Google rejects a 31-character headline at upload. Not truncates. Rejects. So the character count is the first thing worth knowing about generated search ad copy, and Aiter writes to it: 30 characters for headlines, 90 for descriptions, with the counts visible on screen.
Four headline strategies, and when each one wins
The headlines section has four tabs. They are not four wordings of the same idea, they are four different arguments.
Balanced mixes universal, need-targeted and objection-targeted headlines in one set. It is the right first choice, and the right choice for anyone who does not want to think about this. Start here.
Universal produces headlines that work across the widest set of queries. This is the one for volume: setting up twelve campaigns for twelve product lines, or an agency onboarding a client at speed.
Need-targeted speaks to specific pain points. It performs best when your audience knows they have a problem and is searching for a solution to it, which is most non-brand commercial search.
Objection-targeted attacks the reasons people do not buy: cost, effort, risk, switching pain. Underused, and often the highest-yield tab if you have run review research, because reviews hand you the objections in the customers' own words.
A practical approach: generate Balanced first, then generate Objection-targeted, and build your responsive search ad from the best of both. Google wants fifteen headlines and rewards genuine variety more than it rewards fifteen versions of one idea.
Descriptions
Descriptions are the short supporting lines under the headline, each taking a different angle on the product. They generate on demand rather than automatically, so nothing is spent until you ask.
Ninety characters is less room than it looks. The ones that work state one concrete thing rather than three vague ones.
Keywords, in three tiers
The keyword section splits into three, and the labels are doing real work:
Broad brings a lot of traffic, heavy competition, and usually a painful cost per click. Useful for awareness with a budget behind it.
Middle is good traffic at a lower cost with less competition. Described in the app as optimal for small and medium businesses, which is accurate: this is where most sensible spend goes.
Long-tail is low competition, low traffic, cheap. Individually small, collectively where a lot of efficient accounts quietly make their money.
If you have a modest budget, start entirely in the middle and long-tail tiers. Broad keywords are how small accounts spend a month's budget in nine days learning something they could have read here.
Getting better output
Run competitor analysis first. Search ads are a contrast medium. Knowing what your rivals claim is what lets a headline say something they cannot.
Run review research if you can. Objection-targeted headlines built on real complaints outperform ones built on guesses about complaints.
Raise the creativity level for the second batch. Conservative produces safe headlines, which in a crowded auction is a synonym for invisible. Generate one round at Standard and one at Spicy, then pick across both.
What to check before uploading
Read every headline against your landing page. Google cares about the match between ad and destination, and so does the person clicking. A brilliant headline promising something your page does not deliver is an expensive way to buy a bounce.
Then check the claims. Anything superlative, anything with a number, anything a regulator would frown at.
Frequently asked questions
Are the character counts enforced?
Google's are hard limits and generated copy is written to them, with counts shown so you can see the headroom. If you edit copy by hand afterwards, keep an eye on the count.
Can I generate for a specific product rather than the whole site?
Yes, and it usually works better. Create a campaign from that product's page rather than the homepage, and everything downstream is about that product.
How many times should I regenerate?
Two or three batches gives you plenty to build a strong responsive search ad from. Past that you are usually rearranging the same facts, and the better lever is upstream in research or the brief.