Getting our sales into shape

It all really started when a client relationship of over seven years came to an end. It was our longest and biggest, the kind of thing a lot of routine builds up around over the years. When it ended, a big chunk of work vanished from the calendar in one go, and empty space was left in its place.

As an entrepreneur you have two options. Either you sit around mourning what ended, or you use the time on something you should have done long ago. For us, the latter meant sales.

Over the past year we’ve been working our sales and the processes around it forward in a big way, all the way from building lists through cold outreach to proposals and contracts. In this first part I’ll go through what we’ve done and with what tools.

First, get the data in order

Sales without proper data is mostly guesswork. So the first thing we tackled was where we even get information about Finnish companies.

We pull company data from Clevenio, and it was already in use before. Just very narrowly. In practice it was used every now and then to look up individual companies, and that was it. The tool existed, but it was nowhere near fully used.

This is a pretty common situation, by the way. Companies have plenty of software they use for maybe ten percent of what it could do, and the missing ninety percent still gets paid for every month.

Once I started putting more time into sales myself, we took Clevenio back onto the table and looked at what it can do when you use it properly. Finnish company data is in there in the form you actually need it, and you don’t have to build your own scraping or pay through the nose for international databases where Finland is little more than a footnote.

For us, Clevenio has turned out to be exactly the right tool. The biggest win is time: finding contacts that match our ICP takes a fraction of what it used to.

They also have a handy MCP, which in practice means I can pull company data straight through an agent without opening any user interface in between. At this point, building a prospect list is more conversation than clicking.

CRM switched from Severa to HubSpot

These used to run on top of Severa, but we’ve since moved on from it. HubSpot was chosen as the CRM.

In all honesty, there’s no particular love for HubSpot. It just offers a handy API for doing things programmatically, and at this stage that’s the deciding factor. When you build automation and agents around sales, the most important feature of the system is that you can get into it programmatically. Everything else is secondary.

It may well be that this changes again at some point. Tools are tools, and you shouldn’t fall too much in love with them.

Pitches, calls and LinkedIn

Cold outreach has been ramped up in a big way. Pitches have been honed, cold calls made, and people contacted on LinkedIn far more systematically than before.

There’s no magic in this. It’s repetition, measurement, and admitting that the first version of a pitch is almost always bad. Only after you’ve made enough calls do you start to hear what isn’t working in your own story.

A new playbook and clearer goals

We built a new sales playbook and made sales more goal-driven than before. We did sales earlier too, of course, but it wasn’t quite as trackable. There was activity, it just didn’t leave proper data behind.

And that’s the real problem. When you don’t know what works and what doesn’t, all improvement runs on gut feeling. It feels like this pitch lands better, it feels like this industry books meetings more easily. Maybe the feeling is right, maybe not, but there’s nowhere to check.

In practice, the playbook means we’ve written down how we do sales. What happens at each stage, what the next step is, what a good proposal looks like. It’s not fancy and it’s not finished, but it exists and it can be improved. And most importantly, when the work is done the same way time after time, it finally starts producing the kind of data that’s worth looking at.

The other reason for a playbook is scalability. Ideally you can bring extra hands into sales when the situation calls for it, without having to transfer tacit knowledge from one head to another at the coffee table over a couple of months of shadowing. When the work is written down, a new person gets a grip on what’s done and in what order much faster. We don’t need that right now, but it’s good that it’s possible.

Writing the playbook is the easy part, though. Actually using it is still being drilled, and that’s a much slower thing than producing a document. Old habits sit tight, and it’s completely natural to fall back on how things were done before when you’re in a hurry. It takes reps, and reps only come from doing.

More out of meetings

Fireflies and Teams recordings are in use so we get more out of meetings. Before, what a meeting yielded depended on what you managed to write down and what you remembered afterwards. Now the conversation is stored, you get summaries from it, and you can go back to it when writing a proposal.

This is a surprisingly big change. What the customer actually said and what you remember the customer saying are rarely the same thing.

More hands and agents

Technical reviews and proposal writing are no longer just the founders’ job. Our developers now play a bigger part in the sales process, and especially in technical reviews and meetings their input is crucial. Who could better assess a customer’s system or answer a technical question than the person who actually does the work?

Proposals have a new division of labour too. They’re created and updated with Claude, and the finishing touches are done by hand. The agent handles the structure and the repetition; a human makes sure the end result looks like us.

Saying no

Maybe the most important lesson of the whole year. We’ve learned to say no better to the customers and deals that don’t genuinely serve either side.

The more there is in the pipeline, the more room there is to choose. And when you can choose, it pays to pick the deals where we’re genuinely good and where the customer gets the most out of us. Saying no here isn’t refusing, it’s choosing, and it’s a direct consequence of sales finally producing enough to choose from.

It’s also a favour to the customer. If we don’t happen to have the best expertise for that particular problem, everyone wins when you say so directly and can point to someone better. Surprisingly often that conversation comes back around later, on a completely different topic and with better timing.

A proper process helps here more than you’d expect, because then it’s not about gut feeling but about looking at criteria agreed on together.

This was part 1

The year has been an education, and looking back, that client relationship ending was maybe the best thing that could have happened to us. It forced us to change and improve things, so that the company stands on a more sustainable footing.

It may well be that more gets written about this. Especially about how these things were done in the technical sense: how Clevenio’s MCP, HubSpot’s API and the agents actually fit together, and what came up along the way. That’s the side that interests me the most, so I’m not promising anything, but I’m not promising not to write it either.