Google Ads, Microsoft Ads or ChatGPT Ads: where should a UK business start?
by Harry Watts · · updated · 7 min read
There are now three serious places for a UK business to buy search-shaped advertising, and whoever's selling each will tell you theirs is where to begin. My fee doesn't grow with your ad spend, and I charge a reduced rate for each additional platform, so I have no incentive to push your business onto one unnecessarily. Here's the order I'd put them in.
The short version: it's an order, not a choice. Google to prove the offer, Microsoft when the fit arrives, ChatGPT as a measured experiment. The long version is why, and the two mistakes that waste more money than any platform's prices.
The three, in one honest paragraph each
Google Ads carries more than nine in ten UK searches. It has the highest prices of the three for the least mysterious reason: that's where the customers are, so that's where the bidders are. It is also the only platform with enough volume to teach you anything quickly: whether the offer converts, what a customer costs, which search terms produce revenue.
Microsoft Advertising is about 6% of UK searches, but over 12% of desktop search and under 1% of mobile. Clicks run about a third cheaper because fewer advertisers bid, and it holds the one card Google can't match: LinkedIn profile targeting, on ordinary search campaigns. A second channel, not a first one.
ChatGPT Ads reaches 16m+ UK users a month on the free tiers, with no reliable cost benchmarks yet, category bans that exclude several industries outright, and location targeting that reached postcode level in a live UK account on 6 September 2026, six days before OpenAI's documentation said it could. It's a test lane, currently a subsidised one, with OpenAI's £500 spend-match credit landing in UK inboxes.
| Microsoft | ChatGPT | ||
|---|---|---|---|
| UK reach | 9+ in 10 searches | ~6% of searches; 12%+ of desktop | 16m+ users on free tiers |
| Targeting | Keywords, radius, audiences | Keywords + LinkedIn profiles | Context hints; postcodes where supported, UK districts accepted in the picker |
| Click prices | Highest (most bidders) | Often ~a third cheaper | No UK benchmark yet |
| Best for | Proving the offer; almost everyone | B2B, desktop, capped winners | National e-commerce and B2B tests |
| Role | First | Second | Experiment |
Start on Google, almost always
Not because it's best — because it's first in the only sense that matters: learning speed. A new advertiser's real job in the opening months isn't buying customers, it's buying answers. Does the offer convert? What does a customer cost? Which searches pay? Google's volume answers those questions in weeks; the other platforms' volumes answer them in quarters.
And everything flows downhill from Google. The working account imports into Microsoft in an afternoon. The search terms that convert become the raw material for ChatGPT context hints. Nothing transfers in the other direction. Start anywhere else and you're learning slowly on a platform that can't export the lessons.
The exceptions are rare and mostly aren't platform choices: if Google's CPCs genuinely don't clear your cost-per-customer arithmetic, the problem is usually the economics or the offer, and a cheaper platform postpones that conversation rather than winning it.
Add Microsoft when one of two things happens
- Your winners are capped. Profitable Google campaigns losing impression share to budget mean demand you can't afford at Google's prices: Microsoft sells more of the same searcher for about a third less.
- Your buyer sits at a desk. B2B, professional services, considered purchases: the LinkedIn targeting plus the office-desktop audience makes Microsoft clicks cheaper and better qualified.
If neither is true (mobile-first local services above all), skip it without guilt. An eighth more market is only worth an afternoon's setup when your customers are in it.
Run ChatGPT as an experiment, sized like one
Permitted category, national reach, curiosity about where attention is going: those are the qualifications. The discipline is the same one the June verdict set out: a testing budget, chunkier where the potential is high, pixel and Conversions API wired first, judged at 60–90 days on cost per customer. The £500 credit halves the tuition while it lasts; it doesn't change the syllabus. The spend each platform wants first, and the deadlines attached to it, are compared across all three.
The two mistakes that beat any platform's flaws
Splitting a small budget three ways on day one. Every platform's bidding needs conversion volume to learn. Divide £900 into three £300 slices and you've built three campaigns that will all stay stupid, each starved of the data that would have made any one of them work. Concentrate until something converts; diversify from proven headroom.
Running three platforms on one platform's tracking. Google's tag doesn't measure Microsoft; Microsoft's UET doesn't measure ChatGPT; ChatGPT's pixel measures neither. Each needs its own properly wired measurement, under the same consent setup. Skip that and whichever platform you didn't wire will look like a failure regardless of what it delivered: the multi-platform version of flying blind.
The decision, as a flow
- 1. Everything on Google until the offer converts reliably: 60–90 days, judged on cost per customer.
- 2. Capped winners or a desk-bound buyer? Import to Microsoft, cut the bids, install UET.
- 3. Permitted category and national reach? ChatGPT on a testing budget, credit gratefully accepted.
- 4. One number across all of it: blended cost per customer against your target. Platforms are lines in that sum, not teams to support.
Quick answers
Which platform first?
Google, almost always: nine in ten UK searches means it answers "does this work?" fastest, and everything learned there exports to the other two. Nothing transfers back in.
When does Microsoft join?
When profitable Google campaigns are budget-capped, or your buyer is B2B and desk-bound. Then it's an afternoon: import, cut bids, install UET.
How do I split the budget?
Early on, you don't. Concentration feeds the data that makes bidding work. Microsoft gets funded from proven headroom; ChatGPT stays on a testing budget.
Isn't the cheapest platform the smart start?
Cheap clicks reflect absent competition, and competition follows customers. Judge cost per customer: a discounted click from the wrong audience is the expensive kind.
One tracking setup for all three?
No. Google's tag, Microsoft's UET and ChatGPT's pixel/CAPI are separate installations, all wired into the same consent setup. The unwired platform always looks like the failure.
Where your business lands in this
My fee doesn't grow with your ad spend on any of the three, and each additional platform is charged at a reduced rate. A free intro call places you in the flow above in about thirty minutes: what Google needs to prove first, whether Microsoft's audience includes yours, and whether the ChatGPT experiment is worth a slice yet.