Skip to content
Faisal Mahmud TusherDigital marketing, SEO & analytics
with Faisal Tusher

CRM & Marketing Automation › Automation Workflows

Sales Alerts From Marketing Activity

How to design sales alerts from marketing activity: signals worth one, who gets it, real-time or digest, alert fatigue, consent and measuring follow-up.
Guideintermediate10 min readUpdated

Alerts are not scores

Two mechanisms carry marketing activity to sales, and they answer different questions. The lead scoring guide adds fit and intent signals into a score, and when the score crosses a threshold agreed with sales, the lead is handed over. The lead stages and pipeline guide turns that hand-off into a workflow that changes the stage, assigns an owner, creates a task and notifies them, with an escalation rule that reassigns a lead left untouched past the agreed window. Both are about new leads crossing a line.

Alerts are about records that are already across it. Once a contact has an owner, an open deal or a customer account, a score that keeps rising tells the seller little; what they need is to know when one thing happened that changes what they should do today. That is the job of an alert: a single event, on an owned record, with a suggested next action.

Keeping the two apart matters for both. A scoring model that also fires alerts gets tuned to produce alerts, and an alert system built on score changes inherits every quirk of the model. The rule of thumb is that most activity changes the score quietly, and only a short, written list of events interrupts a person.

Signals worth interrupting a seller for

As general practice, an event earns an alert when three things are true: it is rare enough that the seller will read every one, it is specific enough to suggest an action, and acting on it soon is plausibly better than acting on it later. The intent signals guide warns against over-reading: one page view is not intent, interest is not intent, and intent expires. Those warnings apply with more force to alerts, because each one costs a person's attention.

Event on an owned recordWhy it may justify an alertCondition that keeps it rareUsual form
Reply to a marketing emailA person has started a conversationAny reply that is not an automatic responseReal-time, to the owner
Meeting or demo request from an existing contactDeclared intent, the clearest signal there isContact already owned; otherwise it is routingReal-time, to the owner
Return to the pricing or proposal pageEvaluation activity on a live decisionContact on an open deal; first return after a quiet periodDigest, or real-time for late-stage deals
Several contacts from one account engaging in a short periodAn account-level pattern rather than one person's curiosityNamed or target account; agreed definition of engagedDigest to the account owner
Product usage crossing an agreed lineUsage signals for trials or freemium accountsThresholds agreed with salesTask or digest, as the product-qualified leads guide describes
Unsubscribe or complaint from a key contactThe relationship may need attentionContact on an open deal or a customer accountDigest to the owner
Email clickLow on its ownRarely worth an alert; keep it in the scoreScore only

Two of these rows have their own guides. Account-level engagement is part of the joint plan in the account-based marketing guide, which asks for an agreed definition of when an account is engaged enough for outreach and a service level for how fast sales act. Usage signals are routed as the product-qualified leads guide describes, including the caution that a first message which reveals more about someone's activity than they would expect feels intrusive. That caution applies to every alert: the seller acts on the signal, but the message the contact receives should offer help, not recite what they were seen doing.

Who receives an alert, and when

An alert goes to the person who can act on it, which is almost always the record's owner. Use the owner fields the CRM already holds rather than a separate list, so an alert follows the record when ownership changes. Where an account has several people on it, decide by event: a pricing-page return on an open deal goes to the deal owner; a support-related unsubscribe goes to the account manager; usage signals on a customer account go to whoever owns expansion. If the account is already a customer or has an open opportunity, the product-qualified leads guide's rule applies to alerts too: they go to the person who owns that relationship, not to a new seller.

Decide what happens when the owner is absent. An alert to someone on leave or no longer at the company is an alert nobody reads. A deputy or a team queue should receive alerts while the owner is away, and the routing fields that decide lead routing should be the same ones that decide alert recipients, so the two do not drift apart. Recording the owner, fallback and cover for each route in the lead routing rules template gives the alert rules one list to read from. Timing matters as well: an alert that arrives outside working hours waits until morning anyway, so it may as well arrive in the morning digest.

Do not copy managers on every alert. A manager who needs to know whether alerts are acted on needs the report described below, not a second inbox of the same messages.

Real-time alerts, digests and tasks

The form of an alert decides how much attention it takes, and the choice should follow the event, not habit.

Real-time messages are for events where a delay plausibly costs the conversation: a reply, a meeting request, a late-stage buyer returning to the proposal. The inbound lead handling playbook gives demo and pricing requests from new leads the fastest response because intent is highest; whether speed matters as much for events on owned records is something to test rather than assume. Keep real-time alerts few enough that sellers keep notifications switched on.

Digests collect lower-urgency events into one message at a fixed time, usually the start of the working day: pricing-page returns on mid-stage deals, account-level engagement, unsubscribes from key contacts. A digest lets a seller plan the day around the accounts showing activity, and it is far less likely to be ignored than the same events sent one by one.

Tasks are for events that need an action with a due date and a record of whether it was done: a follow-up call after a usage threshold, a check-in after a key contact unsubscribes. A task appears in the seller's normal work queue and can be reported on, which neither a message nor a digest can.

Many teams use all three: a short list of real-time alerts, one morning digest and tasks for anything that must be tracked. Each is an automation and belongs in the inventory the automation governance guide describes, with an owner, a purpose and a measure.

Alert fatigue and how to prevent it

Alert fatigue builds the same way everywhere: each alert is added for a good reason, nobody removes any, and the total grows until sellers filter them into a folder. After that, the important alert is as invisible as the trivial ones. As general practice, five habits keep it in check.

Keep a written list of alert types, each with its event, condition, recipient, form and owner, and treat a new alert as a change that needs a reason. Cap repetition: once a contact has triggered an alert, the same event on the same record should not trigger another for a period the team agrees. Suppress alerts where they do not apply: contacts with no open deal, customers in onboarding, records whose owner has already logged an activity that day. Ask sellers which alerts they act on and which they ignore, at the regular review between sales and marketing, and retire the ignored ones. And review alert volumes per person, because an alert type that is reasonable for one seller can be overwhelming for another who owns many more accounts.

When a deal is won, the alerts should change with it. Prospect alerts stop and customer alerts start, which is one of the jobs of the closed-won workflow alongside the account team notification it sends.

Page-visit alerts depend on tying website activity to a known contact, and that depends on identifiers stored in the browser. The privacy and consent guide explains, from Google's consent mode documentation listed in this page's sources, that where consent is required a tag that sets or reads identifiers must wait for permission, and that a share of visitors do not grant it; declined visitors are not identified. Visits on another device, visits before the person ever submitted a form, and visits by colleagues who never identified themselves may never be tied to the record either.

Three consequences follow. Alerts see part of the activity, not all of it, and should be presented to sellers that way: an alert is a reason to look, not a complete account of what a buyer is doing. The absence of an alert means nothing on its own. And alerts are built only on First-Party Data the business is entitled to use for the purpose, collected under the consent the person gave; this guide does not cover identifying anonymous visitors or companies from their visits, and is not legal advice.

Measuring whether alerts lead anywhere

The tempting measure is alerts sent, or alerts opened. Both are easy to count and both are the wrong target. An alert count used as a goal is Goodhart's law in action, as the Goodhart's law framework describes: once more alerts are the aim, alerts will multiply and their meaning will thin out.

Measure what happens after an alert instead: whether the owner logged an activity on the record, whether a conversation or meeting followed, and whether the deal moved. Compare those outcomes with similar records where no alert fired, and read small numbers with care. Time from alert to first activity is a useful operational measure, in the same spirit as Lead Response Time for new leads, but only as a diagnostic, not a target.

To find out whether speed matters at all, the lead response time test randomises routing speed for new leads and reads meetings held per lead; the same design is the model for testing an alert type, by sending it for some eligible records and not others and comparing what follows. Share the results at the joint review in the sales and marketing alignment guide, where both teams read the same CRM data and agree which alerts to keep.

An office fit-out firm trims its alerts (illustrative)

Illustrative example — not a real result

Tibbalune Office Interiors trims its sales alerts

Tibbalune Office Interiors, an invented office fit-out and refurbishment company, sends its six account managers an alert every time a contact clicks a marketing email. In a typical week that is about 400 alerts, and when asked, the account managers say they have filtered them into a folder they rarely open. The marketing operations lead replaces them with three alert types: a real-time alert when a contact on an open deal replies to any email or requests a site visit, a real-time alert when a contact on a deal at proposal stage returns to the pricing or case-study pages for the first time in two weeks, and a morning digest listing every other open-deal contact who visited those pages the day before. Email clicks stay in the score. Weekly alerts fall to around 25 real-time alerts and one digest per person. Over the next quarter the team tracks meetings booked within a week of an alert against comparable open-deal contacts with no alert: 18 of 70 alerted contacts book a meeting, against 9 of 70 without one. The team reads it as encouraging but too small to settle, keeps the design and runs a randomised comparison the following quarter.

Common mistakes

  • Alerting on every click. Sellers filter the channel and miss the alert that mattered. Keep clicks in the score.
  • Using alerts to replace the score, or the score to fire alerts. Scores hand over new leads; alerts react to single events on owned records.
  • Sending alerts to a fixed list instead of the owner. Alerts go to the wrong person as soon as ownership changes.
  • No cover for absence. Alerts to someone on leave, or who has left, are read by nobody.
  • Real-time for everything. Most events belong in a digest or a task.
  • Presenting alerts as complete. Consent and identifiers mean alerts see only part of the activity.
  • Reciting the signal to the contact. The seller acts on the alert; the message offers help.
  • Counting alerts sent. Measure the conversations and meetings that follow, against records with no alert.

Frequently asked questions

What is the difference between a sales alert and a lead score?

A score accumulates many signals into a number and hands a lead to sales when it crosses an agreed threshold; the lead scoring guide covers it. An alert reacts to one event on a record a seller already owns and asks for a specific action now, such as a reply to a renewal email or a return to the pricing page by someone on an open deal. Most activity should change the score quietly; only a short list of events should interrupt a person.

Should sellers be alerted every time a contact opens an email?

No. The lead scoring guide treats opens as very low or zero weight because mailbox privacy features make them unreliable, and an alert on every open or click is the quickest route to alert fatigue. Keep email engagement in the score and reserve alerts for events that change what the seller should do next, such as a reply, a meeting request or a click on a pricing or proposal link by a contact on an open deal.

Why do alerts miss some of the activity sellers know happened?

Because page-visit signals for a known contact depend on identifiers that, where consent is required, wait for the visitor's permission, as the privacy and consent guide explains from Google's consent mode documentation. Visitors who decline are not identified, and activity on other devices or before a form was submitted may never be tied to the record. Alerts therefore see part of the activity, and should never be presented to sellers as a complete picture of what an account is doing.

Terms used on this page

Frameworks

Where this page is used

Also related

Sources

  1. Consent mode — Google Tag Platform

Related