Close the Loop: Send Lead Outcomes Back to Your Ads
A lead feedback loop sends outcomes like surveys booked back to Google and Meta, so ads optimise for lead quality, not form fills. The steps, and the limits.
Ad platforms optimise for whatever you tell them is a conversion. If that's a form fill, they find more people who fill in forms, not more people who book surveys. A lead feedback loop changes the instruction: it sends lead outcomes (a survey booked, a call back, a poor lead) back to the ads, so spend moves towards the audiences that become jobs.
That sounds like a reporting tweak. In practice it changes what the campaign is trying to do.
Why does optimising for form fills go wrong?
Automated campaigns on Google and Meta look for more of whatever you count as a success. Count form fills, and every form fill looks equally good, whether it came from a homeowner ready to book or someone who tapped through out of curiosity and won't answer the phone.
Here is an invented example for a solar installer in Warrington, with round numbers:
| Example only | Audience A | Audience B |
|---|---|---|
| Ad spend | £2,000 | £2,000 |
| Form fills | 100 | 100 |
| Surveys booked | 5 | 20 |
| Cost per survey booked | £400 | £100 |
To an ad platform counting form fills, these audiences look identical: same spend, same number of conversions. To the installer, one is worth far more than the other. Without lead quality feedback, the platform has no reason to prefer audience B, and nothing stops it putting more budget behind audience A if A's form fills start coming in a little cheaper.
Which outcomes should you send back?
To optimise ads for lead quality, send back the outcomes your team already records, provided they're recorded the same way every time. A short, fixed list beats a long, flexible one. If one rep writes “booked”, another writes “survey Tues” and a third leaves the field blank, nothing useful can go back to the ads.
| Status | What it means | What the ads should learn |
|---|---|---|
| Survey booked | Spoke to the homeowner and agreed a survey or appointment | The strongest early signal: find more people like this |
| Job won | The homeowner went ahead | The best signal of all, but it arrives weeks later |
| Call back | Spoke, interested, wants another time | Real interest, not yet a decision: a weaker positive |
| Not interested | Reached, but didn't want to go ahead | Not a conversion: no reason to find more people like this |
| Poor lead | Didn't fit on the call: the wrong job, or not the homeowner | Spend here was wasted: steer away |
| Invalid number | The number didn't connect | Never a conversion |
Leads still waiting for a first call have no outcome yet, so they send nothing back.
Qualification answers count too. A homeowner who answered the questions on WhatsApp and chose a time for a call has told you more, within minutes, than a form fill does. That signal arrives long before any survey, which matters for reasons covered under the limits below.
The order of priority depends on the business, but the rule doesn't change: tell the platforms which outcomes you want more of, and stop counting the ones you don't.
How do outcomes get back to Google and Meta?
Google supports this directly, through offline conversion import. Its guidance on managing offline conversions describes importing conversions that happen away from the website, such as a sale made over the phone, so you can see which ads led to them. Google recommends enhanced conversions for leads for this, which uses the click ID plus data the person provided.
Meta offers a similar route for lead ads that uses CRM outcomes.
The detail of setting either one up belongs with whoever runs your ad accounts. The decisions that matter to an installer come first: which statuses count, who sets them, and how quickly they're updated after each call.
What does the loop look like, step by step?
Six steps, each one feeding the next.
- Reach: campaigns on Meta and Google put the offer in front of homeowners in the postcodes you cover.
- Personalise: each click lands on a page made for the audience or search behind it. That also makes results easier to read, because each page's leads trace back to one ad. Our guide to personalised landing pages covers how to build them.
- Follow up: the enquiry is answered on WhatsApp within seconds, or by SMS, in the installer's name. The homeowner's answers, and whether they agreed a call time, are the first outcome the loop sees.
- Deliver: the lead reaches the installer's team with every answer attached and a timeline of what happened, and when.
- Outcome: a rep calls and sets a status from the fixed list.
- Back to the ads: the outcomes return to the ad platforms, so spend can shift towards the audiences, searches and pages that become jobs. The loop then starts again with better information.
What are the limits?
A feedback loop is better than optimising for form fills, but it is slower and messier than six neat steps suggest.
1. Outcomes take days or weeks
A survey might be booked the same afternoon or a fortnight later. A job might be won weeks after that. The ads are learning from last month's leads while spending today's budget. Earlier signals, such as answers on WhatsApp and a call time agreed, help bridge the gap, but they're a stand-in for a job, not a job.
2. Small volumes are noisy
If one audience produced ten leads this week and two booked surveys, and another produced ten leads and one booked survey, that isn't evidence the first audience is better. Give the platforms, and yourself, enough outcomes before moving budget on them.
3. A status is only as good as the rep who sets it
If “not interested” sometimes means “I couldn't get through”, the ads learn the wrong lesson. If reps book surveys with anyone to hit a target, the ads learn to find people who agree to surveys and then cancel them. The call itself is better evidence than the status alone, which is why it helps to judge leads from what was said on the call.
None of these limits is a reason to go back to counting form fills. They're reasons to keep the status list short, check statuses against calls, and read results over weeks rather than days. The number that pays wages is cost per job, not cost per lead, and the loop is how a campaign moves towards it.
Where POCKLA fits
POCKLA runs this loop for installers. It runs its own Meta and Google campaigns, sends each click to a page made for that audience or search, and follows up every enquiry on WhatsApp or SMS with AI in your business name. Each lead is exclusive to you and arrives in your CRM, or in POCKLA Desk if you don't have one, with every answer and a timeline attached.
Outcomes are logged in Desk or read from your CRM: booked, call back, not interested, invalid number. Calls made from Desk get free call analytics, so a status can be checked against what was actually said. Those outcomes go back to the ads, so they can learn which audiences turn into jobs. You pay for leads, not clicks. The homepage shows the loop end to end.
Want your lead outcomes to steer the ads? See how the POCKLA loop works →