Build or Buy: When Custom Lead Software Beats a Data Subscription

There are three ways to get a list of companies worth calling. You can buy a list outright. You can rent a data platform and pull records from it every month. Or you can build a pipeline that assembles the list yourself, from sources you choose. All three work. They work in different situations, and the wrong choice for your situation is expensive in a way that takes about a year to become obvious. This is the framework we use when a prospect asks us which one they need — including the cases where the honest answer is "not a build."

The Three Options, Stated Plainly

Strip away the vendor language and the options are simple.

Buy a list. You describe a segment, someone delivers a file. You own the file. It starts decaying the day you receive it, because companies move, people leave, and phone numbers get reassigned. It is the fastest path from decision to dialing.

Rent a subscription. You pay for access to a vendor's database and pull records against their filters. The data is refreshed on their schedule. You keep access as long as you keep paying. The filters are the filters — you work inside the boxes the vendor built.

Build a pipeline. Software that goes to sources — public records, licensing boards, permit portals, registries, directories, sites you name — pulls records on a schedule, cleans them, matches them, enriches them, and drops them somewhere your team works. You own the code and the database. It runs until you turn it off.

Those are different products solving overlapping problems. The mistake is treating them as three price points on one ladder.

When Buying a List Is the Right Call

Buy a list when speed matters more than fit. Specifically:

• It's a one-off campaign. A trade show, a regional push, a product you're not sure you'll sell again next quarter. Building infrastructure for a campaign with an end date is a bad trade.
• You're testing a segment. You think dental practices might buy. You don't know. A list is the cheapest way to find out you were wrong. Test first, build second.
• You have no technical capacity and no appetite to acquire it. A pipeline is software. Software has an owner. If nobody at your company will be that owner, a build turns into a stranded asset.
• You need it this week. Builds take time. A list does not.

If you're not sure the segment converts, buy the list. Prove the segment, then decide whether to industrialize it. We sell lists for exactly this reason, and we would rather sell you a list than a build you didn't need yet.

When a Data Subscription Is the Right Call

Subscriptions get criticized a lot by people selling alternatives. They're good products for the job they do. Rent one when:

• Your ICP matches the vendor's filters. If "software companies, 50 to 500 employees, US, with a VP of Engineering" describes your buyer, that is a solved problem. Every major platform has that filter. Building it yourself is reinventing something you can rent tomorrow.
• You need breadth, not depth. Firmographics across every industry, technographics, org charts, contact data at scale. That breadth is genuinely hard to assemble. Vendors have spent years on it.
• Your team is already trained on one. Tooling switches have a real cost that never shows up in the comparison spreadsheet. Reps who know a platform's search syntax are productive in it. Reps handed a new interface are not, for a while.
• Your volume is low and steady. Below a certain usage, a subscription is just cheaper than owning anything.

If two or three of those describe you, buy the subscription and stop reading comparison articles. This one included.

When Building Actually Wins

Building wins in situations that share one trait: the thing you need does not exist as a product you can buy.

Your ICP is a filter nobody offers

"Roofing contractors who pulled a permit in the last 30 days." "Restaurants that just got a liquor license." "Property owners who filed for a rezoning." "Trucking companies whose DOT authority went active this month." No platform has that dropdown, because that dropdown is your business and not theirs. When your qualifying criterion is an event rather than an attribute, you are outside what subscriptions do. This is most of what we build — see custom lead generation software for how those pipelines are put together.

Your segment is SMB long-tail

The major databases are deep on companies with a web presence, funding history, and a LinkedIn footprint. They get thin fast on the two-truck HVAC company, the independent motel, the single-location clinic. If your buyer is a small local operator, you will hit coverage gaps that no amount of filtering fixes. Those businesses do exist in public records — licenses, registrations, filings, inspections. They're just not in the platform you're paying for.

You need timing, not just identity

Knowing a company exists is worth little. Knowing it did something last Tuesday is worth a lot. Permits, licenses, filings, and registrations are timestamped, and that timestamp is the whole signal. Our PermitPulse work exists because permit data answers "who is about to spend money" in a way a firmographic database structurally cannot.

Your annual data spend is large enough that a fixed cost dominates

We won't quote competitor pricing, so do this yourself: add up every seat, every credit pack, every export overage, and every list purchase from the last twelve months. Compare that number to a one-time build starting at $1,500 plus whatever it costs to keep running. If the recurring number is a multiple of the fixed number, the math is not subtle.

You want an asset instead of an expense

A pipeline you own is a thing on your balance sheet that keeps producing. That matters more to some companies than others. If you're building a business to sell, proprietary data infrastructure is a different kind of line item than a renewed vendor invoice.

The Ownership Argument, Fairly Stated

The pro-build case is easy to overstate, so here it is with the caveats attached.

True: subscriptions stop when you stop paying. Everything you learned about the segment stays with the vendor. Renewal negotiations happen on their terms because switching means retraining a team and re-establishing workflows. A build you own doesn't have a renewal date. It runs on your infrastructure, into your lead database, and it keeps running through a budget freeze.

Also true: software rots. Sources change their HTML. A county puts its permit portal behind a new system. A state moves its licensing search. Formats drift. A pipeline nobody maintains produces silently wrong output, which is worse than producing nothing, because nobody notices for six weeks.

The most common way a custom pipeline dies is not a technical failure. It's that the person who wanted it built leaves the company, and nobody inherits caring about it.

Ask before you build: who checks that this ran? Who gets the email when a source breaks? If the answer is "we'll figure that out," figure it out first. A maintenance arrangement — yours or ours — is part of the build, not an upsell on top of it.

A Build We Run, With Its Running Costs

The comparison is easier with a real pipeline in front of you. PermitPulse, the short-term-rental permit dataset this company sells, is a build of exactly the kind this guide describes: it reads public government registers on a schedule and turns them into one file. On October 23, 2026 it held 283,611 records from 56 registers large enough to have their own page, and 127,924 of those records carry a phone or an email.

What it runs on. One scheduled worker and one database on Cloudflare. Cloudflare's published pricing for the Workers Paid plan starts at a minimum charge of $5 USD per month for an account, and its own worked example of an hourly scheduled worker comes to $6.99 a month (Cloudflare Workers pricing). Infrastructure is the small number in a build like this. The large number is the hours spent keeping sources working.

What changed in one quarter. Between July and October 2026, without any change to the product, the pipeline needed work because the sources changed: Denver's register moved to a new map layer; one county register started blocking requests from cloud servers; a Florida city put its register behind a sign-in token, so pulling from it stopped (a build should stop there, not work around it); map services delivered phone numbers as decimals, which failed phone validation until the values were converted; and several layers that looked like permit registers turned out to be zoning maps or parcel lists and had to be excluded by name. Each of those is a day or less of work. None of them is optional.

What it would cost to buy instead. We won't quote another vendor's price. The useful comparison is structural: no list vendor we know of sells "every short-term-rental permit holder in these registers, refreshed daily", because the product only exists once someone writes the pullers. When the data you need is a set of public registers rather than a business directory, the build is not the luxury option. It is the only option, and the question becomes who maintains it.

Hidden Costs on Both Sides

Every option has costs that don't appear in the initial comparison.

Subscription side. Per-seat creep is the big one — you price it for three reps, then hire two more, and the number moves without a new decision being made. Credit systems meter exports, so the cost scales with usage in ways that are hard to forecast. Data decay is real regardless of who holds the data; a refreshed database is not a correct database. And there's a lock-in cost that only shows up at renewal, when your workflows, your saved searches, and your team's habits are all denominated in one vendor's product.

Build side. Source changes, as above. Scope drift — the pipeline gets extended six times until nobody remembers what it does. Deduplication and matching are harder than they look, especially across sources that spell the same company four different ways; that's usually where enrichment and matching earns its keep. And the orphaning problem, which is not a technical cost at all but kills more pipelines than any bug.

List side. A list is a snapshot. It is correct on delivery and less correct every week after. Budget for the fact that you are buying a decaying asset, and use it fast.

A Checklist You Can Apply Yourself

Answer these honestly. Not aspirationally.

• Can you write your ICP as filters that exist in a product you can buy today? If yes, subscribe. Don't build a worse version of a shipped product.
• Is your qualifying signal an attribute or an event? Attributes are well served by platforms. Events usually are not.
• Are your targets large enough to have a real digital footprint? If they're small local operators, expect coverage gaps.
• Have you proven the segment converts? If not, buy a list and prove it before you build anything.
• What did you spend on data in the last twelve months, all in? Include seats, credits, overages, and one-off purchases.
• Will you still be selling to this segment in two years? Builds amortize over time. Short campaigns don't.
• Who owns the pipeline after it ships? A name. If there isn't one, the answer is no, or the answer is a maintained build.
• Does the data need to land somewhere specific? A CRM, a dialer, a routing rule. Delivery is part of the requirement, not an afterthought.

Three or more answers pointing toward a build is a real signal. One is not.

The Hybrid Answer Most Companies Should Take

The framing of "build or buy" implies a fork. In practice the right move is often both, sequenced.

Buy a list now. Your reps need something to work this week and a pipeline does not exist yet. Work that list, and while you work it, pay attention: which records converted, which criteria mattered, what you wished you could have filtered on and couldn't. That is the specification for the build, and it's a much better specification than anything you'd write in a planning meeting before you'd made a call.

Then build against what you learned. The list bought you time and told you what to build. The pipeline replaces the repeat purchase.

The same logic applies to subscriptions. Plenty of our clients keep one and run a custom pipeline alongside it. The platform covers the broad firmographic work it's good at. The pipeline covers the segment or signal the platform can't see. Nobody has to lose that argument.

How to Decide From Here

If the checklist pointed you toward a list or a subscription, take that. You'll get to work faster and you won't be maintaining something you didn't need.

If it pointed toward a build, the next step is scoping — which sources, what the qualifying logic actually is, where records land, and who owns it after handoff. Builds start at $1,500, and the honest range depends almost entirely on how many sources are involved and how messy they are. Some jobs are one source and a clean schema. Some are eleven sources that disagree with each other.

We work across a range of verticals, and the patterns repeat more than you'd expect — see industries for what we've built before. If you want a straight answer about which of the three options fits your situation, tell us what you're trying to reach and we'll say so, including when the answer is that you don't need us to build anything.