Glossary

The vocabulary the rest of these docs assumes — Adlyse’s own terms first, then the advertising metrics they’re built on.

Adlyse concepts#

ABI score#

Ad Baseline Index — a single 0–100 health grade for an ad account, shown on Ad accounts. Think of it as a credit score for your advertising: higher is better.

• 70–100 — above baseline. Spend is working well; look for chances to scale.
• 40–69 — at baseline. Solid, with room to tighten.
• Below 40 — below baseline. Something needs attention.

It blends three things: how efficiently you turn spend into results, how little of your spend is wasted, and how much room you have to grow. Each score carries a confidence label — low confidence means there wasn’t much recent data, so treat the number as directional. Adlyse recomputes it after each sync.

Agent#

An AI colleague you create and brief yourself: a name, a role, a model, a set of integrations, and standing instructions. Unlike Adlyse AI it’s narrow on purpose, and it can be scheduled or called from a workflow. See Agents.

Angle#

The promise a creative makes — the reason it gives someone to care. Adlyse tags every synced creative with its angle automatically, which is what lets Creative analytics rank promises rather than ad IDs.

Brand profile#

Your positioning, audience, objectives, and constraints, maintained in one place and rendered live into every AI conversation. It replaced the old guidelines/brand.md file. See Brand profile.

Capability gap#

Something Adlyse found but can’t act on — a missing permission, a disconnected account, an unconfigured conversion action. These appear under Needs you on Home with the amount they’re holding back.

Decision queue#

The ranked list of proposals on Home, each with the money at stake and the reasoning attached. It’s the product’s spine: every recommendation, whether from Adlyse AI, a scheduled agent, or Adlyse’s own analysis, arrives here for a human to approve.

Fatigue#

A creative whose click rate is sliding week over week while its cost per conversion climbs — the fingerprint of an ad that’s worn out its welcome. Flagged with a badge in Creatives.

Guideline#

A markdown file in the guidelines/ folder of Files that acts as a standing instruction the assistant always follows. Naming one after a platform (google.md, meta.md) scopes it to conversations where that platform is connected.

Hook#

The opening device of a creative — how it captures attention in the first second, as distinct from the angle it goes on to promise. Ranked in Creative analytics.

Impression share#

The share of available auctions your ads actually appeared in. A drop caused by rank rather than budget means a competitor is outbidding you — see Brand watch.

Proposal card#

A pending recommendation in the decision queue: what to change, the estimated monthly impact, a confidence level, and the reasoning. Nothing is applied until a human approves it, and approving shows a before-and-after preview first.

Safe budget to add#

How much more you could spend per day without your results getting less efficient — headroom you’re leaving on the table because you’re capped by budget rather than by demand. Surfaces as Growth headroom on Home.

Schedule#

A clock pointed at an agent or a workflow. Adds no intelligence of its own; it just means you stop being the trigger. See Schedules.

Skill#

A saved, reusable routine an agent can step into mid-conversation — instructions plus the exact tools the job needs. Lighter than an agent, and not a fixed sequence like a workflow. See Skills.

Swipe file#

Your saved reference ads — competitor ads, your own winners, anything you upload — with a remix button. See Swipe file.

Tool group#

A bundle of related abilities you switch on for an agent — memory, folder access, visualization, artifacts, creative studio, suggestions, data tables. Distinct from integrations, which connect an agent to a source rather than granting an ability. See Tool groups.

Velocity#

How fast new creatives are entering an account. Low velocity means you’ve stopped learning, and performance decays on a lag. Tracked in Creative analytics.

Wasted spend#

Money that isn’t buying results, as a share of spend. How it’s spotted differs by platform: on Google it’s spend going to auctions you’re losing on relevance rather than budget; on Meta it’s campaigns that spent real money with no conversions, or whose cost per result is far above the account’s norm. Platforms without an estimator show no waste column rather than a misleading zero.

Workflow#

A pipeline you draw on a canvas — a trigger, then steps in the order you connected them, including non-AI steps like conditions, API calls, and human approval gates. Choose one when the order of steps is the requirement. See Workflows.

Advertising metrics#

CPA — cost per acquisition#

What you pay for each conversion. Spend ÷ conversions. Lower is better.

CPC — cost per click#

What you pay for each click. Spend ÷ clicks.

CPM — cost per mille#

What you pay per thousand impressions. A proxy for how expensive your audience is to reach.

CTR — click-through rate#

The share of people who click after seeing your ad. Clicks ÷ impressions. A creative-quality signal more than a funnel one.

Conversion#

The action you’re paying for — a purchase, a signup, a lead. What counts as one is configured on the ad platform, which is why a missing conversion action shows up as a capability gap.

Frequency#

How many times, on average, each person has seen an ad over a window. Rising frequency alongside falling CTR is the earliest warning of fatigue.

ROAS — return on ad spend#

Revenue for every unit of currency spent. A ROAS of 3.2 means $3.20 back per $1 spent. Needs revenue tracking to be meaningful — use CPA when you don’t have it.