CRM Software Comparison: Scorecard, Costs and Practical Test

Anyone searching for a CRM comparison ends up with lists of ten, fifteen, or thirty systems. The names repeat themselves: HubSpot, Salesforce, Pipedrive, Zoho, Microsoft Dynamics 365, monday, Bitrix24, CentralStation, cobra. They're usually sorted by features, ratings, and price tiers.

The problem: A feature checkmark says nothing about whether you reach that feature in your daily routine in four clicks or in fourteen. And price information in comparison articles goes out of date faster than the articles get updated. So check prices and plan limits directly on the provider's pricing page on the day you decide, not in a list.

One sentence to take with you: A CRM that nobody on the team maintains is more expensive than no CRM at all. So the question is less "Which system can do the most?" and more "Which system will actually get used at our company?"

Before comparing individual providers, determine which type of software you're looking for. Anyone testing a pipeline CRM against office software with customer management is comparing apples to toolboxes.

Customer file, customer management, CRM: the difference

• Customer file: A collection of master data – name, address, contact person, phone number. Nothing more.
• Customer management: Master data plus history, i.e. notes, documents, quotes, and invoices per customer.
• CRM (Customer Relationship Management): Customer management plus control of sales, such as sales opportunities, stages, follow-ups, and evaluations across all customers.

Anyone searching for "customer management software" or "customer database software" often needs the middle tier – and ends up paying for a full CRM with features that go unused.

Five system types at a glance

The well-known names from comparison lists can be sorted into these types. Do this yourself based on the product pages, because providers keep expanding their products, and a classification made today could already be off by spring.

Rule of thumb: when customer management is enough

A rule of thumb for small businesses: if you have fewer than about 50 active customers a year, don't track sales stages over several weeks, and every order starts with a quote, type 2 is enough. A pipeline CRM pays off more once several people work on open sales opportunities at the same time and you need to know who last called which contact. For more decision help on this, see the article CRM system for small businesses: selection and decision guide.

Award points to each candidate according to fixed weights. The weighting below is designed for small businesses with two to ten people. Shift points if your case is different – but set the weights before you start testing, otherwise the system that looked prettiest on the first click wins.

Knockout criteria before scoring

Check three points first. If a candidate fails here, it's eliminated, no matter how many points it would otherwise score:

1. Unusable export: If more than 2% of the values are lost during export or break apart into unreadable columns, you'll later be locked into the system.
2. No data processing agreement available, even though the provider stores customer data for you.
3. The core workflow is missing: If you write quotes and the system can't represent them, you're testing the wrong product.

Count clicks instead of guessing usability

"Intuitive" appears in every product description. Count instead. Run through these five tasks in each candidate and note the clicks:

1. Create a new contact with company and contact person
2. Log a note about a phone call
3. Create a quote from the contact
4. Set a follow-up in seven days
5. Show all open quotes

Scoring: up to 25 clicks total gets full 15 points, 26 to 40 clicks gets 10 points, 41 to 60 clicks gets 5 points, above that 0. These thresholds are set, not scientifically determined – what matters is that you check all candidates against the same yardstick.

The monthly price on the pricing page is the smallest number in the comparison. Instead, calculate the total costs over 36 months:

3-year costs = license costs (users × price per user × months) + add-on modules × 36 + setup hours × hourly rate + migration effort

Worked example with assumed values: two fictional systems, no real provider prices.

After three years, both systems are only €24 apart, even though System B looks cheaper per month at first glance. Setup makes the difference.

If, on the other hand, the team stays at three people, System A only costs 3 × 15 × 36 + 360 = €1,980 and is thus €744 cheaper than System B. So the calculation tips on a single assumption: whether you'll hire people in the next twelve months. How user licenses affect growing teams is shown in the article CRM software for small teams: role test and costs.

What's often missing from the calculation

• Read-only access: Does the office or tax advisory need their own access, and does that cost a full seat?
• Plan limits: Many plans cap contacts, storage, or automations. Check at what volume you'll slip into the next tier.
• Annual instead of monthly billing: An annual contract can be cheaper but ties you in beyond your trial experience.
• Your own time: Setup hours are not a side issue. Anyone who sets them at €0 systematically favors the more complex systems.

A CRM contains personal data of your customers. Before you import real data, clarify these questions in writing – by email to the provider or from their documentation:

• Where are the servers located, and where are the backups kept?
• Does the provider offer a data processing agreement, and how do you conclude it?
• Which subcontractors are used?
• Can roles be set so that, for example, trainees can't see revenue figures?
• Can individual contacts be deleted completely, not just archived?
• Is two-factor login available for all users?

For the technical side, it's worth looking at the guidance from the German Federal Office for Information Security (BSI): the BSI publishes recommendations on the secure use of cloud services, which you can use to guide your questions to the provider. For the legal assessment of your specific case, a data protection consultant is responsible, not a comparison article.

December is a good time to prepare the selection – and a bad time to test.

Shortlist in December, test phase in January

Do the math: a 14-day trial period that starts on Monday, December 21, 2026, ends on Sunday, January 3, 2027. This includes the nationwide public holidays on December 25 and 26 as well as January 1, and Christmas Eve and New Year's Eve are barely working days at many companies. What remains is six full working days – too few to check the core workflow with real customers.

If you start the same 14-day trial on Monday, January 4, 2027, you have ten working days until Sunday, January 17. In Baden-Württemberg, Bavaria, and Saxony-Anhalt it's nine, because of the public holiday on January 6.

From this follows this schedule:

1. Early to mid-December: Determine the system type, enter the scorecard weights into your own table, check the knockout criteria, select three to four candidates.
2. Between the years: Prepare test data (see next section). This also works when the phone is quiet.
3. From January 4: Start the trial accounts at the same time, so all candidates run under the same conditions.
4. By mid-January: Fill out the scorecard, calculate the cost formula, decide. If the trial has automatic subscription renewal, check in advance when and how the cancellation applies.

Preparing data: the 50-contact set

Don't test with the provider's sample data. Put together a set of 50 real or realistically anonymized contacts with 12 fields each – company, contact person, street, postal code, city, phone, email, customer number, industry, last invoice, note, status. That's 600 values.

Deliberately include tricky cases: umlauts, a company with two contact persons, a note with a line break, a foreign phone number, a duplicate with a typo.

Then the actual test: import the set, export it right away again, and compare both files. More than 12 lost or fragmented values (2%) is a knockout criterion. The turn of the year is a good fit for this, because you're already reconciling the customer list with the year's invoices anyway, and duplicates show up in the process.

Field service: testing without a network

If you or your team work on-site with customers, switch on airplane mode on the smartphone and try three things: search for a contact, create a note, open an address in the maps app. Switch the network back on and check whether the note arrives. In basements, new buildings, and rural areas this is not an edge case. A system that only shows a loading spinner without network access gets zero points on this line.

Too many candidates. Anyone testing eight systems doesn't test any of them properly. Three to four are enough once the system type is already determined.

The boss tests alone. The person who will later create notes every day discovers different obstacles than the one who looks at monthly reports. Have at least one other person run through the five standard tasks.

Double data entry gets overlooked. A pipeline CRM alongside separate invoicing software means customer data in two places. After a few months, addresses no longer match. You spot this problem early if you ask during testing: Where is the invoice created, and where does it get the address from?

The exit isn't checked. Hardly anyone tests the export before buying – and that's exactly when it would be easiest.

If your core workflow is quotes, invoices, and payments, office software with customer management like MeinGewerbe belongs on the shortlist. Check it with the same scorecard as any other system; the weights and knockout criteria apply to us just the same.

Is free CRM software good enough for the comparison?

Yes, as a candidate – as long as you factor the limits of the free plan into the cost formula. Check at what number of contacts or users an upgrade becomes necessary and whether export is included in the free plan. You'll find a structured test specifically for free solutions in the article Customer management software free of charge: the 24-point test.

Is Excel enough as a customer database?

As long as one person works alone and the list stays manageable, often yes. It gets critical as soon as two people make changes at the same time, notes on history get added, or quotes are supposed to be generated from the list. Then version control, permissions, and the link to documents are missing.

What CRM systems are there, really?

Roughly the five types from the table above: spreadsheet or address book, office software with customer management, pipeline CRM, all-in-one platform, and ERP-adjacent or self-hosted suites. The specific providers change, the types stay the same. Assign every name you come across to a type first, then you can quickly sort out what doesn't fit your workflow.

How long should a CRM trial last?

Long enough for the core workflow with real customers: at least one quote, one follow-up, and one export. With a prepared 50-contact set, you can manage that in ten working days per candidate if the trials run in parallel.