> ## Documentation Index
> Fetch the complete documentation index at: https://docs.agentweb.pro/llms.txt
> Use this file to discover all available pages before exploring further.
> ## Agent Instructions
> ---
> About this documentation: Emma is the AI marketing agent built by AgentWeb (the product is also marketed as HireEmma; AgentWeb, Emma, and HireEmma refer to the same product). Emma learns a company's brand, ICP, and positioning, then executes marketing work inside that company's own accounts: research, ad campaigns on Google Ads, Meta, and LinkedIn Ads, LinkedIn and email outbound, organic content, CRM lead enrichment and ICP scoring, and landing pages deployed to the customer's domain. Actions that spend money or post publicly require the user's confirmation.
> For what Emma can read, draft, publish, and spend on per channel, see https://docs.agentweb.pro/capabilities. For an overview of the product, see https://docs.agentweb.pro/what-is-emma. For terminology (skill, work plan, scheduled chat, template), see https://docs.agentweb.pro/concepts.
# Paid ads on a founder budget
> When to start paid ads and when not to, what has to exist before you spend, how to choose between Search and Meta on a founder budget, and how to judge campaigns on qualified leads rather than cheap ones.
Paid is the only motion where being wrong costs money every hour it runs. This playbook covers the part the channel guides don't: whether to start at all, what has to exist before you spend, and how to tell a campaign that's working from one that's just cheap.
For the mechanics of a specific platform, see [Google Ads best practices](/playbooks/google-ads-best-practices) and [Integrations → Meta](/integrations/meta). This page is about the decision around them.
## First: are you ready for paid?
The most expensive mistake early-stage teams make is over-investing in paid before organic is proven. Paid does not discover your message — it scales whatever message you give it, including a bad one, efficiently and at volume.
Run this gate honestly before spending a dollar:
| Check | Why it matters |
| ----------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------- |
| **You've closed customers without ads** | If nobody buys when you explain it personally, ads won't fix that — they'll just reach more people who don't buy. |
| **You know which claim resonates** | You need a headline that already worked in a DM or a post, not a hypothesis. |
| **You know which ICP converts** | Targeting is only as good as the segment definition behind it. |
| **You know roughly what a customer is worth** | Without this, "is \$80 per lead good?" is unanswerable. |
| **[Measurement](/playbooks/measurement) is in place** | Lead-source field, UTM convention, conversion event. Retrofitting this later doesn't work. |
| **You can spend consistently for 6–8 weeks** | Ad platforms need conversion volume to optimize. Stop-start budgets never exit the learning phase. |
**If more than one of these is a "no," go back to Stage 1** of the [pipeline system](/playbooks/pipeline-system). Founder-led outbound and organic content answer the message and ICP questions for free. Paid answers them for money, slowly, and less clearly.
## Choosing where to spend
| Channel | What it's good at | Realistic starting budget | Emma's role |
| ------------------------------- | ------------------------------------------------------------------ | -------------------------------------------------------------------------- | ---------------------------------------------------------------------- |
| **Google Search** | Capturing existing intent — people already looking for what you do | Enough to win a few clicks a day on your terms; B2B keywords are expensive | Full build and manage — campaigns, keywords, bids, budgets, reporting |
| **Meta (Facebook + Instagram)** | Creating demand — reaching people who don't know they need you yet | Enough for the platform to gather conversion data, not just impressions | Full build and manage — campaigns, budgets, creative analysis |
| **LinkedIn Ads** | Precise B2B targeting, at the highest CPMs in the market | Highest floor of the three | Full build and manage — campaigns, targeting, bids, budgets, reporting |
For most B2B founders the order is: **Search first** if people are already searching for the category, **Meta first** if they aren't. Search captures demand that exists; Meta creates demand that doesn't. Running both before either one works is how budgets disappear without a lesson.
**LinkedIn Ads gives you the sharpest B2B targeting and charges for it.** Emma can build and run LinkedIn campaigns end to end, but the CPMs are several times Meta's — so it's usually the third channel you add, not the first, unless your deal sizes are large enough that a handful of the right conversations pays for the whole test.
## Before the first campaign goes live
**Google Ads** requires the Pro plan and a *standard* account your Google login can access directly — access inherited through a manager (MCC) hierarchy doesn't count, and manager accounts are hidden by design. **Meta** is easiest via the Partner flow: grant AgentWeb's Partner ID access to your Ad Account, Facebook Page, and Instagram account.
Both fail at campaign creation if the account has no valid payment method.
Ask Emma for a landing page matched to the campaign — not your homepage. She generates it, deploys it live, creates a CRM lead list named after it, enables public form submission, and injects UTM attribution so every lead carries its source.
Connect your Meta ad account before you ask for the page — a page generated with it connected gets the Pixel and server-side conversion tracking wired automatically. Connecting afterwards means that page stays untracked, and adding the Pixel later records visits without reporting conversions. See [analytics and tracking](/playbooks/analytics-and-tracking#meta-conversions-api-capi).
Ad platforms optimize toward the event you give them. "Pageview" buys you people who view pages. Form submission, demo booked, or trial started is what you actually want optimized for.
Emma has vetted launch-gate work plans — look for **Meta Pre-Launch Gate** under **New Chat → Select a workplan**, or ask her *"what launch-gate plans do you have for Meta?"* These check tracking, targeting, creative, and budget structure before spend starts rather than after.
Decide the failure threshold before you're emotionally invested in the campaign. "If cost per qualified lead is above \$X after 30 conversions, it's off" is a decision. "Let's see how it goes" is not.
## Running it with Emma
All of this happens in conversation. Emma explains what she's about to do and waits for confirmation before any action that affects your ads or spend.
**Launch:**
> *"Create a Search campaign for \[product] targeting \[locations] with a \$X/day budget. Use Maximize Conversions. Write three responsive search ad variations against the pricing page."*
**Check:**
> *"How are my campaigns doing this month? Show spend, conversions, and cost per conversion by campaign."*
**Tighten:**
> *"Show me which search terms triggered my ads last week and add negative keywords for anything irrelevant."*
Negative keywords are the highest-return maintenance task in Google Ads and the most commonly skipped. A weekly pass reliably cuts wasted spend.
**Adjust:**
> *"Increase the budget on my best-performing campaign by 20% and pause the worst one."*
## The loop that makes paid worth running
This is the part most paid guidance omits, and it's the whole reason paid sits inside the [pipeline system](/playbooks/pipeline-system) rather than beside it.
```text theme={null}
Campaign --> Landing page --> CRM lead list --> Enrichment --> ICP score
^ |
| v
+-------- cut, scale, or retarget on QUALIFIED leads ----------+
```
Cost per lead is a channel metric. **Cost per qualified lead** — CPL divided by the share scoring above your ICP threshold — is a business metric, and the two routinely move in opposite directions. Broad targeting almost always produces cheaper leads and worse ones.
Ask for it directly:
> *"For leads from `paid-search` last month: how many, average ICP score, how many above 70, and what did I spend? Give me cost per lead and cost per qualified lead."*
If cost per qualified lead is rising while CPL falls, your targeting is broadening in a way the ad platform considers a success and you should not.
## Scaling and cutting
* **Judge on conversions, not clicks or days.** Roughly 30 conversions before drawing a conclusion; a week of clicks at founder-stage volume is noise.
* **Scale in steps, not jumps.** Raising a working budget by around 20% at a time lets the platform re-optimize. Doubling it resets learning and often destroys the performance you were scaling.
* **Change one thing at a time.** New creative *and* new targeting *and* a new budget means you learn nothing from the result.
* **A live ad's text and destination URL can't be edited** in Google Ads — Emma removes the old ad and creates a new one. Plan variations up front rather than iterating in place.
* **Cut on your pre-written threshold.** The rule you wrote before launch is more trustworthy than your judgment three weeks into a campaign you're attached to.
## Known limits
* **LinkedIn campaign type is immutable** once created, and audiences under 300 members won't serve. Emma asks for your ad format up front and checks estimated reach before building — answer carefully, because fixing it means a new campaign.
* **Insights lag a few hours** on both Meta and Google. Data under about four hours old can show as zero — it isn't broken.
* **Performance Max needs real assets** — at minimum several headlines, descriptions, a square image, a landscape image, and a logo. Thin asset groups underperform and the campaign type won't tell you that's why.
* **Never schedule an unattended campaign change.** Scheduled runs don't pause for approval. Ad spend stays on rung 1 of the [autonomy ladder](/playbooks/operating-rhythm#the-autonomy-ladder) permanently — keep spend decisions in an interactive conversation.
* **X and Instagram organic aren't ad channels here.** X has no ads integration; Instagram organic posting is separate from Meta Ads.
## Next steps
* [Google Ads best practices](/playbooks/google-ads-best-practices) — the channel field guide, mapped to Emma's skills.
* [Integrations → Meta](/integrations/meta) — Partner setup, Pixel, and account checklist.
* [Measurement](/playbooks/measurement) — the attribution setup this playbook assumes.
* [Analytics and tracking](/playbooks/analytics-and-tracking) — installing pixels and tags.