By: Brandon Ponzo | Co-Founder & CEO at swyft
My first job in legal was running growth at a personal injury firm. The assignment fit in three words. Get more cases.
Before I spent another dollar, I wanted to know what was already working so we could do more of it, and what wasn't so we could stop. Reasonable place to start. We just had no way to answer it.
Every case that came in got tagged in one free-text box called Source. That box was the whole system. What lived around it was institutional knowledge in people's heads. Google's big. These referral partners are great, those aren't. No, we don't have billboards, thank God. Some of it true, some of it arguable. None of it queryable, none of it consistent, and all of it resting on the idea that a lead touches your firm exactly once before deciding to hand you their case. Leads don't work that way, especially now. People circle a firm two or three times before they ever pick up the phone, and we were writing all of that down in one box, when we wrote it down at all.
We had a data problem.
So we put in a real CRM and started tracing every lead back to where it actually started. It wasn't cheap and it wasn't fast. No one loves writing checks for plumbing they can't see. But once we could see where cases came from, the picture was humbling. Channels we'd defended for years turned out to be loud and not much else. Sources we barely thought about were quietly carrying the firm. We'd been rewarding the wrong things, because the wrong things were the easiest to see.
We did more of what worked, cut what didn't, and more than tripled the cases we signed without adding a dollar to the marketing budget. We didn't buy more leads. We figured out which of the leads we were already paying for were worth our time, and went and got more of those.
The most expensive guess in the building
What we were missing has a boring name. Lead source attribution. Knowing exactly where every lead and every signed case came from, down to the specific source instead of the rough category.
John Wanamaker, a Philadelphia merchant who built one of the country's first department stores and got called the father of modern advertising for his trouble, supposedly said a century ago that half the money he spent on advertising was wasted. He just didn't know which half. That's the whole problem in one sentence, and most law firms are still living in and paying for it.
In fact, this problem costs more in law than almost anywhere. Legal is the priciest category on the internet to advertise in. Ahrefs ran the ten thousand most expensive keywords on Google and found legal terms make up more than 17% of them, the largest share of any industry. One click on "truck accident lawyer corpus christi" runs about $500 USD. That's just the click, not the lead, and most of those clicks never call. By the time you've signed an actual case you can be four to twelve thousand dollars in, depending on your market. Lead generation is the biggest line item most firms have, and most of them can't tell you what it bought.
Attribution is a resolution problem
I think about it as resolution now. Most firms see their lead sources like a blurry photo. A green blob they call Google, a gray blob they call referrals, and that's about as sharp as the image gets. Attribution is the work of pulling that photo into focus until you can read the license plate. Focus comes in levels.
Start with paid, because it's the easiest. The blob is "Google Ads." One notch sharper is the campaign. Sharper than that, the exact keyword. Sharpest is the keyword plus the landing page plus what happened after the call connected. But if you stop at "we spent forty grand on Google and got some cases," that tells you next to nothing you can act on.
The reason people stop there is last-touch attribution, which is what every ad platform shows you by default. Last touch hands all the credit to the final click before the phone rings.
It's the same mistake as that one Source box at my old firm, rebuilt in software. It assumes a lead arrives in a single touch when almost none of them do. A guy gets rear-ended, googles "what to do after a car accident," lands on your explainer article, reads the whole thing, decides he trusts you. Two weeks later he's ready, types your firm's name into Google, clicks, and calls. Last touch records that as branded search and gives the article that earned him nothing. So next quarter you cut content and pour the money into branded search, and you've quietly defunded the thing that was working. Your budget drifts toward whatever sits last in line, steered by your measurement instead of your market.
There's a gold rush right now to rank inside AI. Firms are paying agencies to get them mentioned in ChatGPT and Google's AI Overviews, and I get the pull. Search really is changing. Roughly six in ten Google searches now end without a click to anyone's site because of AI Overview results like what Gemini displays for a Google search. Seer Interactive measured organic click-through falling from about 1.8% to 0.6%. The floor is lowering rapidly.
Be honest about the size of the prize, though. AI referral traffic to legal sites is growing fast, almost 12x year over year by Previsible's count, and it still adds up to under 1% of traffic. Firms are rebuilding entire content strategies around a channel that today sends them a rounding error, and almost none of them can name a single blog post that ever produced a signed case. They can see traffic all day. What they can't see is which post produced a client. They're chasing rankings on queries they've never tied to a dollar, sitting on years of content they've never brought into focus at the page level. High resolution here is easy to describe and rare to find. Which page. Which post. Which exact question someone typed right before they found you.
The firm name is not the source
The level that matters most in plaintiff work is the one firms botch worst. Referrals.
The best cases we ever signed came from other lawyers. They show up pre-qualified, trust already loaded, sent by a colleague the client respects. Now watch a firm log one. A case comes in from another attorney. Someone on intake opens the referral field and types "Smith Law Group." Next time it's "Smith Law." The time after that, "John Smith." One referring attorney quietly becomes five records that never add up to a number you can use.
And the firm name isn't even the right resolution. The relationship doesn't live at Smith Law Group. It lives with the person who picked up the phone and thought of you. Sometimes that's a partner, but when it's not and that attorney moves to another firm, and attorneys move constantly, your data still says Smith Law Group while the actual relationship walks out the door.
The ten-case test
Want to know where your firm actually stands? Pull your last ten signed cases and name the exact source for each one. The keyword, the page, or the person. Not the channel they all sit under. The specific source.
Ten for ten, you have a real attribution system and you should be proud of it, because almost nobody can do it. Miss a few and you're Wanamaker, guessing which half is wasted, except you're doing it with five-figure chips on every hand.
None of this is really about dashboards or marketing reports. Attribution is how a firm sees. Every decision about where to spend, what to write, which relationships to feed runs straight through it. When the picture's blurry, the decisions are guesses in a nicer outfit.
We didn't triple our caseload by getting better at marketing. We tripled it by finally being able to see, then doing more of what the data said was working. The cases were already out there. We were just blind to which ones we were already winning.
That blindness almost always starts at the first moment, the point where a lead comes in and nobody writes down where it came from. Close the gap there and everything downstream gets sharper. It's one of many rare things the Lawbrokr team is actually building for, which is part of why they're such a great integration partner for Swyft, where lawyers come to leverage their data to make better business decisions. Better data = better business decisions. We're a match made in heaven.
But the software's the easy part. The hard part is deciding the guessing has to stop. So pull your last ten cases and try to name where each one came from. Whatever number you get, that's the real one. That's your firm's vision, uncorrected. Everything good starts with admitting how blurry the picture's been.
