A sponsorship agency whose entire sales engine ran on one inbox, a spreadsheet and a lot of paper.
The situation
Every business runs at least one process that made sense when it started and doesn't now. These processes rarely wear out slowly. They stop all at once, usually at the worst moment. Sometimes it's paper: work on a sheet in a drawer that nobody else can search, count or back up. Sometimes it's a tool or workaround the rest of the world quietly moved on from.
The business
A small sponsorship sales agency, selling billboards, sports kits, stadium boards, match programmes and hospitality on behalf of several client brands. The relationships were real and the product sold.
They email prospects from each client brand's own address, so the approach carries the club's or venue's name. That's the right commercial call, but it means their outbound email depends on domains they don't own.
Everything behind the selling was manual: one personal email account, a spreadsheet and a lot of paper. Renewal dates lived on paper, sale agreements were rebuilt in Word each time, and invoices were re-keyed into the accounts package. One person worked through about 4,000 prospects in BCC batches of 10 to 20. Over 2,000 emails a week, by hand.
It had worked for years. Then, in about a fortnight, it stopped.
The problem
Spam filters are built to catch batch BCC sends from a personal account, and the sending domain had no authentication to prove the emails were genuine. Deliverability collapsed and the account was blacklisted. Replies dried up, and for a business paid on commission, so did the income.
Nothing about how they worked had changed. The world around them had, which is why an expired process is so dangerous: you're doing nothing differently on the day it stops working.
Underneath that, there was no system to send into. No CRM, so no single place for a prospect, sale, invoice and renewal date, and no view of what income was due. The team lost 20 to 30 hours a week to manual work, and taking on another client brand would have meant hiring someone just for the admin.
Proof over promises
open rate on an early campaign, with deliverability restored
a week given back, down from up to 30 to three or four
more client brands the team now has the capacity to represent
ongoing automation subscription, because the layer was built rather than rented
Figures from a five month engagement, May to September 2026. The client is anonymised at their request.
The process
I don't build anything until I understand what I'm building on, so the work ran in stages. Each one ended with a decision, and the client was never committed beyond the next.
A conversation about how the business runs, where the hours go, and what the owner wants to be true in six months.
Working sessions mapping the operation from finding a prospect to renewing them. The result was a process map and eight friction themes, ranked by impact. The owner kept the audit whether or not we went further.
The audit became a set of costed, forecast options rather than one answer. Each listed its own risks, with contingency amounts quoted up front.
Built in milestones, each signed off before the next began.
The decision
Every option sat somewhere between off the shelf and built for you, and most owners are never walked through that choice.
Off-the-shelf software is quick and cheap to start, and a vendor keeps it running. But it was built for thousands of businesses, so you bend your process to fit it, pay a subscription every year, and live with someone else's roadmap: pricing changes, retired integrations, upgrades that break your workflow.
Bespoke software does what your business needs and nothing else, with no licence or seat count. It costs more, takes longer, and someone has to own it.
The answer is usually a mix: off the shelf where it genuinely fits, bespoke where it doesn't. Either way, four risks need pricing.
Does it actually do the job, or does it do 80% of the job and leave a manual workaround nobody mentioned in the demo?
Not the setup cost. What it costs to run in three years, at the volume you hope to reach.
Will it work with what you already own? An integration on a marketing page isn't the same as one that works on your data.
What happens when the vendor changes the product, the pricing, the API, or gets acquired?
I put these to the client in Phase 2, with numbers attached, before anyone committed to anything.
When plans meet reality
One named risk was how to automate the links between the CRM, the email platform and the accounts software. My contingency was a subscription automation platform: quick to set up, with a monthly cost.
The problem was timing. The client couldn't sell properly until a set point in the build, so their income was squeezed at exactly the moment we'd have added a new recurring cost.
So I built the automation layer instead of renting it, working directly with each platform's own interface and using software the client already owned. It took me longer, but it came in within the agreed budget, removed the subscription entirely, and does more than the original scope.
That's what the contingency conversation is for. Nobody can predict exactly what will go wrong, but you can agree in advance how to handle it.
The dependency
Proving the campaign emails were genuine meant adding authentication records to the clubs' and venues' domains, not my client's. Each was run by that organisation's IT provider, so the most urgent job in the project sat with people who had no contract with us.
My client isn't technical, so I wrote instructions each provider's engineers could act on, dealt with them on her behalf, chased where needed, and checked every record myself once they said it was done. It added time, not risk.
Nobody had ever checked whether the emails were landing, and nobody could have. A personal email account gives you no deliverability reporting, so there were no warning signs to ignore.
The contact list had the same blind spot. Nobody checks 4,000 contacts by hand, and the migration turned up around a thousand duplicate or invalid records. Better found on a migration than on a send.
The outcome
Without the change, this wasn't slower growth. It was the end of selling, in a business whose only route to market was an email account that had stopped delivering.
From up to 30 hours a week to three or four, and what's left is monitoring, not labour.
A 56% open rate on an early campaign, from emails that weren't arriving at all.
Opens, clicks and conversions are tracked, so the business can see which approaches and offers work. Decisions that were instinct are now evidence.
Emails no longer go out in hand-built batches, so the team can send far more, to more segments, in the same week. That's more earning potential, on top of the time saved.
Every sponsorship tracks its own expiry and prompts the renewal conversation in good time, so more sponsors renew and each customer is worth more over time.
A bespoke link connects the accounts software they already owned to the new CRM. A closed deal raises its own commission invoice and records its payment with nothing re-keyed, and overdue invoices are easy to spot.
Pipeline, renewals and expected income visible at a glance, for the first time in the life of the business.
The first sale from the first proper campaign came in at £1,000.
Saving time was the obvious win. The bigger one was capacity: selling for four more clients with the same staff is worth far more than an hour saved.
The handover
The last thing I handed over was a written training manual covering every process in the new system. Clients rarely ask for one. It's the most valuable part of the handover.
It trains your current staff and whoever replaces them, so the knowledge doesn't live in one person's head. It keeps everyone working the same way, and it gives your next developer a starting point instead of reverse-engineering the software.
It's also the document you give an A.I. system so it understands how your business works. A business that has written down how it operates can adopt A.I. properly. One that hasn't will spend a fortune finding out.
The lessons
Paper and outdated methods feel stable right up to the moment they are not, and you do not get a warning.
We could have started with the CRM, the bigger prize. It would have been a fine system feeding emails nobody received.
A build quote is half the question. The other half is what you're still paying each month in three years.
Something on your project will go differently to plan. Projects survive that when the conversation has already happened.
Every good decision in this project traces back to a process map made before anything was bought.
Over to you
If one of these gives you an uncomfortable answer, that's usually the start of a useful conversation, not a reason to panic.
Some things genuinely need paper. For most businesses, it's where the week goes: it can't be searched or counted, and only the person holding it knows what it says.
If nobody can tell you why a step exists, beyond the fact that it has always been done that way, that is the one to look at first.
If they left on Friday, what would go with them that is not written down anywhere?
Double entry is the clearest signal that two systems, or two people, are not talking to each other.
This was business process and automation work, not a media project. The same audit starts every engagement.
Common questions
Because the standards changed, not your email. Mailbox providers now expect sending domains to carry authentication records (SPF, DKIM and DMARC). Domains without them used to be tolerated and are now increasingly filtered, so the same message to the same people can stop arriving.
Yes, for two reasons. Repeated large BCC batches from a normal mailbox look exactly like what spam filters exist to catch. And BCC strips out personalisation, so fewer people reply even when the email lands.
From a standard mailbox, you largely can't. There's no deliverability reporting, bounce analysis or open tracking, so businesses find out through falling sales months later. A proper sending platform reports all of it.
They inflate your list, distort every metric and can push you into a higher subscription tier. Sending repeatedly to dead or duplicate addresses is also a spam signal. They build up quietly in any spreadsheet list, because nobody checks thousands of contacts by hand.
It depends what you're automating and for how long. Subscriptions are quick and cheap to start, but the cost never stops and you inherit the vendor's decisions. A build costs more once and nothing to run. For a stable workflow central to how you earn, building often pays back faster than people expect.
Every process, step by step, in language your staff can follow without the person who built it. It should train new starters, keep everyone consistent, and give the next developer a starting point. It's also what you give an A.I. tool so it understands your business.
How your business works, written down: which processes exist, in what order, who does what, the exceptions, and what your data means. Businesses with that written down can adopt A.I. tools quickly. Those running on habit and memory have to write it first.
Here, the build followed a four-week plan in milestones, after the audit and design phase. The variable is rarely the building. It's how fast decisions get made, and how fast third parties like a client's IT provider do their part.
Finding the root cause, then fixing it. Lake District based, working across Cumbria and North Lancashire.
© 2026 Xtreme Exposures Ltd
Privacy Policy