SEM advertising campaign: how to plan, split budget and read one across search engines

An SEM advertising campaign is paid placement against a search query, bought on an auction, and run across more than one engine. The plan has five parts: the query set you want to appear for, the account structure that holds those queries in groups a bidding system can learn from, the budget and its split between engines, the ads and landing pages, and the measurement that tells you which of those to change. Build it in that order. Most campaigns that disappoint were structured before anybody decided what a conversion was worth.

Step one: define the query set and what it is worth

Start from demand you can see. Export Search Console queries to find what already brings you clicks, use a keyword planner for volume and competition, and read the search terms report of any existing account. Then sort every query into three buckets:

• Buying now. Product or service plus a buying modifier: price, near me, buy, quote, hire, best. These justify the highest bids.
• Comparing. Versus, alternatives, reviews, your competitor names. Cheaper clicks, longer path, worth bidding with a clear comparison landing page.
• Learning. How to, what is, definitions. Usually better served by organic pages than by ad spend, with exceptions where your product is the answer.

Then put a number on a conversion. Average order value times gross margin for retail, or lead to close rate times average contract value times margin for services. That number sets your maximum cost per acquisition, and your maximum cost per click follows from it and your conversion rate. Without it, every bid decision is a guess.

Step two: the account structure

• One campaign per thing you want to control separately. Separate budget, separate geography, separate target. Not one campaign per keyword.
• Group by landing page and intent. If two queries deserve the same page and the same message, they belong together.
• Enough volume per group for the bidding to learn. Automated bidding needs conversion data. Fragmenting a small account into forty tiny ad groups starves all of them. Consolidate until each group sees meaningful weekly conversions.
• Brand kept separate from everything else. Brand queries convert at rates that flatter any campaign they sit inside and hide the performance of the rest.
• A negative keyword list from day one. Free, jobs, salary, course, DIY, and competitor terms you do not want. Review the search terms report weekly and add to it.

Step three: budget, and the split across engines

The two mainstream search ad platforms are Google Ads and Microsoft Advertising, the latter serving Bing, Yahoo and partner placements. Running both is normal practice for one practical reason: the Microsoft network usually has less competition, which often means a lower cost per click for the same query, and its audience skews towards desktop and work contexts in many markets. The way to find out what it is worth to you is to import your Google campaigns into Microsoft Advertising, which the platform supports directly, give it a defined test budget and compare cost per acquisition on your own data.

Budget decision	How to set it
Total monthly budget	Target acquisitions multiplied by your maximum cost per acquisition
Split by intent	Majority to buying now queries, a defined slice to comparing, a small slice to testing
Split by engine	Start with a test allocation on the secondary engine, then set it from measured cost per acquisition
Brand spend	Treat as defence, cap it, and review whether organic already covers the query
Test reserve	A fixed percentage held back for new queries, new ad formats and new landing pages

Keep daily budgets high enough that your main campaigns are not budget limited by mid afternoon. A campaign that runs out of money every day is being optimised by the clock rather than by performance.

Step four: ads and landing pages

• Write enough headline and description variations for the ad system to test, each one a distinct claim rather than a reword. Pin only where a legal or brand requirement demands it.
• Put the query phrase in at least one headline. Relevance lifts quality and lowers what you pay.
• Add every extension that applies: sitelinks, callouts, structured snippets, call, location, price, lead forms where they fit your process.
• Send traffic to a page that matches the ad promise exactly. A click on a specific product that lands on a category page is the most common and most expensive leak in SEM.
• Make the landing page fast and put the conversion action above the fold on mobile. Paid traffic is the least patient traffic you buy.
• Test one element at a time on the page: headline, form length, proof, price presentation.

Step five: the five reports to read

• Search terms. What people actually typed. This is where you find both waste and new winners. Weekly.
• Cost per acquisition by campaign and by engine. Against your maximum, from your own system where possible.
• Impression share and the reasons you lost it. Separates a budget problem from a competitiveness problem.
• Conversion lag. How long between click and conversion. Judging a week before the lag has played out causes most premature changes.
• Quality and relevance diagnostics. Where expected click through rate, ad relevance or landing page experience is below average, you have a specific fix rather than a bid problem.

Review weekly, change one thing at a time in campaigns using automated bidding, and let each change sit for a full conversion cycle before judging it.

This page covers planning an SEM campaign across engines. For a step by step build inside Google Ads specifically, see how to build a Google SEM campaign. For day to day account management, match types, negatives and bidding, see PPC management done properly. For how the two definitions of SEM differ, see SEO and SEM compared, and for running both sides as one programme, see organic and paid search.

Where people get stuck on SEM advertising campaign

What is a sensible minimum budget for an SEM campaign?

Enough to buy a statistically useful number of clicks in your niche within a month. Work it backwards: your target conversions, divided by your expected conversion rate, multiplied by your cost per click. If the answer is more than you have, narrow the geography or the query set rather than spreading the budget thin.

Should I run Google and Microsoft Ads at the same time?

Import your Google campaigns into Microsoft Advertising, run them with a defined test budget, and compare cost per acquisition after a full conversion cycle. Keep the budget where the measured cost per acquisition is better, and review the split quarterly.

Broad, phrase or exact match?

Exact and phrase give control and clean data, which is what a new account needs. Broad match works once you have solid conversion tracking and a strong negative list, because the bidding system then has the signal to steer it. Start controlled and widen deliberately.

How long before an SEM campaign settles?

Automated bidding typically needs a learning period of a week or two per significant change, plus your conversion lag. Plan a 30 day first read and a 90 day decision point.

Should I bid on my own brand?

Bid on brand when competitors appear on your brand query, when you want to control the message on a high intent search, or when your organic listing does not occupy the top of the page. Keep it in its own campaign so it never flatters the rest of the account.

A search surface where no one bids

An SEM campaign splits budget between paid and organic across several search engines. A further surface takes a share of the same buyers: assistants that answer in text and sell no placement. The reply names companies, and nobody can buy a slot in it.

Plan for it in the same review. Ask the campaign's top queries as sentences in two or three assistants, and see which brands are named. AI Knows Us runs those questions and gives you the cited sources, so the answer shows up in the report next to your paid and organic numbers.