Click Shift / Campaign Automation · v.0.9 beta

One-to-one impact. At any scale.

Practice areas
ABM
SEO
PPC

Method
Guardrail-driven
Event-instrumented

The best marketers do meticulous, one-to-one work. They research the account. They tune the campaign. They write the brief like the page is for one specific reader. Campaign Automation scales that rigor with bounded autonomy — AI agents acting inside guardrails you set, with every action logged as an event you can measure, audit, and trust.

Live event stream · sample events.clickshift.ca · last 60s
schema: trigger → action → impact ∅ events
§01 — Manifesto

Most "AI marketing" still runs on the approval bottleneck. We replaced it with guardrails.

Most agencies bolt AI onto workflows that still require a human to approve every step. The throughput stays human, the cost goes up, and nobody can quite tell what the AI did.

We built differently. Define the bounds — budget caps, quality gates, content allowlists, staged rollouts — and the agents act inside them. Every action emits an event with its trigger, its target, and its measured impact over a defined window. Guardrails tighten or loosen as evidence accumulates.

The result: the rigor of one-to-one work, applied at the scale only software can reach.

"Approval is a bottleneck. Guardrails are a system. The difference shows up in the throughput, the cost, and the audit trail."

Campaign Automation is the software arm of Click Shift Marketing. Use it with a ClickShift strategist operating alongside you, or stand it up for your own in-house team. Either way, the event log is the deliverable.

— Dave Bussell, founder · ClickShift · Wakefield, Quebec
§02 — Practice areas

Three workflows. One event substrate.

Each practice area has its own playbook, MCP integrations, and guardrails. They share a common event schema — so the proof of value reads the same whether the work is paid search, organic, or account-level.
A / 01

Paid search & social.

PPC · MCP: google-ads, meta-ads

Search-term mining, negative keyword automation, ad-copy variant testing, budget pacing, query intent reclassification — running continuously inside spend caps and conversion guardrails.

  • Negative keyword discovery
  • Ad copy variant testing
  • Budget pacing & shifts
  • Query intent classification
B / 02

Organic & content.

SEO · MCP: search-console, cms

Content brief generation grounded in real query data, internal linking proposals, technical audit running, SERP and intent monitoring. Drafts for editors — not auto-publishes.

  • Content brief generation
  • Internal linking proposals
  • Technical audit automation
  • SERP & intent monitoring
C / 03

Account-based motion.

ABM · MCP: hubspot, salesforce, intent

Account research synthesis, intent signal detection, account-level personalization at the page and outreach level, sales enablement asset generation. The 1-to-1 sales motion, but for 1-to-200 lists.

  • Account research synthesis
  • Intent signal detection
  • Account-level personalization
  • Sales enablement assets
§03 — Product view

What you see when you sign in.

Not another dashboard. An agent control plane — modules on the left, live events in the feed, guardrails always visible. The event log is the product.
clickshift.ca/campaignautomation / workspace / acme-corp
Paid Search / last 7 days
Paid Search · 4 workflows running
autonomy: ● auto
Actions · 7d
1,284
+ 18% vs prior
In-bounds rate
99.4%
+ 0.2pp
Impact · 7d
$8.2k
saved spend
Reversals
3
operator initiated
Live feed · in-product trigger → action → impact
§04 — Modules

Four modules. One guardrail model.

Each module ships with pre-built workflows and MCP connections for its channel. The guardrail layer and event schema are shared — so a rule you write once applies everywhere.
Paid Search & SocialM / 01 · ppc

The continuous optimizer.

Search-term mining, negative keyword automation, bid adjustments, ad copy variant testing, budget pacing — running on a loop, not a weekly review cycle.

Negative mining
RSA expansion
Bid modifiers
Budget pacing
Query intent
Audience split
Organic & ContentM / 02 · seo

The briefing engine.

Content briefs grounded in real SERP and GSC data, internal linking proposals, technical audit queue, SERP change detection. Drafts for editors, not auto-publishes.

Brief generation
Link proposals
Audit queue
SERP watcher
Intent shift
Canonical fix
Account-based motionM / 03 · abm

The account layer.

Account research synthesis, intent signal aggregation across providers, page-level personalization for known target accounts, sales enablement asset drafting on stage transitions.

Research synth
Signal fusion
Page variants
Stakeholder map
Deck drafting
Routing rules
Guardrails & eventsM / 04 · core

The control plane.

The layer that makes the other three safe. Spend caps, quality gates, staged trust, reversibility — plus the event log every action writes to. The audit trail is the deliverable.

Spend ceilings
Quality gates
Staged trust
Reversal paths
Event schema
Weekly report
Native MCP connections · bring your stack
§05 — Configurator

Set the autonomy level. Watch the rules shift.

SuggestReviewAuto · boundedAuto · broad
±5%±20%±40%
24h7d14d30d
Rules the agent operates under
Agent may act automatically on in-bounds workflows without human sign-off.
Any single action may change spend by no more than ±15% per campaign per day.
Impact measured on a 7-day post-action window, confounders flagged.
Every action emits a reversal token — undoable until a human confirms otherwise.
Autonomy expands weekly only where evidence passes the staged-trust check.
At this level, the agent runs the loop. You tune the bounds.
§06 — The schema

If we can't show you the event, we didn't do the work.

A single event is the unit of value in this practice. Here's the shape — simplified, with the fields that matter most for proof and audit.

Three blocks. Every event has them.

Trigger answers why. The signal observed, the threshold crossed, the rule that fired. Without a trigger, an action is just noise.

Action answers what. The exact change, the target, the bounds it stayed inside, the reversal token that lets us undo it. Without an action body, you can't audit.

Impact answers so what. The metric, the window, the delta, and — critically — the confounders we couldn't rule out. Without honest impact reporting, every event is just a press release.

Multiplied across thousands of small actions per month, this becomes the proof layer that lets clients see exactly what AI did, where it worked, and where the guardrails need tuning.

// event.schema · v0.4
// trigger — why this fired
"trigger": {
  "signal": "zero_conv_query",
  "value": 42, // clicks
  "window": "14d",
  "threshold": 25
},

// action — what changed, in bounds
"action": {
  "type": "negative_keyword.add",
  "target": "free trial download",
  "scope": "campaign:cmp_acq_q2",
  "bounds_ok": true,
  "reversal_token": "rv_2k4j8m"
},

// impact — measured, on a window
"impact": {
  "metric": "wasted_spend",
  "delta": -340.18,
  "window": "7d_post",
  "confounders": ["none_detected"]
}
§07 — Access

Three ways to get on it.

Campaign Automation is in private beta. Pick the path that fits — stand up the platform yourself, run it with ClickShift as the operator, or embed it in your own agency delivery.
T / 01
Operator
For in-house marketing teams running their own program.
$2,400/ month · per workspace
  • All four modules
  • Up to 3 ad accounts
  • Up to 25k monthly actions
  • Guardrail configurator
  • Weekly impact report
  • Slack event stream
Request access →
T / 03
Agency OEM
White-label the platform for your own agency's clients.
Custom/ volume-based
  • Multi-workspace tenancy
  • Your brand, your UI
  • Partner success manager
  • Co-branded event reports
  • MCP connector requests
  • Revenue share model
Partner inquiry →

Beta cohort opens quarterly.

We onboard a small group each quarter so every workspace gets the schema review, guardrail design pass, and MCP integration help it needs. Next cohort begins January 2027. Join the list.

Join the beta list →