Cutting a £730 a month software bill to about £150, by checking every keyword and every subscriber
This did not start as a software exercise. It started as a conversation about revenue at Claims Bible, and one of the ways to make more profit is to spend less while what comes in stays the same. That is a less interesting answer than winning a new client. It is also a much more certain one.
I co-founded Claims Bible, so this was my own bill to look at. By the end of September 2026 two tools were taking about £730 a month between them, and neither was priced on what we were getting out of it.
Where the money was going
Semrush was $419.94 a month, about £311, on the Guru plan with a 500 keyword top-up I had bought on 21 September. The top-up is the part worth admitting to. I hit a tracking limit and bought my way past it instead of asking why I was tracking that much.
What that limit was carrying was 947 keywords spread across six projects, plus eighteen scheduled reports. Five of the six projects were from work that had finished months earlier. Of the eighteen reports, I could not find one that anybody read.
Campaign Monitor was £419 a month, £5,028 a year. It is priced by subscriber band, and we sat in the 50,001 to 75,000 band. That means the invoice arrives at the same figure whether you send a campaign that month or not. We were paying for the size of a list rather than for anything that left the building.
What surprised me
The wastage. We had paid about £1,600 in Campaign Monitor fees for a service we were not using. Not under-using it. Not using it.
Semrush was the same thing in a different shape, a plan full of features nobody opened, on a tier I had topped up rather than questioned.
Neither of those happened in one go, and that is the part worth sitting with. Both bills crept, a feature at a time and a band at a time, and because a subscription just turns up on the card every month there is never a moment where somebody has to stand up and defend it.
What I actually checked
I went through all 947 keywords one at a time, against two tests. Is there a firm at the other end who will buy this claim today, and is there real demand for the term in Search Console rather than an estimate in a tool.
That is much slower than filtering a list, and it is the only way the answer means anything. A keyword with volume and no buyer is a cost. A keyword with a buyer and no demand is a guess.
270 terms came out of it. 210 I kept from the existing set, and 60 new housing disrepair terms I added, ahead of a relaunch that has not happened yet, because a baseline is only useful if it exists before the pages go live.
I cut car finance hardest of all, from 108 terms to 41. That is below what my own review had recommended the day before. The funnel is paused and I know when it is coming back, so there is no sense paying to watch 108 positions move in the meantime. I kept the terms still earning clicks, and the FCA redress scheme terms, because that scheme will move whether we are ready for it or not.
Medical negligence came off entirely. I dropped a term that was earning eight clicks a month, which felt wrong while I was doing it, because there is no firm to send a medical negligence claim to. Clicks you cannot convert are not traffic, they are a bill.
Then I deleted five of the six projects, deleted all eighteen reports, dropped the top-up and moved from Guru down to Pro.
The thing nobody tells you about downgrading
Semrush does not take your add-ons off when you downgrade. The 500 keyword top-up stays on the account and keeps billing. The only button offered is one asking you to upgrade and give up your units, which is not a sentence that explains itself.
In my case the leftover add-on was £97.24 a month. That is more than the Pro plan I was moving to. A downgrade done the obvious way would have left me paying more and getting less.
There is no self-service way to remove an add-on on its own. You have to ask support to do it, and until they do, the saving is not real. If you take one thing from this page, take that one. Check your add-ons before you celebrate the new plan price.
The email side
The bigger number is the email one, and part of it is still an estimate, so I am going to be careful about how I put it.
The £419 monthly fee stopped on 1 October 2026, when I switched the Campaign Monitor account to pay per campaign, and the replacement went live on 3 October. That is Amazon SES for the sending, with Sendy holding the list. Sendy is a one off $69 licence running on hosting I already pay for, with the DNS on Cloudflare. SES charges per email rather than per contact, so the bill follows what we actually send instead of how many names sit in a database.
I am estimating about $20 a month. I want to say plainly that it is an estimate. This page argues for not trusting numbers you cannot evidence, so it would be a poor page if I passed a forecast off as a fact. The Semrush saving is banked and invoiced. The Campaign Monitor fee has stopped, so that part of the email saving is real, but what SES costs us is a forecast until the first bill lands.
The part I would have done regardless of cost: only contacts with a recorded opt in have moved across. That pulls the list down, which is where a lot of the saving comes from, and it also means the list I send from is one I can stand behind under PECR. Two reasons pointing in the same direction, which does not happen often.
What it costs now
Semrush is £136.08 a month including VAT, £1,633 a year, with 230 of the 500 keyword slots spare. That is room for the housing pages without buying anything. About £175 a month and £2,100 a year saved, and that half is done.
The email move takes £419 a month down to roughly £15, plus the one off licence. Call it £400 a month and £4,700 a year.
Together, about £730 a month before and about £150 a month after. Around £575 a month, or £6,800 a year, with the SES cost still to be proved by an invoice.
What I gave up
Historical ranking data on the five deleted projects, the Looker Studio connector, the extra scheduled reports, and the advanced tools that only come with Guru. I checked each one before I cancelled rather than after, and I was not using any of them.
That is the only reason this was a saving rather than a cut. The work was in the checking, not in the cancelling. Anybody can downgrade a plan in an afternoon. Knowing which 677 keywords you can stop watching takes going through them, and knowing which eight clicks a month you can afford to lose takes knowing who buys the claim at the end of it.
What I would tell another business
Keep a tight rein on your costs. If more than one person in your business can spend money, make somebody the gatekeeper.
That is not about saying no to things. Duplication is what you get without a gatekeeper. Two people buy two tools that do the same job, both plans grow features nobody uses, and at no individual step was anybody wrong.
The useful part is the question the gatekeeper asks, and it is not “do we need this”. It is “how do we use this without throttling the growth of the business”. That version lets you buy the expensive thing when it earns its place, and it stops you paying for the version of it you are not ready for yet.
Tell me what’s not working.
I’ll tell you whether I can help and roughly what it would involve.