one territory lights up. Rank each rep's candidates by fit tier and then by score, send the first five and hold the rest. A held account stays a candidate while its flag stands. Fit comes first: build the tiers from your ideal customer profile and keep them fixed between weeks.

The dated evidence with every alert

An alert that says only that a score rose asks the rep to trust it. Each row of Intent Scores carries the dated evidence behind the score, so the alert can say what happened at the account and when. Put the evidence in the alert itself and show the score as a detail. Scores are written once, after the week closes, and are never revised, so an alert opened a month later still shows what it showed on the day.

A worked example: the alert volume each threshold produces

The example is illustrative: Acme Robotics is an invented company and every figure is an assumption.

Assumptions:

• Acme Robotics has 2,400 target accounts, each matched to a company ID and shared evenly among 12 reps, 200 a rep.
• Every target account is of strong or possible fit, so the fit gate removes none.
• It alerts on three topics, and each account counts once, by its highest score among the three.
• In the week of the example the accounts fall into the bands of the first table.
• 48 accounts carry the surge flag on at least one of the three topics.
• Of those 48, 36 have had no alert in the previous four weeks.

Each rule sends the accounts that meet it. Divide by 12 for the average rep.

The working: 480 + 216 + 108 + 36 = 840 scored accounts, 216 + 108 + 36 = 360 at 30 or more, and 108 + 36 = 144 at 50 or more. From the first rule to the last the week falls from 70 alerts a rep to 3, and all 840 scored accounts stay in marketing's audience.

The average hides the uneven week. Suppose the 36 fall on the 12 reps as 9, 6, 4, 4, 3, 3, 3, 2, 2, 0, 0 and 0. With a cap of five, the first rep receives 5 and holds 4, the second receives 5 and holds 1, and the other ten are untouched: 31 alerts go out and 5 wait.

To run the count on your own accounts, join your account table to Intent Scores (company_intent_weekly) for the latest week. Below, accounts is your own table with an account_id, a company_id, an owner and a fit_tier. The query takes each account's highest score among your topics and counts, for each owner, the accounts over each line and those with surge set.

-- accounts: your table, one row per target account,
-- with account_id, company_id, owner and fit_tier (1 strong, 2 possible, 3 weak)
with latest as (
  select max(week_start) as week_start from company_intent_weekly
),
best as (
  select a.account_id, a.owner,
         max(s.score) as top_score,
         max(case when s.surge then 1 else 0 end) as surging
  from accounts a
  join company_intent_weekly s on s.company_id = a.company_id
  join latest l on l.week_start = s.week_start
  where s.topic in ('crm', 'sales_engagement', 'marketing_automation')
  group by a.account_id, a.owner
)
select owner,
       count(*) as scored,
       sum(case when top_score >= 30 then 1 else 0 end) as at_30,
       sum(case when top_score >= 50 then 1 else 0 end) as at_50,
       sum(surging) as in_surge
from best
group by owner
order by in_surge desc;

Read the result by owner, because the cap is set for the busiest rep's week. Then build the alert list, where alert_log is your own record of alerts sent, keyed by account and by the week_start the alert was for.

-- alert_log: your record of alerts sent, one row per account_id and week_start
with latest as (
  select max(week_start) as week_start from company_intent_weekly
),
candidates as (
  select a.account_id, a.owner, a.fit_tier,
         s.topic, s.score, s.signals, s.evidence,
         row_number() over (
           partition by a.account_id order by s.score desc
         ) as topic_rank
  from accounts a
  join company_intent_weekly s on s.company_id = a.company_id
  join latest l on l.week_start = s.week_start
  where s.surge
    and s.topic in ('crm', 'sales_engagement', 'marketing_automation')
    and a.fit_tier <= 2
    and not exists (
      select 1 from alert_log g
      where g.account_id = a.account_id
        and g.week_start >= l.week_start - interval '28 days'
    )
),
ranked as (
  select c.*,
         row_number() over (
           partition by c.owner order by c.fit_tier, c.score desc
         ) as place
  from candidates c
  where c.topic_rank = 1
)
select owner, account_id, topic, score, signals, evidence
from ranked
where place <= 5
order by owner, place;

The query keeps surging accounts of strong or possible fit with no alert in the four weeks before, takes each account's surging topic with the highest score, ranks each owner's candidates by fit_tier and score and keeps five. Send the evidence column with each row.

How marketing and sales share intent signals without flooding reps with alerts

Marketing and sales share intent signals without flooding reps with alerts by reading one table through two routes. Marketing takes every scored account as an audience, with no cap. Sales receives only the accounts that cross the agreed line, as capped alerts with their evidence. Both teams see the same scores and agree one line between them.

An audience can take breadth: in the example a campaign runs to 840 accounts as readily as to 144. A rep's week cannot, since each alert asks for research and a tailored first touch.

The alert is pushed and everything else is there to pull. Put the latest score, the surge flag and the evidence on the account record, where a rep preparing a call finds them. Between weekly scores, new rows of Company Signals for an account already alerted join the open alert as fresh evidence and raise no second one.

Account engagement, intent signals and lead scoring: marketing and sales alignment on alerts

Account engagement, intent signals and lead scoring measure different things, and marketing and sales alignment comes from giving each a job. Engagement and lead scores record what people at an account do with you. Intent signals record what the company does elsewhere. A sales alert fires on an agreed combination, never on one score alone.

In Salesforce's Marketing Cloud Account Engagement a prospect's score measures implicit interest by giving points to activities such as opening an email or submitting a form, and a letter grade from F to A+ records how well the prospect fits your ideal customer profile. HubSpot's lead scoring tool builds engagement scores, fit scores and combined scores, and labels a combined score from A1 to C3, the letter for fit and the number for engagement.

Demandbase measures engagement at the level of the account. Its glossary defines Engagement Points as tracking the amount of engagement from target accounts, and its standard journey stages place an account in the Marketing Qualified Account stage at a Pipeline Predict Score of 85 percent or more, or at 100 or more Marketing Engagement Points in the past 3 months.

Three combinations decide the alert.

• Surge and engaged. Alert first: the account is moving and already knows you.
• Surge, not engaged. Alert as a prompt for outbound, with the evidence as the reason for the call.
• Engaged, no surge. Leave it to lead scoring and its own threshold.

Keep the measures in separate fields, since a single summed number hides which part moved. The guide to adding company signals to lead scoring covers the join, the weights and the double counts to avoid.

The marketing to sales handoff: B2B practices for intent signals, lead scoring and sales alerts

At the marketing to sales handoff, B2B teams keep alerts readable with four practices: a written lead scoring threshold for a sales-ready lead, a written rule for the intent signals that raise an account alert, one alert for each crossing, and feedback from sales that moves both lines. Marketing owns the thresholds and sales owns the feedback.

Salesforce's training on a lead qualification model gives, as rules of thumb, a score threshold of 100 points and a grade threshold of B- to start from, each to be adjusted up or down on feedback. HubSpot's guide to building lead scores walks through a handoff in which a workflow enrols a contact when its engagement score reaches the High threshold, assigns a rep and sends that rep a notification, with re-enrolment turned off so that a contact is routed to sales once.

An intent alert needs the same agreement. A checklist for the two teams:

1. The line. The topics, the score floor, the surge flag and the fit tiers, written down and dated.
2. Once for each crossing. One alert for an account in four weeks, enforced from the alert log.
3. The cap. A number for each rep each week, and the rule for held accounts.
4. The response. What a rep owes an alert: a first touch or a recorded reason within five working days.
5. The disposition. Three outcomes a rep can record: worked, wrong fit, wrong time. Wrong fit moves the tiers, and wrong time moves the topics or the floor.
6. The review. Each quarter both teams read the alerts sent, the share worked and the meetings booked, by topic, and move one setting at a time. Treat a falling share worked on steady volume as an early sign of fatigue: lower the cap first.
7. The record. The score, the flag and the evidence stored with every alert, so your record shows what raised it.

How Fokals delivers it

Intent Scores are written weekly and Company Signals daily, and both are delivered by REST API, as bulk files or through MCP. Each weekly row is written once and never revised. The score is company-level, so an alert chooses the account and your own records choose the person. The data dictionary defines every field used above.

Frequently asked questions

The queries and code on this page are examples to adapt. Test them in your own environment before you rely on them.

What this page says about the products it names was checked against their public documentation on 7 October 2026. Product and company names are trademarks of their owners. Fokals is not affiliated with them or endorsed by them.