Thryv and Zuper alternatives, and the question a demo never answers: what can a machine reach?

Start with what anybody can repeat in ten minutes, because it frames everything after it. On August 23, 2026 we asked each vendor two questions in the only way that produces a checkable answer: we requested its pricing address and its developer address with a declared research user agent, followed every redirect, recorded the status code, and read the returned document for printed rates and for documented endpoints. Where an address returned nothing we tried the obvious variants, the trailing slash on Zuper's pricing path and three developer hostnames on Sera's domain, so that a 404 here means we looked rather than that we guessed.

Read the table down the last two columns and the shape of the category appears. The vendor you can price is the vendor you cannot build on. The vendor you can build on will not tell you what it costs. The third one tells you two different things depending on whether the thing reading the page has eyes.

Thryv describes itself in its own page title as a marketing and sales platform, and its home services product page is consistent with that. The feature list it leads with there is online listings management, new customer follow-up and estimates, automatic payment requests, recurring maintenance plans, prebuilt email and SMS campaigns, search engine optimization and review management. What that list leaves out is the entire operational half of a service business. No dispatch board, no route optimization, no technician mobile application, no time logged against a job, no parts and inventory ledger. We are reporting what Thryv chooses to lead with on its own home services page rather than auditing a feature matrix, and that is the point: this is a company telling you what it is for.

Zuper's field service page is the mirror image. It sells work order detail, dispatch automated on availability, proximity and skill, a drag and drop schedule, GPS routing with geo-fencing and ETA alerts, technician time and mileage tracking, estimates converted into jobs, and invoicing synced into Zoho Books, Xero and QuickBooks. Sera covers the same ground with two additions worth noticing, margin pricing on every job and a customer membership program, which are the two things a residential trade actually makes money on.

A shortlist containing Thryv and Zuper together is therefore two shortlists that a search result page merged. That matters commercially rather than pedantically. The operator who buys Thryv to fix a dispatch problem has not overpaid, they have bought a product that does not contain the fix, and the renewal conversation twelve months later is the first place anyone notices. If the problem is that the phone goes unanswered and jobs are not booked, that is the intake layer, and we have written on where a rented receptionist stops and on what Avoca's own documentation does not promise. If the problem is technicians driving past each other, that is field service management, and no marketing platform will touch it.

Thryv's pricing page returns 200 and renders a plan comparison. A reader sees a monthly and annual toggle, a Starter tier at 99 dollars a month, a Signature tier marked Recommended at 399 dollars a month, and a third column that says Custom with the words tailored pricing and solutions where a number would be. Two figures on the page, one of them the entry price and one of them the one most buyers are steered to.

The same document also carries structured data, the machine-readable block that an assistant or a shopping crawler reads instead of the rendered layout. That block advertises seven offers, each of them worded as a starting-at figure. Business Center at 244 dollars a month. Local Marketing and Ads at 255. Full Visibility and Automation at 299. Kickstart at 646, with a one-time onboarding fee of 250 dollars. Ignite at 881, same onboarding, plus 500 dollars of implementation if you are on Keap. Accelerate at 1,475 on the same terms. And a Thryv Workforce Center payroll add-on at 49 dollars a month plus 7 dollars per active staff member.

We counted the overlap rather than eyeballing it. None of the seven names in the structured data appears anywhere in the visible text of that page. The word Accelerate does appear twice in the rendered copy, and both times it is the verb, once in a headline about accelerating growth and once in a product banner. Neither of the two visible plan names appears anywhere in the structured data. And none of the six structured-data prices appears in the visible text.

The honest reading is narrower than the temptation. This is not evidence that anybody is being misquoted, and those seven offers may well be real things Thryv sells through a different page or a sales call. What it means is that the answer to the question how much does Thryv cost now depends on which channel asked, and the two channels do not agree at a single figure. On a page where a growing share of buyers will never see the layout at all, because they asked an assistant and the assistant read the block, the structured data is the price. Anybody quoting a Thryv figure back at a rep should say which one they read.

The second thing on that page worth carrying into a build conversation is the phrase full CRM, Keap or Thryv Business Center, and the 500 dollar implementation fee attached to the Keap option. That is not a billing footnote. It decides whether the documented API applies to your records at all, which is the subject of the next section.

developer.thryv.com returns 200 and is titled the Thryv Developer Portal. It is a real portal with real tooling: generated references, Postman collections, code samples, an OAuth guide. Read what it documents and every REST API on it belongs to Keap, the CRM Thryv acquired. The listed references are the Keap Core REST API v2, described there as 177 endpoints across 43 resource groups, the Keap Core REST API v1, the Keap Deals API at 28 endpoints on a separate base URL, and a legacy XML-RPC API. The quickstart printed on the portal's own front page issues a request against https://api.infusionsoft.com/crm/rest/v2/contacts.

That is a perfectly reasonable thing for a company to ship after an acquisition, and 177 endpoints is a serious surface. It also means the answer to can we build on Thryv depends entirely on which CRM your account sits on, which is a question the pricing page settles in the phrase quoted above and which no demo will raise unprompted. If your records live in Keap, the documented API is your API and the base URL you will be integrating against is Infusionsoft's. If your records live in Thryv Business Center, that portal is not documenting your data, and what you can reach becomes a question for your account team rather than something you can read before you sign.

The practical instruction is short. Before anyone scopes a Thryv integration, establish in writing which CRM the account is provisioned on, and get the base URL your data will actually be served from. A scope written against endpoints your account does not expose gets rewritten in week one, which is the same failure we described in the Housecall Pro playbook where the public API is gated behind one specific plan.

Zuper's developer documentation is the most complete of the three by a wide margin, and parts of it are ahead of where most field service vendors are. Authentication is an x-api-key header. Base URLs are regional rather than global: you POST your company login name to accounts.zuperpro.com/api/config and the response returns a dc_api_url for the data centre your account was provisioned in. The reference covers work orders, scheduling, routing, timelogs, estimates, invoices, inventory, purchase orders and approvals, with the job module alone running to dozens of documented operations.

Two things in there are unusual enough to be worth naming. Zuper publishes an llms.txt at its documentation root, an index written explicitly for AI agents, and it serves a markdown version of any documentation page if you append .md to the URL. And it ships a Model Context Protocol server, which its documentation describes as 290 tools across 28 modules, exposed over server-sent events at mcp.zuperpro.com/sse and authenticated with an x-api-key header plus an x-account-region header naming your data centre. The setup instructions cover Cursor, VS Code, Windsurf and the ChatGPT agent. The troubleshooting section states two limits plainly: your client has to support SSE-based servers rather than streamable HTTP, and, in its own words, Zuper MCP is not supported in Claude.ai. We are reporting the count and the constraints from Zuper's documentation, not from an account we hold; we have not exercised those 290 tools.

Now the sentence that should decide the shortlist. The same documentation states that the API rate limit varies based on your subscription plan, ranging from 200 requests per minute to 700 requests per minute, returned to you in an X-Rate-Limit header, with a 429 and a Retry-After when you exceed it. That is a three and a half times spread in how much traffic your integration is allowed to make, and it is set by a plan whose price Zuper does not publish anywhere. The ceiling on what you can build is priced, and the price is not disclosed.

For a single location running one schedule, none of that binds. For anyone syncing several brands nightly, backfilling history, or running an agent that reads job state on every inbound call, the difference between 200 and 700 requests a minute is the difference between a design that works and a design that queues. It belongs in the first pricing conversation, in writing, alongside the seat count. Ask which tier your quote sits in and what rate limit comes with it.

One more item from Zuper's own homepage. Its AI menu lists five agents. A CSR agent, a field agent and an AR agent are presented as available. A marketing agent and a sales coach agent are both labelled coming soon. If the demo that sold you was built around either of those two, you are buying a roadmap, and roadmaps are worth exactly what the contract says they are worth.

Sera Systems puts its pricing on its homepage under a heading that calls it simple pricing, and unlike most uses of that phrase in this category, it is. Core Essentials is 399 dollars a month and includes unlimited admins and up to four technicians. Every technician above four is 149 dollars a month. Thryv prints two figures as well, but they are flat plan prices with no stated unit, so Sera is the only one of the three whose card lets you compute a real annual figure at your own headcount without asking anybody.

Here is that arithmetic. The rates are Sera's; the technician counts are ours, chosen to span a one-truck shop through a consolidated multi-brand operation. Crossover is the fixed commission fee divided by the monthly subscription, rounded up to the next whole month, against the two fee bands from our own pricing page.

The row that matters is not the cheapest one. It is how fast the line moves. A four-technician shop would take nine and a half years of subscription to reach our smallest commission fee, which is a way of saying the question should not be asked at that size. A twenty five technician operation reaches it in thirteen months. Fifty reaches it in seven. Nothing about the business changed between those rows except headcount, and headcount is the meter the vendor bills on, which is why the build conversation in this vertical arrives with growth rather than with ambition. The same arithmetic across other verticals and other published rate cards is laid out on renting AI against owning it, and it is the page to read before this one if the question is purely financial.

What Sera gives up in exchange for that transparency is the thing Zuper leads with. There is no developer portal on its site, no published REST reference, and the three obvious developer hostnames each redirect to an administrator login and return 404. Its integrations page routes to a Zapier listing. Zapier is a real integration path and it handles a great many workflows perfectly well. It is also a trigger and action surface rather than an API, which in our experience means three things when a build sits on it: you are limited to the operations the connector chose to expose, a third party now sits in your data path, and bulk reads of historical records are slow and metered in a way a direct API is not. If you can see the build coming, that constraint is worth pricing at the point of purchase rather than discovering later.

We commission custom AI systems for a living and we would rather say this plainly than sell around it. If you are running four technicians on Sera at 4,788 dollars a year, or on Thryv's entry tier at 1,188 dollars a year at its printed monthly rate, there is no version of a fixed-fee commission that returns your money. Our smallest band is 45,000 dollars. Against the four-technician row above, that is more than nine years of subscription. The correct move at that size is to buy the software, configure it properly, and spend the difference on technicians and trucks.

Two things move the answer, and neither of them is revenue. The first is technician count, because that is what these vendors meter on, and the table above shows the crossover collapsing from month 113 to month 13 across a headcount range many operators cross in three years. The second, and the one that actually produces the calls we take, is the number of separate systems holding pieces of the same customer. One brand on one system does not need a custom layer. Four brands on three systems, with the phone answered somewhere else and the memberships tracked in a spreadsheet, has a problem no single vendor sells a product for, because the problem sits between their products rather than inside any of them.

That second case is a different page and we have written it: which field service system a platform should standardize on works through when to force one golden master and when a reporting layer above several systems is the cheaper answer. The point worth carrying here is that it is not a question about Thryv or Zuper. It is a question about how many of them you own.

Which system holds the record, and what is the base URL for it. On Thryv this is the Keap question, and it is answerable in one email. On Zuper it is the data-centre question and the answer arrives from a single call to the account configuration endpoint. On Sera the honest answer today is that there is no documented base URL and the path runs through Zapier. None of these is disqualifying. All three change what a build costs, and all three are cheaper to learn now.

What rate limit does this quote come with. Zuper is the only one of the three that states a number and it states a range rather than a figure, tied to a plan tier it does not price. Ask for the tier and the limit in the quote, in writing. The same discipline applies to every system in this vertical, and our integration playbooks for FieldEdge, ServiceTitan and Workiz take the same question to each of those products.

What leaves with you. A rate card tells you the cost of staying. It says nothing about the cost of going. The contract question is whether you can extract job history, customer records and technician time in a structured form without a professional services engagement, and it is worth settling before the first invoice rather than during a migration. We took the same question to BuildOps and published what its own contract says about your data.

What we would build on top of any of the three is not a replacement for it. It is the layer these products do not sell because it sits between them: intake that routes on data living in more than one system, a reporting spine that reconciles brands running different software, and the specific workflow where your operation differs from the template the vendor shipped. How that runs, week by week, is on the process page, and what it costs is on the pricing page, both stated before any call. We have not commissioned a build on Thryv, Zuper or Sera, and this page is research rather than a case study; the home services work we can point at is instrumented and published on the home services resource.

Related reading.

On the rest of the home services stack: what we would build for a home services platform, the consultants platforms actually shortlist, and what this work costs in this vertical. On the systems the larger brands run, what each ServiceTitan Pro add-on does and where it stops, ServiceTitan Pro against a custom build, and five things platforms get wrong about ServiceTitan AI. On the integration surfaces themselves, the playbooks for Housecall Pro, FieldEdge, ServiceTitan and Workiz. On intake, which is the layer Thryv is closest to, the receptionist cost ceiling and the Avoca alternatives. On what a contract lets you take with you, the BuildOps reading. On the buy-or-build arithmetic underneath all of it, renting AI against owning it, off-the-shelf against a commission and generic SaaS AI against a commission. On consolidating an estate, which system to standardize on and the 100-day add-on playbook. The maturity assessment places your operation before any of it, the wider vertical set sits on the industries hub, the rest of the series is on the comparisons hub, and if you would rather talk it through than read another shortlist, talk to us.