Sean Boots

Technology, public services, and people. But mostly people.

The “missing middle” in software procurement

Version history

This year’s FWD50 conference was a couple weeks ago. Like previous years, it’s home to a lot of interesting conversations on technology, governments, and society; it also serves as group therapy for public servants who are trying to change their slow-moving organizations. I wrote some reflections on it last year. Both last year and this year, the conference took place online, and the FWD50 crew sets the bar very high for running virtual conferences that are genuinely interesting and fun to be a part of.

One new event this year was a game show-inspired “pitch competition”, where public servants pitched ideas for policy changes or other initiatives that could better enable digital work in government. My friend and colleague Meg Beretta (a digital government legend in her own right) organized and co-hosted it, and the other participants (from across federal and Ontario public service organizations) came up with really interesting and inspiring pitches. Just getting to meet and chat with them “backstage” was a real highlight:

My pitch was about procurement. And also about urban planning, as a shameless way of combining two of my favourite topics. Below is a recap of the presentation; as always, I’d love to hear your thoughts!

The missing middle

Three photos collaged together: on the left, in greyscale, a detached suburban house on a street corner in greyscale. In the middle, superimposed in colour, a neighbourhood in Paris with 6-story residential blocks on each side of a café-lined pedestrian street. On the right, an under-construction 40-story skyscraper in greyscale.
Caption: Photos I’ve taken over the years showing three different neighbourhood building types (from left to right: Whitehorse, Paris, and Ottawa). See my full presentation slides here

In the urban planning world, people use “missing middle” as a term to describe how neighbourhoods are designed and built. In North America, most neighbourhood housing tends to be either single-family detached homes, like you’d see in most suburbs, or – in large cities – really giant condo skyscrapers. There’s small, huge, and not much in between.

In other parts of the world – Paris, for example – a lot of neighbourhoods are 5- or 6-story housing blocks. They’re dense enough to support a vibrant community, there are little restaurants and cafes everywhere and you can easily walk around, and it makes parks and public transit all work really well. Montreal is one of the only North American cities that has a similar vibe, and people love it.

That’s the metaphor – there’s a “happy medium” for urban design, and in North America, we don’t have nearly enough of it. What does that look like for software procurement?

Why it matters

Let’s start with what we’re trying to accomplish. I’ve basically dedicated my career to helping public servants get the tools and equipment they need to be really really great at what they do. I want public servants to have software and tools just as good as people who work at Shopify or Google, no matter where they work in government. Anyone who has worked both in government and in the tech sector knows that, in most departments, there’s a giant gap. In government, we lose talented people all the time because our default tools are so bad. That’s what I want to fix.

If you want to do good work, you need good tools. In today’s day and age, where we’re all working online from home, wherever we are across the country, a lot of that involves software products that we access online.

Some of these tools are free, but a lot of them are online subscription services. Tech folks call this “software as a service” or SaaS. These are things like Slack, GitHub, Trello, Figma, or Mural that you access through a web browser and that let you collaborate in useful ways with other colleagues.

For a long time, in the federal government, you couldn’t even access these tools through the firewall; that fortunately started to change in 2018 thanks to a policy update from the GC Office of the CIO. There was another really useful change this year from the Office of the Comptroller General (the group that does procurement policy), saying that yes, you could accept the standard terms and conditions for low-cost, low-risk tools, which means: now you can even pay for them. All of this is really helpful.

The missing (procurement) middle

What’s the missing middle for procurement?

If your tool is inexpensive and your team is small, amazing, you can put the subscription cost on a departmental credit card and call it a day.

And, if your tool is really, really expensive and sold by one of the really large IT vendors, no problem, there are contracting vehicles in place from Shared Services Canada (SSC) or Public Services and Procurement Canada (PSPC) that you can get it from.

In other words: If I want to spend $20 bucks a month, good to go on a credit card. If I want to spend $5 million dollars a year? Good to go on an SSC or PSPC vehicle or supply arrangement.

What’s missing? The in-between.

For example: you’ve got a team of 150 people. There’s an online tool that everyone in the tech industry uses. It costs $30 per person per month. That means in a year, you’re going to spend about $60,000 dollars on this tool. How do you do that? You can’t.

What are the barriers?

The barriers here are specific to the federal government (in Canada), but I wouldn’t be surprised if other governments and jurisdictions have similar issues.

In federal regulations – the Government Contracts Regulations or GCRs – I can (generally) sole-source something if it’s less than $40k after taxes. This $40k limit is per program, and is usually interpreted as over all time (i.e. the lifetime of the program), which means that any subscription service will run into this limit eventually.

When you’re trying to buy software-as-a-service products, some procurement folks actually still consider them a “good”, picking up on the word “software” – because back in the day, you bought software in a box from Staples. In that case, the limit would be $25k after taxes. Even less room! And yes, “software as a service is a good” is a sentence that would only be spoken in government.

(Fortunately, as an unexpected side effect of PSPC moving from GSIN categories to the United Nations Standard Products and Services Code or UNSPSC, there’s now a clear category for “Computer software rental or leasing services” which covers software-as-a-service products… as a service.)

So: I’ve got an online tool that costs $60k per year.

I can’t sole source it, since it’s more than $40k. I could do a standalone competitive procurement, with a Request for Proposal (RFP). Here’s a few of the challenges there:

  • It’ll take 6 months to 2 years to happen
  • The company that makes the tool the team actually wants might not even bid
  • The winner of the RFP will be whoever writes the most convincing bid, not who makes the best tool
  • I have to know from the very start how many licenses I need, several years into the future, and if my team grows more than that I’m totally out of luck anyway.

Not to mention: my procurement and legal folks will say, you need to set up custom contract and liability terms with the vendor, “to protect the Crown”. And the vendor will say (this is a true story) we won’t bother negotiating custom terms for anything less than $250k USD. It’s not worth the cost of their lawyers otherwise. And my procurement people will say, sorry, in that case, you’re out of luck.

(I’ve heard recently, from more grizzled procurement experts than I, that the most realistic workaround for this is simply to spend the $250k, by buying more expensive licenses or committing to a longer time period. Which is …not exactly an improvement? Locking ourselves into spending more money because we won’t be more procedurally flexible is not a good look.)

I’m trying to spend $60k, and I still can’t.

If I wanted to spend $5 million, I’d get in touch with SSC or PSPC’s procurement folks, who would say: yep, here’s our standing offer with Microsoft, here’s our standing offer with Amazon Web Services, here’s our standing offer with (insert large company here).

Even for medium-large companies, they might say: here’s our supply arrangement for Atlassian products, or here’s our supply arrangement for Adobe products, where three or four companies you’ve never heard of are reselling the products from those companies to government customers. These are considered competitive procurements (and therefore not subject to the $40k limit) because several authorized-reseller “middlemen” competed to be on the supply arrangement. (Classic.) But if the software tool you want doesn’t have government resellers, and doesn’t have a supply arrangement, once again, you’re totally out of luck.

What do you do instead? You stop using the tool that everyone loves, and all your best people quit and go work for Shopify. Not great.

What should change?

When I pitched these to the judging panel at FWD50, one of the judges quite aptly pointed out that my description of the problem was great, but the solutions I suggested weren’t necessarily realistic. This is true! These changes, at the least, would involve a procurement regulation change, which would require a consultation period and Cabinet approval. All the more reason to get the ball rolling sooner than later, that’s what I’d say.

Here’s what I’d recommend:

  1. Clarify that the $40k sole-source limit in the Government Contract Regulations is per year. That way it’s way clearer how subscription services fit against the thresholds, and it still nicely fits the intent behind that particular exemption (of it not being cost-effective to procure at such low dollar values).

  2. Increase the sole-source limit for services from $40k to $100k, again, per year, to make more room for medium-sized teams to sole-source online tools. That still fits, just barely, under the whole bunch of international trade agreements so all good there.

  3. Add an exception that, if a company only offers its product via credit card purchases, you can do a credit card purchase even above regular departmental limits. In most of these cases, 99% of a company’s customers are not governments. They’re not going to bother with contract negotiations and invoices. If we can’t buy their stuff, that’s our problem, not theirs. Governments need to get used to being small, unimportant customers, and to paying for things on a credit card like everyone else does.

Friends of mine in procurement would say: you should just compete it! Or convince the company to set up resellers and a supply arrangement. But the truth is, these companies aren’t going to. 99% of their customers are not in government. Doing a competitive procurement means, 1) spending a ton of time and effort on things that aren’t delivering high-quality services, and 2) at the end of the day ending up with a different product than the one that all my tech and design people actually want to use.

Plus, I don’t have one online tool to procure, I have seven that will run over the sole-source limit sooner than later. And in a year, the best-in-class tool for a given use-case will change, and my team will want to use the new one from a different company instead. There’s no way our existing procurement processes can keep up.

So that’s the situation:

  • If I want to spend $200 on a credit card, no problem.
  • If I want to spend $5M via an SSC or PSPC vehicle, also no problem!
  • If my online tool has authorized resellers and a supply arrangement, also no problem.
  • But if it’s above the sole-source limit and doesn’t have resellers, I am 100% stuck. It’s the elusive 5-6 story medium density neighbourhood that doesn’t exist in North America.

And to fix it, my pitch is:

  • Change sole-source limits to be time-based, for subscription and recurring products. So, $40k per year instead of $40k in general.
  • Bump the sole-source limit for services as far up as you possibly can, which given trade agreements is probably $100k/year.
  • Explicitly bless paying for things on credit cards and accepting the commercial terms when it’s the only option available from a given vendor, even up to that $100k/year limit.

Procurement: it’s the best (?)

Hello from year four, we’re right on time:

Would the changes I’m suggesting open the door to more questionable sole-source contracts (e.g. for $99k instead of the just-under-the-limit $39k contracts you frequently see nowadays)? It’s possible, although limiting it to subscription services that are already-complete deliverables (versus, $100k of “working on an IT project that we’ll finish eventually”) would certainly help. Would it enable more “shadow IT”? Yes, but that’s a feature, not a bug. Would it make public servants more productive and effective? 100%.

I don’t want my team to be the only crew in government that can use modern tools; public servants deserve them no matter where they work and no matter what they do.