For bottom-of-funnel Search, I do not start by asking AI for 15 headlines. I give it the keyword theme, landing page, offer, proof, and constraints – then make every visible position do one job.

My default RSA structure is:

I usually pin these roles to their positions. Where I have several strong variations, I pin two or three assets to the same position so Google can still rotate within the role.

That is a deliberate trade-off. Google generally encourages more asset flexibility because pinning reduces eligible combinations. This framework prioritizes message order for tightly themed, commercial-intent ad groups. It is not a universal rule for every Search campaign.

Ad copy’s job is to sell the click, not the product. The landing page earns the scroll and makes the full case.

That makes message match the central lever:

The searcher sees continuity instead of a bait-and-switch. Google sees a closer relationship between the query, ad, and destination. Relevance and trust improve together.

A responsive search ad can contain:

More slots do not automatically mean a better ad. If the offer supports six strong headlines, six are better than 15 padded variations. The goal is useful coverage, not a completed progress bar.

Use the searcher’s language. For BOFU queries, clever wordplay usually loses to immediate recognition.

I pin this role to HEADLINE_1. The line does not need to repeat every keyword. It needs to represent the ad group’s dominant buying intent accurately.

The second position answers: why this result instead of the other ads on the page? Pair the differentiator with evidence whenever possible.

Only use proof the business can substantiate. “Best,” “leading,” and invented customer counts are not differentiation.

I pin two or three approved variations to HEADLINE_2 when the account has enough real proof to test.

The third position should make the next step feel clear and low-risk.

Match the CTA to the landing page. Do not promise a free trial when the page offers only a sales call.

I pin approved CTA or risk-reversal variations to HEADLINE_3.

Without pinning, an exact-match line, two CTAs, or two proof claims can appear together in an awkward order. Pinning lets the RSA vary the words without changing the argument.

The cost is fewer combinations. That is acceptable when sequence matters more than maximum machine freedom. If you do not need the sequence – or the ad group mixes broader intent – leave more assets unpinned.

Explain the outcome and validate it with a real number, customer signal, or concrete capability.

Example: “Send unlimited emails from $9/month. Trusted by 10,000+ growing teams.”

Address the barrier most likely to stop a BOFU click.

Example: “Switch in days with free migration. Cancel anytime.”

As with headlines, you can pin more than one variation to DESCRIPTION_1 or DESCRIPTION_2. Each description should still stand on its own because Google may show one or two descriptions.

Do: “Project Management Software.”

Avoid: “Supercharge Team Productivity.”

A BOFU searcher is scanning for confirmation, not decoding a campaign concept.

Do: “CRM for Real Estate Teams.”

Avoid: “Customer Management Solution.”

Specific language makes the intended buyer obvious.

Do: “Rated #1 by 10,000+ Users” – if true and supportable.

Avoid: “The Best CRM Software.”

The AI may be excellent at plausible-sounding proof. Plausible is not permission.

Do: “From $49/User/Month.”

Avoid: “Affordable Pricing.”

A concrete price, timeline, guarantee, or migration promise reduces uncertainty.

I default to Title Case for headlines because it scans cleanly in Search results. This is a house style, not a Google requirement. Clarity comes first.

Start with the live keyword theme, current ads, landing page, and offer. If the ad group contains several unrelated intents, fix the structure first. One RSA cannot answer three different buying jobs coherently.

Provide only claims the landing page or business can support: pricing, trial terms, customer counts, ratings, security, migration, guarantees, implementation time, or a real differentiator.

Ask the AI to flag missing evidence instead of inventing it.

Review combinations within each pinned position. Look for duplicated meaning, unsupported proof, a CTA that conflicts with the page, or a line that depends on context it may not receive.

HireOtto supports 3–15 headlines, 2–4 descriptions, a final URL, two optional display paths, and headline or description pinning. New RSAs are created paused, giving you a review checkpoint before traffic reaches them.

Confirm the account, campaign, ad group, ad ID, status, headlines, descriptions, and pins. Then review the final URL, display paths, policy details, and Ad Strength in Google Ads before enabling.

Treat Ad Strength as creative feedback, not the strategy. Deliberate pinning may work against its flexibility recommendations. Performance still needs to be judged with impressions, clicks, conversions, conversion value, CPA, ROAS, and incremental volume as appropriate.

The framework above is the strategy. HireOtto handles the account work around it from the same AI conversation, so approved copy does not have to be manually rebuilt in Google Ads.

HireOtto does not decide search intent, invent proof, choose the offer, or determine whether message control is worth the pinning trade-off. Those are practitioner decisions.

The advantage is continuity: one AI workflow can move from live account context to approved execution, while the final Ad Strength, URL, display-path, and policy review stays in Google Ads.

Get started here: https://docs.hireotto.com/setup/connect-ai-tool

Assume a tightly themed ad group targets “email marketing software.” MailPilot is a fictional example, and its price and proof are illustrative—not claims to copy.

Position mapping:

The original visual uses Drip as an illustrative example; its proof, price, and offer should not be treated as current product claims. The structure is the point: confirm the category, establish a reason to trust the product, and remove friction without trying to explain the entire platform in 30 characters.

The RSA sells the click, but assets can answer the next BOFU questions directly on the results page.

Use only claims and destinations that are live. HireOtto can also create, link, reuse, unlink, and report on extension assets.

Before editing, list the current RSA and preserve the complete headline, description, and pin set. In HireOtto, updating headlines or descriptions replaces the complete corresponding set. It is not a one-line append operation.

Do not edit a live RSA solely to chase Ad Strength. Change it because you have a message hypothesis, new evidence, or a genuine coverage gap.

No. Google allows 3–15, but this method starts with roles, not slot count. Add another headline only when it gives a position a genuinely useful variation.

No. This is my BOFU framework for tightly themed ad groups where the order of the argument matters. Broader campaigns may benefit from more unpinned flexibility.

Yes. Pinning two or three variations to the same position preserves the role while allowing rotation within it.

It can. Pinning restricts combinations, while Ad Strength rewards flexibility and asset variety. Use pinning because you need message control – not because the score approves of it.

Yes. It can create RSAs with headline and description pinning and update existing asset sets. New RSAs are paused. For updates, provide the complete desired headline or description set and preserve the pins you want to keep.

The current ad listing returns status, headlines, descriptions, and pins. Check Ad Strength, final URLs, display paths, and detailed policy information in Google Ads during the final review.

AI is useful here because it can turn approved inputs into position-specific options, enforce character limits, surface repetition, and reconstruct complete update sets.

The practitioner still owns search intent, truthful proof, the offer, the pinning trade-off, and the launch. That division of work is the point: accelerate the review loop without automating judgment out of the process.